ijerph5

Med-MDPI/ijerph_5/ijerph-17-15-05607.txt

@@ -0,0 +1 @@
+Intimate partner violence (IPV) is a worldwide public health problem. Here, a bibliometric analysis is performed to evaluate the publications in the Intimate Partner Violence (IPV) field from 2000 to 2019 based on the Science Citation Index (SCI) Expanded and the Social Sciences Citation Index (SSCI) databases. This work presents a detailed overview of IPV from aspects of types of articles, citations, h-indices, languages, years, journals, institutions, countries, and author keywords. The results show that the USA takes the leading position in this research field, followed by Canada and the U.K. The University of North Carolina has the most publications and Harvard University has the first place in terms of h-index. The London School of Hygiene and Tropical Medicine leads the list of average citations per paper. The Journal of Interpersonal Violence, Journal of Family Violence and Violence Against Women are the top three most productive journals in this field, and Psychology is the most frequently used subject category. Keywords analysis indicates that, in recent years, most research focuses on the research fields of “child abuse”, “pregnancy”, “HIV”, “dating violence”, “gender-based violence” and “adolescents”.Intimate partner violence (IPV) is a common and worldwide health concern [1,2,3,4,5,6,7,8]. According to the World Health Organization (WHO), IPV includes “any behavior by an intimate partner or ex-partner that causes physical, sexual or psychological harm, including acts of physical aggression, sexual coercion, psychological abuse and controlling behaviors” [9]. According to a WHO report in 2013 [10], over one in three women worldwide have experienced physical and/or sexual partner violence, or sexual violence by a non-partner. IPV levels vary in different regions due to a variety of cultural, economic level, social system, and religious reasons, with the highest prevalence in Africa, the Eastern Mediterranean and the South–East Asia Regions, followed by the Americas. High-income regions, the European and the Western Pacific Regions have a relatively low prevalence [10]. Since IPV is associated with many serious physical and mental health consequences: physical injury [11,12,13,14], post-traumatic stress disorder [15,16,17], HIV infections [18,19,20,21], induced abortion [22,23,24], alcohol use disorders [25,26,27,28,29], adolescent pregnancy [30,31,32,33], dating violence [30,34,35,36,37], and more, scholars from many countries have been participating in the study of IPV and how to prevent the violence [38,39,40,41,42,43,44].According to the literature, America is the earliest region to study IPV. E.J. Alpert. proposed in 1995 that physicians can play an important role in the early intervention of IPV through querying women who were treated for emergency care [45]. Over the ensuing few years, many strategies were proposed to prevent IPV, such as training programs [46], abuse screening [7,38,40], and reducing poverty and alcohol consumption [41]. Since the WHO released the “World report on violence and health” in 2002, more and more countries joined the IPV research. The collaborations between regions or countries also are increasing.Recently, bibliometric analysis has been an effective tool to quantitatively analyze academic publications to evaluate the research trends in different research fields, such as health care science services [47,48,49,50,51], Psychology [52,53], Economics [54,55], Energy [11,51,56,57] and Ecology [58,59]. Bibliometrics, first proposed by Alan Pritchard in a paper in 1969, is defined as “the application of mathematics and statistical methods to books and other media of communication” [60]. To our knowledge, this is the first time assessing the IPV research field using bibliometric methods. The aim of this research is to provide a broad overview on the IPV research area, including the following aspects: (1) the main contributors: country, institute, research group; (2) collaboration patterns: cooperation between countries; (3) the most productive journals; (4) top papers with highest citation numbers; (5) research trends by analyzing the author keywords. This study demonstrates the research focuses and hotspots of IPV research, which enable readers to understand the trajectories, key elements on the theoretical and practical contributions, and the future challenges of IPV.The analysis was based on the papers related to “Intimate Partner Violence” which were obtained from the Science Citation Index-Expanded (SCI-E) and Social Sciences Citation Index (SSCI) during the period from 2000 to 2019. The data was retrieved through the “Web of Science Core collection” by searching the title, abstract, author keywords and KeyWords plus with the search formula of “Intimate partner violence” or “Intimate partner abuse” or “spous* violence” or “spous* abuse” or “wife violence” or “wife abuse” or “husband violence” or “husband abuse” on 20 July 2020. The data of the top 25 authors in “Intimate Partner Violence” and citation analyses were acquired on 20 July 2020. Keywords and international cooperation were analyzed using the Derwent Data Analyzer (DDA) software. The Impact Factor (IF) for each journal was determined according to the report from the 2019 Journal Citation Reports (JCR). Note that some related publications that do not use the above search formula may not be included in this analysis.All told, 13,515 papers met the search criteria mentioned above, including 13 article types. They were: articles (11,450), reviews (925), meeting abstracts (550), editorial materials (333), proceedings papers (278), early access (188), letters (94), corrections (41), book reviews (32), book chapters (19), news items (9), reprints (4) and a retracted publication (1). The vast majority of articles and reviews were published in English (12,044; 97.325%), followed by Spanish (180; 1.445%), Portuguese (65; 0.525%), German (32; 0.259%), French (20; 0.162%), Turkish (14; 0.113%), Slovenian (6; 0.048%), Italian (4; 0.032%), Croatian (4; 0.032%), Polish (2; 0.016%), Korean (1; 0.008%), Afrikaans (1; 0.008%) and Hungarian (1; 0.008%). The following analysis was only based on the articles and reviews which were the majority of the publications in this field. Figure 1 shows the annual analysis of published papers and the number of countries. It is clear that the number of annual publications and countries have been increasing at a relatively high rate since 2002. This is attributed possibly to the report from the WHO [9]. Until 2019, 151 countries or regions have participated in IPV research.The top 30 most productive countries in the IPV research field are shown in Table 1. The USA led the list with the most publications (7947) and highest h-index (149). Canada was in the second position, but the amount of publications is only 12% of that from the USA. Other productive countries include the UK (899), Australia (631), Spain (554), South Africa (513), and Sweden (352). Switzerland took the first position of average citations per paper (71.73). The UK is listed in the second position (29.72), followed by South Africa (29.04), Uganda (24.39), the USA (24.19), India (23.29) and Bangladesh (22.06).Shown in Table 1, all the countries from Africa had a very high share of internationally collaborative papers, especially Kenya and Uganda. Ten European countries held a relatively high share of cooperative publications. Especially, Switzerland had an 84.17% share of co-author papers with other countries or regions. It is worth mentioning that, although the USA was the most active country—collaborating with 119 other countries or regions—over 80% of the papers published independently were from the USA. Altogether, most productive countries had frequent cooperation with other countries or regions.The academic collaboration network of the top 15 most productive countries is shown in Figure 2. Derwent Data Analyzer (DDA) software was applied to draw the network diagram on the basis of a co-occurrence matrix. The size of the nodes is according to the number of publications and the thickness of the connecting lines represent the frequency of cooperation. It is clearly demonstrated that the USA cooperated most frequently with South Africa, India, the UK, and Switzerland with strong collaboration relationships. Furthermore, the USA, the UK, Australia, South Africa, Germany, and Switzerland had the biggest collaboration network within the top 15 most productive countries.A total of 6684 institutes have participated in the study of IPV. The top 20 productive institutes, which were from the top four most productive countries, are shown in Table 2. Among them, seventeen institutes are located in the USA, one in the Canada, the UK and Australia respectively, which indicates again that the USA dominates the IPV research area. The University of North Carolina ranks first in terms of total publications, followed by Johns Hopkins University and the University of Michigan. The London School of Hygiene and Tropical Medicine holds the first position for average citations per paper (ACCP). Harvard University has the highest h-index value. It is worth noting that there is no institute from Asia, Africa or Oceania on this list. We expect more countries will increase their funding input and strengthen international and domestic cooperation to prevent IPV.Additionally, we analyzed the share of cooperative publications between institutes (see Table 2). It can be seen that all 20 of the most productive institutions have very high collaboration rates, especially the University of California-San Diego and Harvard University. It suggests that IPV research requires the cooperation of multiple institutions such as: universities, hospitals, and sectors of government and non-government.The top 25 most productive authors are shown in Table 3, based on the number of publications. J.C. Campbell led the list with 151 papers followed by J.G. Silverman (94) and G.L. Stuart. (87). Regarding the average citation per paper, C. Watts ranked first with 71.92 followed by R. Caetano (64.27), and R. Jewkes (62.19). The highest h-index was achieved by J.C. Campbell (43). Among these top 25 productive authors, 18 authors were in the USA, two in the UK and Spain, and one in Canada, South Africa, and Australia, respectively.Twelve thousand three hundred and seventy-five papers related to IPV have been published in about 103 research areas in SCI and SSCI databases, among which the top 20 are listed in Table 4. ‘Psychology’ ranked in the first position in terms of the total publications and h-index, followed by ‘Family Studies’, Public Environmental Occupational Health’ and ‘Criminology Penology’. ‘General Internal Medicine’ led the list of the ACCP (45.38), followed by ‘Neurosciences Neurology’, and ‘Pediatrics’.The 12,375 papers related to IPV research during 2000–2019 were published in 1454 journals. The top 50 journals in terms of the number of total publications are shown in Table 5. Approximately 48% of the papers were published in these top 50 productive journals. The top five journals produced 2590 papers with a 21.15% share of the publications. A bubble chart of the top 50 productive journals by year is shown in Figure 3. The Journal of Interpersonal Violence, the Journal of Family Violence and Violence Against Women were the top three most productive journals, with a sharp increase in IPV research outputs during the last decade. It can be clearly seen that there were few articles sparsely distributed in most of the top 50 journals from 2000–2007, however, there has been a rapid growth in publications since 2008. It is clear to see that more and more research workers have contributed to the ‘International Journal of Environmental Research and Public Health’. It is worth noting that Lancet (IF = 60.392), one of the most authoritative academic journals in the world medical community and one of the most influential SCI journals, is in the 46th position on the list. This suggests that IPV is a popular and important research area.To elucidate the main focus and research trend of IPV research, 10,085 author keywords from 12,375 papers were analyzed. The raw data were cleaned to ensure that keywords with the same meanings were represented by one unified word. Among the author keywords, 6794 (68%) were used only once and 1279 (13%) used twice. However, the top 30 most used author keywords appeared 16,614 (35%) times. The large number of once-only author keywords may indicate a wide range of interests in IPV research. As a bubble chart can clearly express in 3D values, using the bubble size as the third dimension, one can be applied to track research frontiers [51,61,62,63]. The top 30 author keywords by year are shown in Figure 4. Using visual bubble charts, the development trend of research can be clearly presented. Note that the number on the bubble represents author keyword occurrence frequencies and the number of publications.Although the citation impact of a paper depends on many factors [64], it is still a measure of its influence in this research field. The top 20 most highly cited publications are presented in Table 6. The most highly cited paper was “Health consequences of intimate partner violence.” published in the Lancet by Campbell, J.C. It led the list of total times cited with 1865 and held the second position for annual citations. “The Epidemiology of Depression Across Cultures” [65], authored by Kessler and Bromet, took the first position for annual citations with 128.00. “Prevalence of intimate partner violence: findings from the WHO multi-country study on women’s health and domestic violence” authored by Garcia–Moreno, C., et al., ranked in the third position with annual citations of 106.46.Among these top 20 papers, eight were published in Lancet, and one in Psychological Bulletin, the American Journal of Preventive Medicine, the Annual Review of Public Health, Jama-Journal of the American Medical Association, the Clinical Psychology Review, the Archives of Family Medicine, the Annals of the New York Academy of Sciences, Pediatrics, Child Abuse and Neglect, Aggression and Violent Behavior, the Archives of Internal Medicine, and the Bulletin of The World Health Organization, respectively. The USA contributed nine of them, followed by South Africa (4), Switzerland (3) Canada (2), Australia (1) and Ireland (1), and again indicated that the USA was the leading country in this research field. It is worth noting that three papers from South Africa were related to the study of the relationships between IPV and HIV infection and prevention in South Africa. Through analyzing the publications about IPV, we found that 1547 papers were associated with HIV research and a 34% share of the publications was related to Africa. It suggested that more and more scholars agree that HIV and IPV are related to some extent [19,21,66,67,68].One hundred and fifty-one countries contributed 12,357 publications to Intimate Partner Violence (IPV) research from 2000 to 2019, indicating that IPV is a global public health problem and attracting worldwide attention. It is clear that the number of publications has increased steadily since the WHO released the first World Report on Violence and Health in 2002. The number of papers from 2010–2019 was 9884, which represented 80% of the total number of publications. Also, the number of countries which were involved in IPV research increased every year, except for several fluctuations, which indicates that more and more countries have put their efforts to study and prevent IPV.North America, Western Europe, and Australia were the most active regions in the research of IPV. This was further confirmed by the most active institutions and authors. There was no institute from Asia and Africa in the top 20 most productive institutions. China and India, as the world’s most populous countries, had very low productivity. One possible reason might be the traditional culture difference, funding input, and economic level. Another possible reason is that while the WoS database is comprehensive, some journals published from India, China, and other Asian and African countries are not indexed in WoS. Furthermore, as most SCI papers are published in English, some non-native English-speaking researchers might not produce high quality papers due to the language problem to some extent. These thoughts might explain the low productivity from Asia and Africa.The obvious change in the number on the bubble of the author keywords showed the trend of IPV research: “intimate partner violence” (4399 times) was the most frequently used keyword and increased sharply during the last ten years (2008–2019), followed by “domestic violence” (2166 times), “child abuse” (985 times), “violence” (650 times), “sexual violence” (611 times), and “HIV/AIDs” (605 times). It is worth mentioning that, among the top 30 author keywords, five were related to “woman”, including “women”, “pregnancy” “violence against women”, “battered women” and “women’s health” and two were related to adolescents and children, including “adolescents” and “child abuse”, which indicates that the biggest victims of IPV are women and children. Studies on the impact of children and young adolescent’s exposure to IPV have attracted great attention from the scholars over the last two decades [33,34,36,69,70]. This trend reflects on the one-in-four of the total 12,357 papers being related to children. Additionally, “child abuse” “pregnancy”, “HIV”, “dating violence”, “gender-based violence” and “adolescents” were used at a very low frequency during 2000–2007 but increased rapidly during the last decade, which might be the new emerging research direction.The top 20 cited publications are shown in Table 6. The article with the highest citation was a review article published in 2002 and discussed the increased health problems caused by IPV [14]. Overall, five papers were published in 2002. Therefore, 2002 was a milestone year in IPV research. The article “The world report on violence and health” analyzed and summarized the main points of the first report on violence and health released by the WHO in 2002 [4]. “Prevalence of intimate partner violence: findings from the WHO multi-country study on women’s health and domestic violence” was published in Lancet in 2006 and discussed the prevalence of IPV in 10 mainly low and middle-income countries [5]. The report, “Prevalence and health effects of intimate partner violence and non-partner sexual violence”, was released by the WHO in 2013 and demonstrated that 30% of all women from over 80 countries have experienced violence by an intimate partner [10]. The WHO, with other agencies, launched a RESPECT women program to prevent violence against women in 2019 [71].Here, we presented a general overview of the Intimate Partner Violence (IPV) research area in terms of leading countries, institutes, and research trends. The USA definitely led IPV research with the most publications and highest h-index. There was no doubt that more and more countries have been participating in IPV research. Since IPV is a world health issue, we expect that, as more and more researchers join this research area, more results will be published based on the collaboration between different research groups all over the world, which will continue to make an effort to stop or prevent IPV. Needless to say, how to prevent IPV more effectively is still a big challenge, although many scholars have made various suggestions to intervene or stop IPV. Furthermore, more and more researchers have recognized that IPV is associated with women’s vulnerability to HIV. We expect more research will focus on these areas.This study can help potential researchers to quickly understand IPV globally. It also can provide useful information for relevant research in terms of identifying the research trends and potential collaborators, for example. Additionally, this study can help policy makers improve policymaking to prevent IPV.Author Contributions: Y.W. (Yuehua Wan) designed the study. J.C. and H.F. responsible for data collection. Y.W. (Yanqi Wu) analyzed, interpreted the data, and wrote the manuscript. All authors have read and agreed to the published version of the manuscript.This research was funded by Humanity and Social Science Project of Zhejiang Education Department (Y201738172) and Zhejiang Provincial Natural Science Foundation of China (LS18G03012).The authors declare no conflict of interest.The number of the publication and number of countries by year.Collaborative relationships among the top 15 most productive countries.Bubble chart of the Top 50 productive journals by year.Bubble chart of top 30 author keywords by year.Contribution and impact of the top 30 most productive countries in Intimate Partner Violence (IPV) research.Note: TP total paper, TC total citations, ACCP average citations per paper, SP Share of publications. nCC number of cooperative countries or regions.The Top 20 most productive institutions for publications, citations, and h-indices during 2000–2019.Note: TP total paper, TPR% the percentage of articles or journals in total publications, TC total citations, ACCP average citations per paper, SP Share of publications.Contribution of the top 25 authors in IPV research.Note: TP total paper, TPR% the percentage of articles of journals in total publications, TC total citations, ACCP average citations per paper.Contribution of the Top 20 research areas in IPV research.Note: TP total paper, TPR% the percentage of share publications, TC total citations, ACCP average citations per paper.Contribution of the top 50 most productive Journals in IPV research.Note: IF: impact factor.The Top 20 most cited publications in IPV research field during 2000–2019.Note: Total Citation/Year: Total Citation/ (2019-Publication Year).

Med-MDPI/ijerph_5/ijerph-17-15-05608.txt

@@ -0,0 +1 @@
+Childhood anterior cruciate ligament (ACL) injuries—which can pose a major risk to a child’s sporting career—have been on the rise in the last few decades. Dynamic knee valgus (DKV) has been linked to an increased risk of ACL injury. Therefore, the aim of this study was to analyze the acute effects of an ACL injury prevention protocol (ACL-IPP) and a soccer-specific fatigue protocol (SSFP) on DKV in youth male soccer players. The research hypothesis was that DKV would be reduced by the ACL-IPP and increased by the SSFP. Eighteen youth male soccer players were divided according to baseline DKV. Those with moderate or large DKV performed a neuromuscular training protocol based on activation of the abductor and external rotator hip muscles. Those with little or no DKV performed a soccer-specific fatigue protocol. DKV was assessed using the single-leg squat pre- and post-protocols in both legs. The ACL-IPP significantly decreased DKV during single-leg squat (p < 0.01, effect size = 1.39), while the SSFP significantly increased baseline DKV in the dominant leg during single-leg squat (p = 0.012; effect size = 1.74). In conclusion, the ACL-IPP appears to acutely reduce the DKV in youth male soccer players, and the SSFP seems to acutely increase the DKV in those players who showed a light or no DKV in a non-fatigue situation. By using the SSFP, it may be possible to determine which players would benefit from injury prevention programs due to increased DKV during game scenarios, while hip abductor and external rotator neuromuscular training may be beneficial for players who have moderate and severe DKV during single-leg squat under non-fatigued scenarios.Jumping is one of the most common actions in sports. The vast majority of sports practices require jumps and explosive movements in the execution of their main sporting gestures. Thus, these skills can be considered as performance factors [1]. However, the landing pattern seems to influence to a great extent the forces received by the joints involved, especially the vertical forces [2] and therefore the risk of injury [3]. The type of injury in each sport varies, although particularly in soccer, the lower body is by far the most affected in all age ranges and performance levels [4,5]. The knee and the ankle appear to be the areas with the highest prevalence of injury in this sport [6] and nearly one-third of these injuries have been reported to be due to poor knee function [7]. In fact, between a third and a quarter of the soccer injuries occur without contact [7,8], which is quite worrying.The anterior cruciate ligament (ACL) rupture is one of the most severe and prevalent injuries in soccer and ball sports [9], occurring mostly in noncontact situations [10,11]. Furthermore, ACL rupture in soccer becomes even more important, as it seems to be one of the most complex injuries to treat and the one which disables the athlete the longest [12,13,14]. In addition, even after a proper ACL reconstruction and rehabilitation, individuals often have impaired strength, proprioception, stability, balance and neuromuscular control [15], as well as an increased risk for ACL re-injury [16]. Aside from this increased re-injury risk, 59–70% of injured soccer players appear to develop knee osteoarthritis, with total knee arthroplasty required in 15% of those cases [17,18]. In fact, many of the injured players are not able to return to their pre-injury level of performance [19], which is extremely relevant.Therefore, it is obvious that the ACL injury affects not only the performance or the quality of life of those involved, but also the economic burden on health systems, with estimated costs of around US$26 billion per year in the United States, including the treatments dedicated to reconstruction and rehabilitation [18,20]. Furthermore, it should also be noted that the number of ACL injuries in children and adolescents has increased considerably in the last years [21,22]. Due to the musculoskeletal immaturity of this population it seems that even more attention should be paid than in adults, since an injury at such a young age could have unexpected complications and even drastically limit the child’s future sports career [23].Multiple theories regarding ACL injury (e.g., quadriceps shear force, axial loading or knee hyperextension) have been proposed in previous literature, although it is currently stated that the main mechanism of injury involves more than one plane of movement [24]. Thus, different studies have showed that knee valgus and the tibial rotation could be the main causes of ACL injury [24,25]. They are caused mainly in landings or abrupt changes of direction, in which the reaction forces with the ground may be five to seven times the body weight [26]. Dynamic knee valgus (DKV) is a modified pattern of movement or alteration in the alignment of the lower limb, mainly observed in the frontal plane [27] and with knee abduction load predicting 70–80% of ACL injury risk [3]. It should be noted that the occurrence of DKV is more pronounced in the female gender [28], although this does not mean that there is no risk in the male population [29]. Several factors have been analyzed as triggers of this alteration in knee movement, but two of the recent factors that have shown some evidence have been a reduced ankle dorsiflexion [30] and a deficit of strength or impaired activation of the abductor and adductor hip muscles, in particular a weakness in the abductors and external rotators of the hip [31,32]. Recent evidence suggests that knee and ankle bracing may reduce DKV [33,34].The literature has demonstrated certain benefits and a reduced risk of ACL injury using ACL injury prevention programs [35]. Specifically, programs focusing on neuromuscular and proprioceptive enhancement have been shown to reduce the risk of ACL injury by 51–88% [36,37,38]. However, to the best of our knowledge, all preventive training programs proposed in the existing literature have been based on treatments lasting from several weeks to an entire season. Specific warm-up exercises have been shown to be effective in tolerating greater demands or requirements in sports practice and reducing the risk of injury [39]. Indeed, strengthening the hip abductor muscles has been proposed in several ACL injury prevention programs [40,41], although always in conjunction with other exercises and never in isolation. A recent study has shown that weakness of the hip abductor musculature (e.g., gluteus medius) predicts knee abduction moment and thus the risk of ACL injury [42]. Therefore, we hypothesized that a specific neuromuscular training of the hip abductor muscles during the warm-up would be capable to acutely decrease the knee abduction and the DKV during sports practice. This would be of great practical relevance in terms of reducing the risk of injury in the short term, without prejudice to the absolute importance of continuing to carry out, simultaneously, a long-term injury prevention program.It is also widely recognized that most injuries, not only those related to the ACL, occur in the final stages of sports performance, which coincides with the presence of muscle fatigue [43]. Since muscles contribute to joint stability, neuromuscular fatigue has also been proposed as another risk factor for non-contact ACL injuries [44,45,46]. However, a recent review has concluded that the fatigue protocols published in the literature do not uniformly alter lower extremity biomechanical factors, due in part to the heterogeneity of the protocols and tasks proposed and suggests further research in this regard [47]. In addition, the few studies that have analyzed the effect of fatigue on DKV in pre-pubertal male children have used a bipodal drop–jump task as assessment method [48,49], while some studies have shown that one-leg tasks (e.g., such as a single leg squat) may be more useful in discriminating DKV because it requires greater stability and neuromuscular control [50,51]. Therefore, the objective of this study was to analyze the acute effects of an ACL injury prevention protocol (ACL-IPP) and a soccer-specific fatigue protocol (SSFP) on DKV in youth male soccer players. The research hypothesis was that DKV would be reduced by the ACL-IPP and increased by the SSFP.A convenient sample of 18 youth male soccer players (age: 12.51 ± 0.87 years; weight: 48.72 ± 9.71 kg; height: 159.34 ± 9.74 cm; BMI: 19.12 ± 2.30 kg/m2), from categories U11 and U13, was recruited for this study. All had at least 6 years of training experience in amateur competitive level, training 3–4 days per week. To be included in the study, participants should have not suffered musculoskeletal injuries in the last six months. Parents or guardians signed an informed consent form detailing the purpose of the study and the protocols and procedures to be used. All the procedures were in accordance with the Declaration of Helsinki (ethical approval number: UA-2018-11-15, Research Ethics Committee of the University of Alicante).Before the pre-intervention evaluations, a standardized and guided warm-up was performed, consisting of joint mobility, light continuous running and dynamic stretching. Evaluations were conducted on an individual basis. The frontal plane of the single-leg squat (SLS) test on both legs—dominant and non-dominant—was recorded at different times during the intervention, with a high-definition camera with 4 K recording technology. The camera was placed 3 m away from the athlete and at the height of the subject’s knee above the ground, using a tripod. Prior to the recordings, three anatomic landmarks were bilaterally identified on athlete’s lower limbs with 2-cm-diameter markers. Afterwards, the videos were analyzed by two specialists with the 2D motion analysis software Kinovea v.0.8.15 (Kinovea open source project under GPLv2), which has demonstrated its validity and reliability in the literature for measuring angles and distances [52].First, an ACL injury prevention protocol (ACL-IPP) with elastic bands was performed, focusing on neuromuscular and proprioceptive function of the gluteus medius. Five minutes before and after the protocol, the performance of the SLS test was recorded to analyze the pre–post-ACL-IPP differences. Second, and on a different day, participants who did not show DKV performing the SLS test participated in a soccer-specific fatigue protocol (SSFP), expressly designed for the study. Before the fatigue protocol and after reaching a fatigue level between 9–10 in the CR 0–10 scale [53], they performed the SLS, which was recorded to analyze the pre–post-SSFP differences.The SLS was the chosen method for evaluation, as some authors have suggested that one-leg methods are better than two-leg ones at discriminating DKV [50,51]. The evaluation of each leg consisted of 3 trials, obtaining the average of the three with a variation coefficient of less than 10% [54]. Intraclass correlation coefficients (with 95% confidence limits) were calculated for each observer, and these demonstrated high and excellent values of relative reliability (0.902, 0.896 and 0.857, for DKV basal values, post-ACL-IPP and post-SSFP, respectively). Markers were placed in the anatomic areas of interest (i.e., anterior superior iliac spine; the midpoint of the tibiofemoral joint, on the patella; and the talocrural joint, on the frontal ankle area, at the level of the malleolus) for subsequent video analysis. The frontal-plane projection angle of the knee valgus was defined by the angle formed from a linear line that connects the anterior superior iliac spine with the midpoint of the tibiofemoral joint and a second line connecting the midpoint of the tibiofemoral joint and the talocrural joint. The maximum degree of DKV was evaluated, analyzing the maximum tibiofemoral angulation (frontal plane) in relation to the Q-angle, which is defined as the angulation in the anatomic reference position [55]. The difference (δ) between these two variables was used as the dynamic value of each participant, measured in degrees [31]. To stratify the sample according to the level of DKV presented in the basal situation, the total angulations were divided into three proportional ranges. Thus, participants were classified according to the following criteria: null or light DKV (0° ≤ δ ≤ 16.2°); moderate DKV (16.3° ≤ δ ≤ 32.4°); severe DKV (δ ≥ 32.5°).Prior to the SLS test, the participants performed a bilateral knee flexion from the standing position until they reached 60° of knee flexion, measured by a digital goniometer (Digital Baseline Absolute + Axis Goniometer, Model 12–1027, version 7–08, Fabrication Enterprises, Inc., White Plains, New York, NY, USA). In that position, a string was placed in contact with the knee, which was the reference depth that the participant had to reach in the SLS test. From a one-leg standing position with arms crossed on the chest, the participant was instructed to perform the SLS, doing a knee and hip flexion, trying to keep the trunk upright. The depth of the squatting position was individually standardized using the string barrier placed previously [56]. In order to homogenize the performances, the athletes did not receive any information regarding the horizontal displacement of the knee in the execution of the test, beyond keeping the whole foot in contact with the ground, the arms crossed on the chest and the trunk as straight as possible.Only participants previously listed as moderate or severe DKV were included in this protocol (n = 10; age: 12.68 ± 0.86 years; weight: 45.57 ± 7.44 kg; height: 157.83 ± 7.14 cm; BMI: 18.31 ± 2.43 kg/m2). With the objective of analyzing the acute effect that neuromuscular and proprioceptive exercises focused on the hip abductors could have on the DKV of the knee, an ACL-IPP developed specifically for this study was carried out. The exercises were always performed in the same order, with a single series of each exercise, with a one-minute recovery between exercises. Ten repetitions of the knee band squat exercise, 10 repetitions for each side of the side-steps exercise and 5 repetitions each leg in the Bulgarian split squat exercise were performed.To standardize the squatting depth, each athlete was previously asked to perform a squat slowly, until he reached a knee angle of 90°, measured by the digital goniometer. Taking that measurement as a reference, a bench was placed at this height and they had to touch the bench in each repetition. An elastic band was placed around the knees of the participant, who had to perform the squatting exercise keeping the hip, knee and ankle aligned, preventing the elastic band from pulling the knees inward.With a rubber band around the knees and in a standing position and the knees semi-flexed, the participants performed lateral displacements, causing tension in the knee against the movement. The participants had to keep their hips, knees and ankles aligned, preventing the elastic band from pulling the knees inward.The starting posture was a one-leg standing position with this leg on the floor and the other supported behind, on a bench at a previously defined height by the length of the participant’s tibia (e.g., distance between the lateral malleolus and the external femoral condyle). From that position—and with an elastic band around the knee of the supporting leg—the participant had to perform the movement up to a knee flexion of 90°, measured by the digital goniometer, avoiding the displacement of the knee inwards produced by the band.Only participants previously categorized as light or no DKV were included in this protocol (n = 8; age: 12.73 ± 0.95 years; weight: 54.40 ± 13.25 kg; height: 164.04 ± 9.47 cm; BMI: 19.86 ± 2.55 kg/m2). To analyze whether fatigue could increase levels of DKV, a soccer-specific fatigue protocol developed explicitly for this study was carried out. The protocol consisted of a ball possession between two teams formed by two players each one, in a limited area of 15 × 15 m. One team had to keep possession of the ball, while the other had to avoid it. Every minute and by means of a whistle, all the players left the ball and performed a sprint up to a certain previously established point (with a cone), located 15 m away from the playing area. Then, they continued with the ball possession, following this procedure uninterruptedly until each individual athlete reached a fatigue level between 9–10 in the CR 0–10 scale. Figure 1 shows an example diagram of this protocol.The descriptive data of the study (age, weight, height and BMI) are shown as the mean ± standard deviation. The normality of the sample was checked by the Shapiro–Wilk test. Since the assumption of normality was not met in all variables, Wilcoxon test was used to check for differences. The effect size (ES) was calculated by the Hedges’ g, by means of the formula: g=M1−M2SD*, where SD* is the pooled and weighted standard deviation. Due to the small sample size, the Hedges equation was corrected and multiplied by [(N−3N−2.25)N−2N]. Pre–post protocols differences (Δ) in each protocol and differentiated by leg dominance, were calculated in percentage values. Spearman correlation coefficients were calculated to analyze the relationships between age/anthropometric variables and all performance variables in the tests and protocols performed. All the analyses were performed using SPSS, v.25 (IBM Corp., Armonk, N.Y., USA). A value of p < 0.05 was established to determine statistical significance. Post hoc power analysis was conducted where significant differences were found between interaction effects [57].Table 1 shows the average pre–post intervention values of the two protocols performed (ACL-IPP and SSFP), differentiated by leg dominance. No statistically significant differences were found between dominant (DL) and non-dominant leg (NDL) in the pre and post-ACL-IPP assessments (p = 0.260, p = 0.721, respectively). No statistically significant differences between dominant and non-dominant leg were found in the post-SSFP assessments (p = 0.674), although they were found in the pre-SSFP assessments (p = 0.028).Figure 2 shows the effect sizes (ES) of the two protocols (ACL-IPP and SSFP) differentiated by leg dominance, as well as the pre–post differences (Δ) in percentage. According to Rhea [58], the following criteria of the effect size interpretation were followed: g < 0.25 as trivial; 0.25 < g < 0.50 as small; 0.50 < g < 1.0 as moderate; and g > 1.0 as large. The values obtained in the post hoc power analysis were: 0.992 to ACL-IPP DL and NDL, 0.997 to SSFP DL and 0.475 to SSFP NDL.Table 2 and Table 3 show the correlations between age/anthropometric variables and pre–post-ACL-IPP and SSFP, respectively, both in DL and NDL.Figure 3 shows the statistically significant correlations found between: (a) age and pre-ACL-IPP DL; (b) age and post-SSFP DL; (c) weight and post-SSFP DL; (d) height and post-SSFP DL. In addition, a significant correlation was found between BMI and post-SSFP DL. No statistically significant correlations were found between other variables related to age/anthropometric variables and ACL-IPP/SSFP (p > 0.05).ACL injury prevention is especially important in soccer, where many players fear ACL tears [9,12,13,14], its complications and injury recurrence [15,16,17,18,19]. One of the most important findings of the present study is that ACL-IPP significantly decreases the DKV similarly on both legs during the SLS test performance (62.57% and 53.34%, in dominant and non-dominant leg, respectively). This finding could be a contributing factor for decreasing the risk of ACL injury in sports related to landings and sudden changes of direction [2,3,24,25,26].To date, the literature has only shown results from long-term ACL injury prevention programs in youth athletes, which have lasted between 4–10 weeks. These have reported from 18% to 67% reductions in DKV in youth male and female players of different ball sports [37,47,59,60,61,62,63,64,65]. However, the current study is based on an acute intervention as part of the specific warm-up. This makes our results highly relevant in practice, since using the ACL-IPP as part of the warm-up would be able to significantly decrease the risk of ACL injury in training or competition in the short term. This does not mean that a longer-term injury prevention program should be discontinued, but rather that the two could be perfectly compatible, with the advantages of both short-term and long-term prevention.On the other hand, several studies have shown a 23.24% to 389.47% increase in DKV following different fatigue protocols [66,67,68,69,70,71,72]. These variable results seem to be due to the great heterogeneity of the fatigue protocols, as well as the DKV evaluation technique [73]. In reference to this heterogeneity, it was suggested that the level of fatigue [70,74] and the specificity of the fatigue exercise [75] can influence the kinematics and DKV. That is why our SSFP was designed for being as specific, intense and similar to real competition situations as possible, increasing DKV in both the DL and NDL (356.59% and 49.34%, respectively). Remarkably, it should be indicated that these increases were obtained in participants who did not have a small DKV at rest. Thus, the DKV presented by the athlete after the SSFP, could probably be very similar to the presented in a competition match, which seems to be far from the value in non-fatigue situations. Therefore, it seems that the assessment of DKV in male youth soccer players should not only be carried out in a non-fatigue situation, but also in fatigue situations [68]. This would provide a more accurate understanding of the player’s actual risk of suffering an ACL injury, which would be of great practical relevance in the area of injury prevention.The greatest increase in the DL compared to NDL may be due to the type of activity-specific fatigue protocol applied. Since the SSFP is intended to simulate real competition, it is likely that participants will tend to use their DL to a greater extent, causing increased fatigue in this limb. This selective or localized fatigue is unlikely to occur with nonspecific soccer fatigue protocols. This may support the results of Daneshjoo and Mohseni [76], in which they also observed an increased DKV in the DL following a soccer-specific fatigue protocol in youth male. Therefore, it would probably be advisable to work on the improvement and prevention of DKV unilaterally and independently [77]. Since it seems that the values of DKV differ between both legs, it would be suggested that the dynamics of each leg should be considered individually in male youth soccer players.In addition, our results seem to indicate an inverse significant correlation between the DKV presented in the dominant leg before the ACL-IPP and age. This may suggest an increased risk of injury at early ages, which has also been previously suggested [78]. Our study has also found direct correlations between weight, height and BMI with DKV in the dominant leg after fatigue. This may suggest that lower height and weight at early ages may reduce the risk of ACL injury [79]. This is probably not comparable to other age ranges, since the increase in muscle mass as maturation progresses causes body composition to vary [80].To the authors’ knowledge, this is the first study focusing into the analysis of the acute effects of an ACL injury prevention program through a specific warm-up of the hip abductor muscles to reduce the DKV in male youth soccer players. Although our data are quite promising, it should be noted that our sample size was limited. However, our study was not performed with an a priori power analysis and was likely underpowered. It is proposed that future research will be able to confirm and reinforce our results using a larger sample size and an a priori power analysis, as well as analyze whether the ACL-IPP could have long-term effects. It is also suggested that future lines of research try to elucidate whether these benefits are equally applicable in the female gender and/or in other age groups. It is finally suggested that future research should examine whether the joint implementation of the ACL-IPP as part of a specific warm-up and a long-term injury prevention program may achieve better results than both performed separately. This would help to extend the range of practical application of ACL injury prevention programs, including ideally a combination of short- and long-term approaches.The use of an ACL injury prevention program (based on hip abductor and external rotator neuromuscular training) as part of a soccer-specific warm-up appears to acutely reduce DKV in male youth soccer players with increased baseline DKV values during a single-leg squat. The use of a soccer-specific fatigue protocol resulted in larger baseline DKV values (especially in the dominant leg) and further validation studies may help to establish it as a protocol to detect players that require additional neuromuscular training for the prevention of DKV during game scenarios. Therefore, detection and appropriate prevention of DKV through sport-specific exercise may hold promise as a means of preventing knee injuries in male youth soccer players.Conceptualization, M.A.G.-L., J.M.C.-T. and M.G.-J.; methodology, M.A.G.-L., J.M.C.-T. and M.O.-N.; software, M.A.G.-L., J.M.C.-T. and M.O.-N.; validation, M.A.G.-L., J.M.C.-T. and M.G.-J.; formal analysis, M.A.G.-L. and J.M.C.-T.; investigation, M.A.G.-L. and J.M.C.-T.; resources, M.A.G.-L., J.M.C.-T. and J.T.-M.; data curation, M.A.G.-L. and M.O.-N.; writing—original draft preparation, M.A.G.-L., J.M.C.-T., M.G.-J. and M.O.-N.; writing—review and editing, M.A.G.-L. and J.T.-M.; visualization, M.A.G.-L. and J.M.C.-T.; supervision, J.M.C.-T. and J.T.-M.; project administration, M.A.G.-L. and J.M.C.-T.; funding acquisition, J.M.C.-T. All authors have read and agreed to the published version of the manuscript.This research received no external funding.No acknowledgments.The authors declare no conflict of interest.Soccer-specific fatigue protocol (SSFP) diagram.Effect sizes of protocols differentiated by leg dominance.Significant correlations found between: (a) pre-ACL-IPP DL and age; (b) post-SSFP DL and age; (c) post-SSFP DL and weight; (d) post-SSFP DL and height. ACL-IPP—ACL injury prevention protocol; SSFP—soccer-specific fatigue protocol; DL—dominant leg.Average pre–post intervention data in the protocols differentiated by leg dominance.Note: CI—confidence interval; LL—lower limit; UL—upper limit; ACL-IPP—ACL injury prevention protocol; SSFP—soccer-specific fatigue protocol; DL—dominant leg; NDL—non-dominant leg. * p < 0.05; ** p < 0.01.Correlations among age/anthropometric variables and ACL-IPP in DL and NDL.Note: BMI—body mass index; ACL-IPP—ACL injury prevention protocol; DL—dominant leg; NDL—non-dominant leg. ** p < 0.01.Correlations among age/anthropometric variables and SSFP in DL and NDL.BMI—body mass index; SSFP—soccer-specific fatigue protocol; DL—dominant leg; NDL—non-dominant leg. * p < 0.05; ** p < 0.01.

Med-MDPI/ijerph_5/ijerph-17-15-05609.txt

@@ -0,0 +1 @@
+Healthcare services are facing challenges in increasing their efficiency, quality of care, and coping with surges in demand. To this end, some hospitals have implemented lean healthcare. The aim of this systematic review is to evaluate the effects of lean healthcare (LH) interventions on inpatient care and determine whether patient flow and efficiency outcomes improve. The review was performed according to PRISMA. We used six databases to search for studies published from 2002 to 2019. Out of 5732 studies, 39 measuring one or more defined outcomes were included. Hospital length of stay (LOS) was measured in 23 studies, 16 of which reported a reduction, turnover time (TOT) decreased in six out of eight studies, while the turnaround time (TAT) and on-time starts (OTS) improved in all five and seven studies, respectively. Moreover, eight out of nine studies reported an earlier discharge time, and the boarding time decreased in all four cases. Meanwhile, the readmission rate did not increase in all nine studies. Lastly, staff and patient satisfaction improved in all eight studies. Our findings show that by focusing on reducing non-value-added activities, LH contributed to improving patient flow and efficiency within inpatient care.In addition to the constant demand to improve their quality of care, hospitals are facing challenges in increasing their efficiency [1], reducing costs [2], and coping with surges in demand, while providing greater value to patients. Inefficiencies such as inadequate resource utilization and poor patient flow, might contribute to care delays and overcrowding, therefore affecting the safety of patients, staff/patient satisfaction, and the overall care quality [3,4]. Patient flow is the movement of patients through care settings [5]. It encompasses the physical resources, medical care, and internal systems required to get patients from the admission to the discharge while preserving quality and patient/staff satisfaction [6]. Both inpatient and outpatient care present opportunities to increase efficiency [7]. Hence, efficiency measures and performance indicators are paramount in the survival of healthcare systems [8]. Some measures used in outpatient care include emergency department length of stay (LOS) [9,10], the waiting time to see a healthcare professional [11,12], the waiting time for treatment [13], the waiting time for triage [14], and patients left without being seen [15,16], among others. Conversely, some common indicators within inpatient care are the hospital length of stay (LOS) [17,18], boarding time [3,19], and discharge order time [19,20]. Likewise, major areas of inefficiency that hospitals are trying to reduce are present in the perioperative workflow [21]. Hence, additional indicators include turnover time (TOT) [8,21,22], turnaround time (TAT) [23,24,25], and on-time starts (OTS) [8,26,27]. A related outcome is the readmission or revisit rate [28,29]. Within inpatient care, hospital overcrowding has become a widespread problem, with constrained bed capacity and admission bottlenecks having negative impacts on quality and safety. Inpatient hospital services represent 20% of medicare spending, whereas outpatients account for 8% [30].The patient admission scheduling problem has been revisited from different approaches. Through operations research, the authors in [31] propose a model considering LOS, admission, discharge time, and constraints on the utilization of operating rooms for patients requiring a surgery. Similarly, an algorithm is proposed to scheduling patient admissions more efficiently, taking into account the medical needs of the patients as well as their preferences [32]. In [33], the authors try to maximize patient satisfaction, taking into account the expected LOS and the room overcrowding risk. In an attempt to deal with both cost issues and quality, healthcare providers have been looking outside the healthcare area for guidance and inspiration [34]. To increase their efficiency, hospitals are implementing lean healthcare (LH) in their processes, with a focus on eliminating waste [35] while increasing the value for patients. Healthcare value has different definitions [36]. In this study, such values are considered “activities that enhance the quality of healthcare and promote patient well-being so as to achieve better outcomes” [37]. In connection with this, LH divide activities into either non-value added (NVA) or value added (VA) [38]; the VA activities contribute to fulfilling patient needs, whereas the NVA activities use unnecessary space, time or resources and do not meet patient needs [38,39]. LH contributes to exposing NVA activities and taking action to reduce or eliminate them [40]. Similarly, waste is anything other than the minimum quantity of space, equipment, or staff time that is necessary to add value to a service or product [41]. The term “lean” initiates from the Toyota Production System (TPS) [42], which aimed at increasing processes efficiency. The TPS entered the medical sector in the early 2000 s commonly known as lean healthcare [43,44]. Applying the TPS was recognized as an effective strategy to improve outcomes and lower costs by incrementing the efficiency of hospital-based clinical care [45]. In the U.S., a survey found that about 70% of hospitals implement LH or similar approaches [46]. Since its introduction, LH has been implemented in virtually all hospital departments, including cardiology, surgery, and intensive care units (ICUs) [21,28,29,47,48,49,50,51,52,53].LH is not free from difficulties, including adjustments in transferring the principles and tools to a new setting [25], as well as methodological restrictions at the implementation phase [54,55]. Different authors have summarized the effects of LH through systematic reviews (SR) with different approaches, such as LH within emergency departments [56], quality improvements in surgery [57], lean-six sigma in surgery [58], lean-six sigma in radiology [59], and lean-six sigma in the healthcare industry [60], whereas others have focused on care efficiency measures [61], contextual aspects and change mechanisms [62], lean facilitators [63], and the positive impacts of LH [64]. Additionally, reviews on LH provide thematic analyses [65], updates [44], and operational definitions [66]. Still others are focused on hospital waste management [67], the choosing wisely approach [68], sustainability [69], leadership and management [70], and safety and patient care [71]. To the best of our knowledge there is not a SR focusing on inpatient care and outcomes related to patient flow and efficiency. Complementarily, different authors have proposed models analyzing the relationship between LH and performance outcomes using structural equation modelling [72,73,74,75] and confirmatory factor analysis [76].Notwithstanding these studies, research on the effect of LH on efficiency and patient flow within inpatient care still remains in its early stages. To address this gap, our research aims to classify, organize, and summarize evidence regarding the effects of LH on efficiency and patient flow outcomes within inpatient care. To contribute to the body of knowledge on LH, we conducted a systematic review to determine whether LOS, TOT, TAT, OTS, boarding time, discharge times, and readmission rates are improved with a LH intervention. In addition, we reviewed the changes in satisfaction of patient and staff as secondary outcomes. The remainder of this paper is organized as follows: Section 2 describes the methodology adopted for this systematic review. The summary of results is presented in Section 3. Section 4 widely discusses our results. Finally, limitations and conclusions are presented in Section 5 and Section 6, respectively. For this systematic review we registered a protocol on the International Prospective Register of Systematic Reviews (PROSPERO; Ref CRD42019134287). The review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [77,78]. Figure 1 presents the flowchart of the stages involved in the selection process, while the resulting PRISMA checklist summarizes all the requirements covered (see online Supplementary Table S1). The subsequent subsections discuss the methodology.For the search we used the following databases: PubMed-Medline, CINAHL, The Cochrane Library, Scopus, Web of Science, and Ebsco. In addition, we searched for grey literature on OpenGrey, Grey Literature Report, Google Scholar, and ProQuest. An initial search was performed to develop a search strategy based on the Peer Review of Electronic Search Strategies (PRESS) [79]. The ultimate search strategy is depicted in the Supplementary Data S1. We utilized components of the Effective Practice and Organisation of Care (EPOC) group’s search strategy, combined with selected MeSH terms and free text terms of PICOS elements (population, intervention, comparator, outcome, and study design). We collected studies published between 2002 and 2019 in English. Likewise, we examined the references of the retrieved articles to look for additional studies. We re-ran the search before the final analysis.We selected studies whose main intervention was LH (also named as TPS or lean) within inpatient care from both the private and public sectors, studies improving patient flow, and studies providing sufficient data (in the study or by email). In addition, studies addressing similar interventions such as six sigma, rapid improvement event (RIE), or Baystate Patient Progress Initiative (BPPI) were also selected. We classified interventions as implementation strategies according to the taxonomy of the Cochrane EPOC Group [80], particularly in the continuous quality improvement subcategory. Conversely, the exclusion criteria for studies were defined as follows: studies considering ambulatory times, studies not involving a patient flow-related outcome (e.g., medical device efficiency, supplier efficiency, and medical device manufacturer efficiency), studies lacking data, and literature on pharmacologic interventions. We searched for randomized controlled trials (RCTs), controlled before–after, and quasi-RCT studies. Additionally, we included a case-control, cohort, and pre–post studies. Cross-sectional studies, abstracts, surveys, and opinion papers were excluded.We categorized the main outcomes as utilization of services, access to services, and healthcare resources use [81]. For the former, we reviewed the changes in the length of stay (LOS) for patients admitted (i.e., the time from arrival to bed to discharge from the hospital). For the perioperative process, we reviewed changes in the outcome on-time starts (OTS), measured by the change of the percentage of starting a procedure on time; the turnover time (TOT), measured as the interval in minutes between patient departure and the arrival of the subsequent patient to the OR; and turnaround time (TAT), measured as the interval in minutes between the conclusion of surgical dressing and surgical incision of the subsequent.As for access to services, we analyzed the boarding time, measured as the time patients spent since the decision to be admitted until being assigned to a bed, and the discharge order time, measured as the change in the percentage of early patient discharges or discharge orders before noon. Finally, we also reviewed the changes in readmission or re-visit rates to the hospital. As secondary outcomes, we reviewed the changes in satisfaction of patient and staff.Each study was independently screened by two reviewers to identify the title, abstract, and keywords. The rate of disagreements was 14% and was solved by means of discussion. Then, two reviewers retrieved and assessed the full texts of the relevant studies based on the inclusion and exclusion criteria. If reviewers did not reach an agreement, then a third reviewer evaluated the study. The raw data were extracted by one reviewer and examined by a second reviewer; these data included the authors’ names, titles, publication years, settings, studies length, study designs, countries, participant demographics, intervention and control conditions, outcomes, and details for the risk of bias assessment. Finally, we used standardized forms to tabulate and organize the collected data. Given the heterogeneity of the studies in terms of their study designs (mainly observational and descriptive), settings, and outcomes, we were unable to pool the results and conduct a meta-analysis. Therefore, we provide a comparative summary of findings for the main outcomes using the measures of effect (i.e., the means, medians, or percentages) in the same way as they were reported.The risk of bias was assessed by means of the Cochrane’s tool ROBINS-I (Risk Of Bias In Non-randomized Studies of Interventions) [82,83]. We used this tool because most of the research was observational and assessed the systematic difference between the results found from a non-randomized study of interventions (NRSI) and a pragmatic randomized trial [84]. The judgment criteria were comprised of five levels (no information, critical, serious, moderate, and low) for each of the seven bias domains covered by the ROBINS-I tool [82]. Two reviewers independently used the algorithm from ROBINS-I to reach an overall risk of bias (RoB) judgment for each study; if a difference persisted between the reviewers, then a third reviewer evaluated the research and came to an agreement.We found 5732 studies in the initial phase. Duplicates removal resulted in 2314 potentially pertinent papers. Then, in the screening phase, 776 studies were removed after applying the exclusion criteria. Moreover, 836 LH interventions underwent a full-text review. Yet, 797 studies were excluded for the reasons stated in Figure 1. Ultimately, the systematic review comprised 39 LH interventions. For settings, 21 of the LH interventions were conducted in the operating room/surgical units, whereas five focused on the emergency department (ED) and two on the intensive care units, among others. In addition, most of the studies were conducted in the U.S. (n = 27), Spain (n = 2), and the Netherlands (n = 2); similarly, early LH interventions seem to have arisen in 2004, yet there is an increase after 2011 (see online Supplementary Tables S2 and S3).Sixteen studies informed a LOS decrease following LH interventions, with 105.85 days representing the longest reduction [85]. Conversely, only seven studies stated no variation after the interventions [20,29,86,87,88,89,90]. For turnover time (TOT), six studies confirmed a decrease, whereas only two studies reported no change [26,27]. Moreover, the turnaround time (TAT) decreased in the five studies that addressed it. Boarding time was only evaluated in four LH interventions, with better results in all of them [3,19,29,91].The readmission rate was evaluated in nine studies, with none reporting an increase. Seven of them showed no change, and two studies reported a reduction [90,92]. The outcome on-time starts (OTS), was measured in seven studies, all of them with positive results. For discharge order time, eight studies reported an earlier discharge time, while one study reported no change [93]. Table 1 shows the direction of the findings for the main outcomes. In addition, main outcomes, statistics, and descriptions (when available) are summarized in Table 2. Likewise, an extensive summary of findings is provided in Table S4 (online Supplementary Material), including five studies that fulfilled our requests for information. Interestingly, only eight studies reported a measure of patient satisfaction and staff satisfaction respectively, all of them reported an improvement after the intervention. Other important outcomes reported included the percentage of day of surgery cancellations [8], reduction in surgery times [94], reduction of time to surgery [95], reduction of unnecessary instruments delivered to the OR [94,96], and savings in nursing time [97]. Additional measures of efficiency after LH intervention included an increase in capacity for extra patients [97], additional ED bed hours per day [91], an increase in capacity of open beds per day [19,98], a reduction in transfer due to lack of beds [29], improved OR utilization [8,26,27], and an increase in surgical admissions receiving appropriate perioperative antibiotics [86]. Finally, mortality was measured in six studies, four of which reported no change after LH intervention [17,18,28,29], while two studies reported a decreased [87,98].We found 25 interventions of lean and six sigma (LSS) within inpatient care, predominantly showing positive results. For work teams, 33 out of 39 interventions used multidisciplinary teams, most of these studies provided positive results in their outcomes. Meeting organizational, regional or national standards/targets for the reported outcome was only discussed in 9 out of 39 studies. Regarding the types of studies, 34 were pre–post studies, among which two used controls. Meanwhile, the remaining five were cohorts. None of the research involved RCTs. Finally, in terms of risk of bias, 28 interventions were assessed as moderate and 11 as serious (see online Supplementary Materials, Table S5).The outbreak of COVID-19 added more pressure to healthcare organizations, who face an ongoing challenge to improve efficiency and meet an increasing demand for high quality of care and lower costs. The objective of this systematic review was to evaluate the effects of lean healthcare interventions on efficiency and patient flow outcomes within inpatient care. Six out of seven outcomes presented an overall improvement, while readmission rates did not increase after the LH intervention. An extensive summary of findings is provided in Table S4 (online Supplementary Materials).First, in 23 out of 39 studies, LOS was the most common process-related outcome. This finding is consistent with previous studies [44,111]. LOS is a general measure of hospital efficiency [86] and is commonly related to costs reductions when the LOS is reduced [99]. We found mixed results for LOS across the departments of the hospitals in which LH was implemented. For cardiology, two studies [47,99] reported a reduction, while one study [90] reported a non-significative reduction. Accordingly, for orthopedic and trauma, five studies showed a reduction in LOS [48,92,95,98,102]. Meanwhile, two studies did not reduced the LOS [87,88]. For both studies conducted in the ICU, one reduced the LOS [17], while the other reported no change in the ICU [29]. Interesting, when the main goal is to reduce infections within surgical units, LH contributed to reducing the LOS in two out of three studies [18,100], while one study [86] reported no change. Moreover, when LOS was reduced, other associated benefits were reported, such as cost reductions [17,95,99,100], a better return of investment [48], increased savings [47,90,93,98,102], or an earlier discharge order [3,19].Major areas of inefficiency that hospitals try to reduce are found within the perioperative workflow [21]. The perioperative process includes the preoperative, operating room, and the postoperative departments, all of which have to run like a well-oiled machine to improve performance and achieve positive outcomes [112]. In this regard, preoperative throughput is an important element in achieving the perioperative goals for the first case’s on-time start (OTS) [105,113,114]. In this research, all seven studies improved their on-start times after LH intervention. This is consistent with the results of a previous study [109]. Thus, the preoperative use of LH leads to substantial improvements in OTS. An inefficient preoperative department can delay the start of surgery and impact the patient flow throughout the day [112], which can affect other outcomes, such as TOT and TAT, which are important performance parameters for the perioperative process. We found that LH led to reductions of TOT in six studies, the largest being a 50% reduction [104]. In contrast, two papers reported no significant changes in TOT, but other benefits, such as reductions from patient in room to procedure starting or OTS [26,27]. All four studies measuring the turnaround time (TAT) in the OR reported an improvement [21,23,24,101], with 20 min being the largest reduction [21]. Similar outcomes were obtained in the reduction of TAT for pathologists [25], as well as the medication TAT [18]. A comparable tool, the plan-do-study-act cycle [115], was also used to reduce the turnaround time [116]. Remarkably, all studies reporting an improvement in either OTS, TOT, or TAT, used multidisciplinary teams during the intervention. Other factors that might affect the perioperative process, include patient-related variables, personnel unavailability [117], surgeon variables [24,118], the workflow in the anesthesia preparation tasks [22,101,103], the type of the previous operative case (emergency), the waiting time for trays or patients arriving late [119], if a scheduled gap existed between the cases [21], patient family and social support [109], and even the weather [109].Commonly targeted processes are the discharge and admission processes, due to the fact that these processes are almost always unnecessarily long and can have a large impact on the throughput of patients [111]. For the admission process, we found that the boarding time decreased in all four studies [3,19,29,91], in which 2.1 h was the longest reduction time, resulting in a boarding time of 5.5 h [19]. In the literature, boarding times ranged from 2 h (or less) to 24 h (or more), with medical/surgical patients experiencing shorter boarding times and behavioral patients experiencing longer boarding times [120]. It is recommended that boarding time frames not exceed 4 h in the interest of patient safety and quality of care [120].The availability of beds is key to reducing the boarding time by improving inpatient discharge timing [121]. Therefore, focusing on the time of discharge may be the least disruptive and most effective way to address constrained bed capacity [20]. Both premature and delayed discharges not only worsen health outcomes but also increase costs. Premature discharges can lead to costly readmissions, while delayed discharges use up limited hospital resources [122]. Despite increasing the number of hospital beds will not solve completely the problems of overcrowding [122], it can affect patient health since delays in bed access compromise patient safety [123]. In addition, healthcare providers should plan their capacity to minimize the risks associated with occupancy rates exceeding 90% [124], i.e., bed shortages and higher rates of infection [122]. The factors affecting the discharge time include the hour of admission [107], preparation for discharge order [106], and the preparedness and cooperation of patients and their families [28].Interestingly, none of the nine studies measuring readmission rates reported an increase after the intervention. Seven studies reported no change at 30 or 90 days and two studies reported a statistically significant reduction in readmission rates [90,92]. Using lean-six sigma, the readmission rate for heart failure patients was reduced up to 19.0% [90], while the US average heart failure readmission rate was 24.6% [125], and that in the UK was 17.8% by 2016 [126]. Recently, the emergency readmission within 30 days of discharge for all patients was found to be 14.4% in England [127]. To this end, healthcare organizations and governments have implemented financial incentives to improve the readmission rates, either to exceed the reduction of targets [128] or to avoid exceeding a threshold of emergency readmissions [129].Patient satisfaction and experience are reported in only in 8 of 39 selected studies. This is contrary to our expectations since LH is considered a factor for improving the flow of patients and thus associated to increasing patient satisfaction [37,130,131,132,133]. Two studies used the Press Ganey assessment survey [89,91], while the other studies used self-developed surveys, including electronic surveys in combination with interviews [23,92,101].Moreover, the literature suggests that healthcare professionals also notice LH benefits, such as an increase in their satisfaction [112,134,135,136,137] and empowerment [41,138]. Although satisfied patients and healthcare workers are prerequisites of sustainable high-quality care [139] and conversely disengaged healthcare workers are by far the main reason for lean failure [140], only 8 out of 39 studies measured staff satisfaction. With a hospital staff turnover rate of 17.8% by 2019 in the US [141] (which was reported to range from 15% to 36% in previous years) [142], this lack of evidence in the valuation of staff satisfaction following LH interventions suggests that generating the ideal staff experience has been absent from many LH transformations [140]. Among studies reporting staff satisfaction, the used instrument included the safety attitudes questionnaire (SAQ) and the operating room educational environment measure survey [21].Improving patient flow in the perioperative environment is challenging but has positive implications for both staff members and the facility [101]. Thus, multidisciplinary perioperative teams [8], perioperative benchmark practices [48], and benchmarking meetings [95] have shown good results. Despite LSS research is abundant in the OR, there are minimal studies in a preoperative context [112]. LH and SS support the development of a clinical pathway [95,102] and thus reduce LOS [143]. The lack of relevant standardization was a common theme among pathways for patients or standard work for caregivers [48]. LOS, complications, patient satisfaction, education, and readmission have all been improved by the standardized pathways developed via LSS applications [26,143]. The purpose of such standards is to set expectations [144], care protocols, and staff roles [145], thereby reducing reliance on memory [144]. For this purpose, the standardization of care alongside evidence-based guidelines might enhance value in healthcare [146].Operating rooms are linked to significant costs in most hospitals [147] but are potentially the most lucrative part of many healthcare systems [103]. With surgical care representing about a third of all healthcare spending [148], and with healthcare costs continuing to rise [149,150], healthcare enterprises are focusing on eliminating inpatient operational inefficiencies and waste to reduce unnecessary costs [93]. Thus, the OR represents an area with an opportunity to optimize work flow and supply use [151], and to maintaining an economically viable institution [22]. By applying process improvement methodologies, such as lean and six sigma, across an entire surgical suite, hospitals are attempting to improve efficiency [22]. These efforts appear to be a war on waste, which would be justified by the need to decrease costs that are not indispensable for patient care [152]. This intrinsic connection between LH and cost reductions/revenue increases was clear in about 43% of the studies (17 out of 39 studies). However, this number is still low, which suggests difficulties in interpreting such results into costs/savings, either due to a lack of personnel or training in this topic.We found 25 out of 39 studies combining lean with tools and principles of the six sigma methodology, suggesting that such integration offers a more robust approach to improving speed, quality and costs, increasing customer satisfaction, and maximizing shareholder value [153,154]. While lean focuses on reducing waste and NVA activities, six sigma focuses on reducing process variation by following the DMAIC approach (Define, Measure, Analyze, Improve and Control) and by using statistical tools. In this way, both methodologies complement each other. Lean-six sigma also provides useful frameworks to help hospital staff identify causes of delays in their own institutions [155]. This combination outperforms the use of only one methodology. Nevertheless, this integration tends to be composed of larger private hospitals with more resources for quality improvement [156]. In this sense, if the goal is to maximize quality improvements and cost savings, then LH interventions or similar methodologies (e.g., Virginia Mason Production System) must occur institution-wide, i.e., in both ambulatory care and inpatient settings [45]. We found one intervention using lean-six sigma and BPPI [19] and other using lean and RIE [8] as examples of others combinations of lean and complementary tools. BPPI is a multi-disciplinary, institutional, performance improvement initiative with the goal of decreasing ED walkouts and boarding hours, inpatient LOS and increasing the number of patients with written discharge orders before noon. On the other hand, the so called Kaizen blitz, or rapid improvement event (RIE), is a focused, fast performing and significant changes initiating activity used for general modification and redesign of observed processes and identified problems [157]. Both BPPI and RIE might differ in scope, but complement lean interventions by improving patient flow and efficiency outcomes. Similarly, other studies have shown that multimodal interventions result in improved outcomes [158,159,160].Most studies used the value stream map tool to represent both the current and future states of patient flow, thereby confirming the great importance and usefulness to identify VA and NVA activities. This importance has been pointed out by other studies [134,161,162,163,164]. Other common tools include standard work, i.e., a concept whereby each work activity is precisely described with a specific cycle time, task sequence, and other steps involved within the process [95], and is considered a prerequisite for flow [10]; the 5’S program, which is used to eliminate clutter [165] and enhance the standardization of stock, as well as to systematically organize the unit and streamline the documentation processes [97]; cause and effect analyses, which are used to map the possible causes of a problem into categories [90,166]; and Kaizen, which is a philosophy of continuous incremental improvement over time and space [29,99,104]. These findings are similar with those reported in previous studies [44,62,161], and sustain the assertion that most LH interventions focus more on tools related to assessment and improvement and less on processes-monitoring tools.Even though lean theory assumes a holistic view [62,167], most interventions occurred in a particular process or department, rather than in the whole organization. According to our results, OR accounts for the area with the largest amount of LH interventions (21), while none were implemented in a whole healthcare organization. This is consistent with the results from [65]. Therefore, small and focalized improvements support organizations sustain momentum, and any early achievement is vital to keep people from becoming dispirited [27]. Organizations with more experience could perform larger and longer projects.The time frame of most studies was longer than one year (32 out of 39). However, only around 15% of all studies conducted a follow-up process longer than one year. These results hinder to confirm the sustainability of the achievements, and may be related to the “project fatigue” in hospitals since so many difficulties within their facilities need attention [168]. Hereafter, a brief follow-up analysis might not be an appropriate indicator of improvement. Some other characteristics that might compromise the sustainability of LH achievements include increased patient volume [138], a poor understanding of the organizational context [169], insufficient space and time for coherent team co-operative improvement, the tension between promoting staff ownership and providing direction [88], the incomplete or slow adoption of the interventions [93], and a lack of standardization [48]. Naturally, to fully realize the potential benefits of LH, organizations need to minimize the impact of such barriers and capitalize on facilitating conditions that are specific to their local contexts [134]. Once institutional rules and dogma are changed, culture and workflow improve [103].The successful implementation of lean or any other improvement framework requires that the hospital and medical leadership all be strong supporters of the methodology, speak the same process improvement language, and are able to generate support and resources [10]. Besides, when LH is properly executed and is owned by the frontline workers, it can yield improvements in care metrics [9] because the employees are trained to become project leaders for improvement [170]. Engagement and empowerment alone, however, will not drive or sustain such improvement. They must be combined with a new breed of leadership that focuses on patient outcomes and performance measurements as a key motivator [171], along with effective communication and team work [118]. Here, senior leadership plays a critical role [25] by developing supporting structures such as visual control, goal deployment, short daily meetings, two-way communication flow, and a system of continuous improvement [172]. Additionally, the contributions from team members with different perspectives and disciplines was noted in most of the reviewed studies, showing better performance in patient flow indicators. Thus, the team—whether multidisciplinary teams [17,21,100], interprofessional teams [27], work stream teams [22], value teams [20], project teams [91,98], or Kaizen teams [99]—is vital in getting ‘‘buy in’’ from all the stakeholders involved [27], principally because LH continues to sustain a multidisciplinary problem-solving perspective, as demonstrated by the joint ownership of performance measures [1].Healthcare organizations are currently subject to compliance with standards, targets, and benchmarks, which serve as a reference of minimum performance levels for safety and patient flow [144]. However, the timeframes and metrics for patient throughput differ widely in both practice and literature [120]. For example, the average LOS in hospitals for acute care among OECD countries is 6.5 days, with Turkey (4.1 days) being the shortest and Japan the longest (16.2 days) [173]. In this research, the average LOS before LH was 22.9 days, and 12.5 days after the intervention, but these values depend on many factors, such as patient variables [174], treatments, and settings; e.g., after LH, the LOS in the rehabilitation ward was 58.3 days [85], 22 days in the ICU [17], and 5.3 days in the trauma center [95]. Despite the contexts for standards or target compliance, we found that few studies discussing meeting local or national standards [47,87], national rates [90], organizational goals [3], targets [8], internal benchmarks [89], or consortium benchmarks [28]. Instead, the specified goal was commonly to improve performance.In terms of risk of bias, 72% of the interventions were assessed as moderate and the rest as serious. None were evaluated as critical or as low risk since only exceptionally will a non-randomized study of interventions (NRSI) be assessed as at low risk of bias due to confounding [82,83]. Our risk of bias analysis is consistent with [82], which anticipated that most NRSI will be judged as at least at moderate overall risk of bias. The relative high number of studies with moderate or serious risk of bias might be debatable; however, it signifies our decision to include all those studies meeting the criteria for inclusion and provide a general viewpoint of the LH phenomenon, as recommended when risks of bias vary across studies [175]. Regarding the domains of bias, all studies were evaluated as serious in the bias due to selection of participants. Selection bias occurs when some eligible participants, or some follow-up time of some participants, or some outcome events, are excluded in a way that leads to the association between intervention and outcome in the NRSI differing from the association that would have been observed in the target trial [84]. In this research, none of the selected studies involved RCTs; moreover, the selection of participants into most of the studies was related to intervention and outcome. Finally, the lack of data prevented us from adjusting these types of bias as indicated by [82].Our study has several major limitations. Firstly, differences in data (patient volume, settings, and data gathering/processing approaches) and the multi-component nature of LH, limit us to generalize the results. Secondly, studies’ heterogeneity and the risk of bias prevented us from carrying out a meta-analysis to determine causal relationships. Thirdly, the majority of the studies were observational pre–post designs. Thus, the lack of randomization, the lack of matched comparison groups, and the potential existence of confounding variables limited the outcome improvements from being causally related to the LH interventions. Moreover, the lack of reliable measures of confounding domains led to the bias due to confounding. Both baseline confounding and time-varying confounding were common in NRSI, and along with bias due to selection of participants, might represent a limitation when estimating the true effect. Finally, there is a possibility that the “Hawthorne effect” led to the improvements reported in the studies, even though the changes in outcomes, as demonstrated in the statistical tests, suggest that the results were more probable due to the LH intervention.On the basis of our findings, we have summarized the main results from the LH interventions within inpatient care. As stated by most authors, LH guided and facilitated the identification of non-value activities in their processes, thereby facilitating actions to reduce them, while improving the efficiency of service. According to our findings, excessive LOS is critical for both patient safety and hospital costs, hence, delays in some procedures might lead to an extended stay and thus increase discomfort among hospitalized patients and compromise the capacity of beds. To achieve this goal, an efficient perioperative process should have a high OTS rate since delays and cancellations lead to underutilized facilities and dissatisfaction among personnel.Bearing in mind the dimensions of quality of care [176], according to our evidence, LH contributes to the provision of efficient and accessible service through a reduction in the length of stay and outcomes associated to the length in time of activities related to the perioperative process and inpatient care, such as TOT, TAT, OTS, boarding time, and discharge time. Moreover, our findings suggest that LH does not contribute to the changes in readmission rates and highlight the important relationship between capacity and demand. By reducing the time-length of the outcomes reported, healthcare professionals increased their capacity, which is crucial to improving the flow of patients to meet demands. In this regard, LH is an important support and, by using a complementary tool (such as six-sigma) that focuses on variation reduction, might help level patient flow and solve more complicated problems, as long as the organization provides support. Likewise, if properly supported, LH might contribute healthcare organizations comply with targets and standards associated to timely and effective care (throughput).Notwithstanding the improvement in outcomes related to efficiency and patient flow, indication of the LH effect on patient/staff satisfaction is still scarce among studies; similarly, although more studies are translating the obtained achievements of LH into savings, there is still a gap to fill.We suggest considering variables that might affect inpatient processes such as economic, cultural, and regional characteristics. Additionally, relevant tools and techniques and critical success factors in the implementation of LH within inpatient care should be evaluated. Despite the largely positive findings of LH intervention, caution should be taken in generalizing such findings. Consequently, additional research involving both high quality observational studies and randomized controlled trials is also recommended.The following are available online at https://www.mdpi.com/1660-4601/17/15/5609/s1, Table S1: PRISMA checklist, Data S1: Search strategy, Table S2: Geographical distribution of studies selected, Table S3: Distribution per year of studies selected, Table S4: Summary of findings of LH, and Table S5: Risk of bias.Conceptualization, C.Z.-L and D.T.; methodology, D.T. and Y.B.-L.; formal analysis, C.Z.-L., D.T., Y.B.-L., J.L.-R., S.O.; investigation, C.Z.-L., D.T., A.P.-S., and G.T.; writing—original draft preparation, C.Z.-L., D.T., J.L.-R., and A.P.-S.; writing—review and editing, S.O. and G.T.; supervision, D.T., Y.B.-L., S.O., A.P.-S., and G.T. All authors have read and agreed to the published version of the manuscript.This research received no external funding.This study was supported by Mexico’s National Council of Science and Technology (CONACYT), the PRODEP Program (Programa para el Desarrollo Profesional Docente, para el Tipo Superior) and the Universidad Autónoma de Baja California.The authors declare no conflict of interest.Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart.Direction of Findings per Main Outcomes.Note. A mark check indicates interventions reporting an outcome. For the outcome direction, (-) denotes that an outcome decreased, NC means no change in outcome, and (+) means the outcome increased. LOS indicates length of stay; TOT, turnover time; TAT, turnaround time; N, no report. The last name of the main author and the publication year are shown.Main Outcomes of Lean Healthcare Intervention.Note. OR indicates operating room; RIE, Rapid improvement event; ED, Emergency department; TPS, Toyota Production System; BPPI, Baystate Patient Progress Initiative; mo. Months; h, Hours; min, Minutes; TS, Thoracic surgery; GYN, Gynecologic oncology surgery; Gen/CRS, General and colorectal surgery. The last name of the main author and the publication year are shown.

Med-MDPI/ijerph_5/ijerph-17-15-05610.txt

@@ -0,0 +1,6 @@
+Haze pollution has a serious impact on China’s economic development and people’s livelihood. We used data on PM2.5 concentration, industrial energy consumption structure, economic development and transportation in Beijing-Tianjin-Hebei and surrounding cities from 2000 to 2017, and analyzed the spatial effect of industrial energy consumption structure and traffic factors on haze pollution by using spatial autoregressive model (SAR) and spatial error model (SEM). The results indicated that: (1) The global spatial correlation analysis showed that haze pollution had a significant positive spatial correlation, and the local spatial correlation analysis showed that the high-high clusters of PM2.5 were located in the south and middle of the region; (2) The change of industrial energy consumption structure was highly correlated with haze pollution, namely, the increase of industrial energy consumption led to the deterioration of environmental quality; (3) The change of economic development was highly correlated with haze pollution. There was no clear EKC relationship between haze pollution and economic development in Beijing-Tianjin-Hebei region and surrounding cities. However, the relationship was similar to inverted U-shaped curve; (4) The change of traffic jam was highly correlated with haze pollution, namely, the increase of fuel consumption per unit road area led to the deterioration of environmental quality. Based on the above results, from the perspective of space, the long-term measures for haze control in Beijing-Tianjin-Hebei and surrounding cities can be explored from the aspects of energy conservation and emission reduction, industrial transfer, vehicle emission control, traffic restrictions and purchase restrictions.A wide range of haze has covered central and eastern China in recent years. Haze pollution has brought serious health effects to China, and its economic cost has been rising year by year [1]. Chinese government highly values ecological and environmental protection. Guided by the conviction that lucid waters and lush mountains are invaluable assets, the country advocates harmonious coexistence between humans and nature, and sticks to the path of green and sustainable development [2] (pp. 279–285). In 2013, the Chinese government issued the implementation of action plan for the control of air pollution in the Beijing-Tianjin-Hebei region and surrounding areas. The aim was to lower air pollution in the region through joint control of regional air pollution [3].Spatial econometric panel data model was developed by the cross-section data of multiple time nodes. Anselin [4] introduced spatial effect into the cross-section data model, and proposed Spatial Autoregression Model (SAR) and Spatial Error Model (SEM). Anselin et al. [5] proposed the Ordinary Least Squares (OLS) for the benchmark models of SAR and SEM, and developed the Lagrangian Multiplier (LM) statistic to test the autocorrelation of the spatial lag term and the spatial error term. Using spatial statistical models with fixed effects items and random effects items, spatial econometric panel data model controls the individual heterogeneity.Space effect, coal burning, industry and automobile exhaust are important factors contributing to the rise of PM2.5 concentration [6]. The impact of coal burning and industrial production on pollution is essentially related to the industrial energy consumption structure [7], while automobile exhaust belongs to the transportation sector [8]. The distribution of PM2.5 has unique spatial and temporal characteristics [9,10], and its transmission characteristics within and between regions have significant positive correlation [11]. This study combines the three factors of industrial energy consumption structure, economic development and transportation to discuss the haze problem [12,13,14,15,16], and attempts to analyze the problem from the perspective of space. For the causes of haze pollution, scholars, especially natural science researchers, have conducted in-depth component analysis, tracking and monitoring. Anselin [17], Gray and Shadbegian [18], Bateman et al. [19] specifically discussed the importance of spatial factors to environmental economy research. Rupasingha et al. [20], Maddison [21] and Chen et al. [22] used the spatial econometric method to discuss the relationship between per capita income and air pollution, and the results showed that the use of spatial variables greatly improved the accuracy of the measurement model. Xue and Geng [11] using data from the China National Environmental Monitoring Centre (CNEMC) found that PM2.5 transmits across regions. Hosseini and Kaneko [23] used SDM to study the atmospheric conditions of 129 countries, and proved that pollutants could spread to neighboring countries, and a significant inverted U-shaped EKC curve could be seen after controlling the spatial relationship. Ding et al. [24] adopted SDM to test whether there was EKC trend between PM2.5 and economic growth in Beijing-Tianjin-Hebei region, and the results showed that economic growth and PM2.5 presented an obvious inverted U-shaped EKC curve. Compared with the non-spatial model, the turning point of EKC is more likely to occur when considering the spatial effect. Ma et al. [25] used SEM and SAR to analyze the interaction effect of inter-provincial PM10 and the influence of industrial structure, and found that industrial transfer lacked long-term impact on reducing pollution. Taking Beijing as an example, they found that traffic jam and spatial factors were important reasons for severe pollution.Pollution control policies should not only be analyzed in a single region, but also in urban agglomeration or the whole region [26,27] (pp. 123–155). Chen et al. [1], Adgate et al. [28], Liu et al. [29] and Bell et al. [30] pointed out the distribution of PM2.5 is spatiotemporal heterogeneity. Although the methods used in various studies are different, and limited by the variability of PM2.5 components and subjective factors of researchers, it can be concluded that PM2.5 sources in key Chinese cities mainly include three aspects: coal burning sources, industrial sources and automobile exhaust, of which the contribution of automobile exhaust is about 10% to 30% [9]. Poon et al. [6], Han and Hayashi [31] and Wu et al. [32] pointed out that although the private car ownership rate in China is lower than that in the United States and other western countries, with the increase of per capita income, this rate will keep growing and pose a serious threat to China’s environment, especially the urban environment. Zheng and Huo [33], Barth and Boriboonsomsin [34,35], Greenwood et al. [36] and Jerrett et al. [37] illustrated the connection between urban traffic congestion and carbon emissions from a microscopic perspective, and discuss the importance of urban spatial structure for easing congestion and improving the environment. In previous studies, Chinese scholars pay more attention to the impact of energy structure and economic development on the environment. Research on the impact of transportation on China’s environment started late and focused mainly on carbon emissions. Most of the studies used static panel model, ignoring the mechanism of spatial interaction. Referring to the existing research experience and combining with the spatial measurement method, this study takes the newly defined heavily polluted area (Beijing-Tianjin-Hebei and surrounding cities) as the research object, and uses the spatial panel data from 2000 to 2017 to analyze the influencing factors of PM2.5 from three aspects: industrial energy consumption structure, economic development level and traffic jam.The area to be studied is the Beijing-Tianjin-Hebei region and its surrounding areas, a total of 28 cities. These cities are among the most polluted areas in China. The Chinese government proposed the concept of ‘26 + 2’ cities in 2017, which means that 26 cities, together with Beijing and Tianjin, constitute the key areas of air pollution control in China. The location of the region in China is highlighted in Figure 1.Tobler’s First Law of Geography stated that ‘Everything is related to everything else, but near things are more related to each other.’ [38]. On this basis, spatial econometrics abandoned the traditional null hypothesis that the space does not matter for economic relationships, and added spatial dimension to the econometric model to analyze the data more accurately.Due to the existence of Tobler’s First Law, a large number of literatures focus on the spatial correlation between adjacent regions. The spatial correlation of PM2.5 can be determined by measuring the global Moran’s I index. The formula is as follows:(1)Μοran’s I=∑i=1n∑j=1nwij(xi−x¯)(xj−x¯)S2∑i=1n∑j=1nwij, S2=∑i=1n(xi−x¯)2n
+        where wij is the spatial weight matrix; xi and xj is the observed value of the space units; n is the number of cities in the study region; x¯ is the average value; S2 is the variance. Moran’s I is between −1 and 1. When Moran’s I is greater than 0, it means there is a positive spatial correlation between space units; when it is equal to 0, it means that there is no spatial correlation; when it is less than 0, it means that there is a negative spatial correlation. The setting principle of w is: (2)wij={1 when region i is adjacent to region j0 when region i is not adjacent to region j0 when i=j
+        where ‘adjacent’ includes diagonal adjacent. In other words, as long as two regions have a common boundary or intersection point, they are defined as ‘adjacent’.The global Moran’s I reflects the overall spatial correlation. Anselin [39] pointed out that the overall evaluation may ignore the atypical characteristics of local areas. It is necessary to use the local correlation index (LISA) to evaluate the specific correlation of local areas and whether there is a significant spatial cluster, which can be tested by measuring the local Moran’s I. The formula is as follows:(3)Moran’s Ii=(xi−x¯)S2∑j=1nwij(xj−x¯)
+        where Moran’s Ii measures the correlation of PM2.5 between region i and its surrounding areas. When Moran’s Ii is greater than 0, it means that regions with similar values cluster together, which is manifested as high-high or low-low spatial clusters. When Moran’s Ii is less than 0, it means that the regions with different values gather together, which is manifested as high-low or low-high spatial outliers.The spatial econometric panel data model of the subject can be established after verifying that the outcome has spatial correlation. Spatial Autoregressive Model (SAR) and Spatial Error Model (SEM) are established to analyze the spatial effects of industrial energy consumption structure, economic development and traffic factors for PM2.5. Based on the traditional econometric model, a spatial autoregressive term is added to construct a spatial autoregressive model to analyze the spatial spillover effects of variables. The model is defined as follows:(4)lnPMit=α1ESit+α2lnGDPit+α3TJit+ρwijlnPMit+δit+υit+εit,εit~N(0,σit2)
+        where ln PMit refers to the concentration of PM2.5 in year t of city i; GDPit is the GDP in year t of city i; ESit is the measurement index of industrial energy consumption structure in year t of city i; TJit is the measurement index of transportation in year t of city i; ln is the logarithm of the variable; δit is the time effect; υit is the individual effect; εit is the disturbance term. ρ is the coefficient of spatial variable and represents the degree of spatial spillover effect. ρwijlnPMit refers to the relationship of PM2.5 between one city and its neighboring cities.The spatial error model indicates that the spatial spillover effect results from the disturbance term. The model is defined as follows:(5)lnPMit=α1ESit+α2lnGDPit+α3TJit+δit+υit+εit,εit=λwijεit+μit,μit~N(0,σit2)
+        where λ is the spatial variable coefficient of the disturbance term. The meanings of other symbols are consistent with those in the spatial autoregressive model.The outcome variable is the logarithm of PM2.5 concentration in each city from 2000 to 2017. Data are drawn from the Atmospheric Composition Analysis Group [40] at Dalhousie University uses NASA satellites and ground stations to monitor and record global PM2.5 concentration. The publicly available original data are raster data with a resolution of 0.01° × 0.01°. This paper collected the PM2.5 concentration of ‘26 + 2’ cities in Beijing-Tianjin-Hebei region and surrounding areas from 2000 to 2017. In order to reduce the interference of heteroscedasticity, the values of PM2.5 concentration was logarithmically processed to form panel data for spatial correlation analysis.The explanatory variables were industrial energy consumption structure (ES), economic development level (lnGDP) and traffic jam (TJ). The structure of industrial energy consumption is the ratio of output value of high-consumption coal industry to GDP. This study collected the industrial output data of eight energy-intensive industries, which included the production and supply of electricity and heat, petroleum processing, coking and nuclear fuel processing industry, ferrous metal smelting and rolling industry, non-metallic mineral products industry, mining and washing of coal, chemical raw materials and chemical products manufacturing, nonferrous metal smelting and rolling industry, and papermaking and paper products [25]. The industrial output of eight energy-intensive industries were added up, and the ratio of them to the current year’s GDP was calculated. Data were obtained from the Statistical Yearbook (2001–2018) [41,42,43,44,45] of 28 cities and China Industrial Economy Statistical Yearbook (2001–2018) [46]. The transportation variable captures the degree of traffic jam. This is defined as the ratio of the consumption of gasoline and diesel oil of urban residents to the urban road area (t/m2). In 2006, the Intergovernmental Panel on Climate Change (IPCC) published the relationship between the speed of private cars and the amount of petrol consumed per mile in the Guidelines for National Greenhouse Gas Inventory [47]. It pointed out that the slower a car goes, the more petrol it consumes. In traffic jams, the petrol consumption is almost twice that of normal driving. In recent years, the rapid increase of the consumption of gasoline and diesel oil is partly due to the increasing rate of private car ownership. On the other hand, serious urban traffic congestion is also an important cause. This paper attempts to use the consumption of gasoline and diesel oil of urban residents to measure the degree of traffic jam. The higher the fuel consumption per unit road area, the higher the degree of traffic jam. Ma et al. [25] took traffic congestion as an independent variable and carried out a spatial analysis of the influencing factors of haze pollution, and found that the indicator of traffic congestion was not significant in whole China. This study made a spatial analysis of Beijing-Tianjin-Hebei region and surrounding cities to further explore the impact of traffic jam on haze pollution. This paper has also tried to use traffic pressure, namely the ratio of the number of private car ownership to the length of regional highways, as a traffic factor for spatial regression analysis, but the regression result is not ideal. There is a high autocorrelation between traffic pressure and traffic jam. After comparing the two regression results, traffic jam was selected as the explanatory variable representing the traffic factor. Data were obtained from the Statistical Yearbook (2001–2018) [41,42,43,44,45] of 28 cities and the Statistical Communique on National Economic and Social Development (2000–2017) [48] of 28 cities.Table 1 reveals the results of the global Moran’s I for PM2.5 in Beijing-Tianjin-Hebei region and surrounding cities. As shown in Table 1, the global Moran’s I were highly significant at p < 0.01. For a city with high/low PM2.5 concentration, there was at least one city with high/low PM2.5 concentration adjacent to it. From 2000 to 2017, this positive spatial correlation fluctuated between 0.3–0.5 and reached its peak in 2007, indicating that this positive spatial correlation developed steadily during this period. Moran scatter plot of PM2.5 in Beijing-Tianjin-Hebei region and surrounding cities is shown in Figure 2. The horizontal axis represents the standardized PM2.5 concentration value, and the vertical axis represents the spatial lag value of the standardized PM2.5 concentration value. The scatter plot takes the average value as the axis center, and the first and third quadrants respectively represent the high-high and low-low spatial clusters. The global Moran’s I is positively correlated, which means that the second and fourth quadrants of high-low and low-high spatial outliers are atypical observation areas. From 2000 to 2017, apart from 4–8 cities in the second and fourth quadrants, most cities were in the typical observation area, which further indicated that the spatial positive correlation of PM2.5 had a strong stability [49].The local Moran’s I statistic was suggested in Anselin [39] as a way to identify local spatial clusters and local spatial outliers. Figure 3 reveals the results of the cluster map for PM2.5 in Beijing-Tianjin-Hebei region and surrounding cities in 2000, 2008 and 2017. The local Moran’s I were significant at p < 0.05. As shown in Figure 3, the low-low spatial cluster was located in the west of the region (Taiyuan and Yangquan). In 2000, the high-high spatial clusters were located in the south of the region (Zhengzhou, Kaifeng, Xinxiang, Hebi, Puyang and Heze), and gradually moved to the middle of the region (xingtai, Dezhou and Cangzhou). From 2000 to 2017, the frequencies of high-high spatial cluster in Xingtai, Zhengzhou, Dezhou, Heze and Kaifeng were all higher than 7, while the frequencies of low-low spatial cluster in Taiyuan and Yangquan were higher than 10. According to the results, the high pollution concentration areas in Beijing-Tianjin-Hebei region and surrounding areas were mainly located in the south and the central part, and showed a trend of transferring from the south to the central part.The results of the global Moran’s I showed that PM2.5 concentration had a significant positive spatial correlation in Beijing-Tianjin-Hebei region and surrounding cities. To account for this spatial econometric panel data model was estimated to analyze the influence of industrial energy consumption structure, economic development and traffic jam on PM2.5. Table 2 showed the results of Lagrange Multiplier (LM) of spatial autoregression model and spatial error model, and further compared the suitability of these two models. The results showed that LM(lag) and LM(error) in SAR and SEM models were highly significant at p < 0.001. Further observation of the results of R-LM(lag) and R-LM(error) showed that R-LM(error) was significant at p < 0.001, but R-LM(lag) was not significant, indicating that the spatial error model was more suitable for this study than the spatial autoregression model. The Hausman test result showed that the statistical value of Hausman test was 83.75. The null hypothesis of random effect was rejected at p < 0.001. It is obvious that the fixed effect model should be selected for analysis. Compared with other models, the value of R2 in model(4) showed that the model had a low fitting degree, so model(5) and model(6) were selected as the reference models for analysis. The values of λ in model(5) and model(6) were both greater than 0, indicating that PM2.5 had spatial spillover effect. According to model(6), for every 1% increased in PM2.5 in surrounding cities, the PM2.5 in the central city increased by 0.651%. The spatial spillover effect was obvious.According to models (5) and (6) in Table 2, the variable of industrial energy consumption structure was significant at p < 0.001. The change of industrial energy consumption structure was highly correlated with haze pollution. For every 1% increase in ES in the region, PM2.5 increased by 0.131% or 0.134%.Energy conservation and emission reduction policies have an impact on PM2.5. From 2000 to 2007, the PM2.5 concentration in Beijing-Tianjin-Hebei region was on the rise. In 2003, China stepped into the market-oriented heavy industrialization stage, and heavy industry was the ‘big consumer’ of coal [4]. As a result, coal consumption accounted for an increasing proportion of total energy consumption, which was an important reason for the rising trend of PM2.5. From 2008, the central and local governments began to implement a series of energy conservation and emission reduction policies. For example, in 2013, the Ministry of Environmental Protection set a target of reducing PM2.5 concentration by 25 percent by 2017 in Beijing-Tianjin-Hebei region, compared with 2012. In 2014, the Beijing-Tianjin-Hebei region jointly set the goals to give top priority to environmental protection in the integration process and gradually realize unified emission standards. In 2015, Beijing stipulated that consumers across the city should be encouraged to buy energy-saving products, and eligible consumers should be given consumption subsidies. From 2008 to 2017, in the process of promoting the transfer of investment from high-polluting industries to low-polluting industries, the proportion of coal in the energy consumption structure of all cities decreased.Industrial transfer affects the regional industrial structure and further affects the concentration of PM2.5. As shown in Figure 3, the high-high spatial cluster of PM2.5 were mostly in the south and central regions. In the process of industrial structure transfer in Beijing-Tianjin-Hebei region, a development pattern gradually formed from ‘Beijing and Tianjin as the exporter, Hebei, Shanxi and other neighboring cities as the importer’ to ‘Beijing-Tianjin-Hebei region as the exporter, central and western China and northeast China as the importer’. These industries were mainly resource-based, and most of them were high-consumption and high-pollution industries. When the capacity of pollution control remains unchanged or the speed of capacity improvement is far lower than that of industrial cluster, the positive externalities of economic development and the negative externalities of pollution concentration will occur together.According to models (5) and (6) in Table 2, the variable of lnGDP was significant at p < 0.001. The change of economic development was highly correlated with haze pollution. For every 1% increase in lnGDP in the region, PM2.5 increased by 0.057% or 0.055%. Due to the existence of Environmental Kuznets Curve (EKC), the impact of economic development on haze pollution needs to be further analyzed.EKC emphasizes that in the early stages of a country’s economic development, pollution levels rise as per capita income increases. When economic development and income reach a certain level, the further growth of income will lead to the improvement of environmental quality or the reduction of pollution. There is an inverted U-shaped relationship between the variation trend of pollutants and per capita income. Given the difficulty of collecting data, most studies used GDP instead of income. Figure 4 reveals the results of the relationship between lnGDP and lnPM2.5 in Beijing-Tianjin-Hebei region and surrounding cities. From 2000 to 2006, with the development of economy, the haze pollution level showed a fluctuating rising trend, which reached the peak in 2006 and then showed a fluctuating decreasing trend, indicating that there was no clear EKC relationship between haze pollution and economic development in Beijing-Tianjin-Hebei region and surrounding cities, but the relationship was similar to inverted U-shaped curve. At present, with the development of the economy, the level of haze pollution shows a fluctuating decreasing trend.According to models (5) and (6) in Table 2, the variable of transportation was significant at p < 0.001. The change of traffic jam was highly correlated with haze pollution. For every 1% increase in TJ in the region, PM2.5 increased by 2.637% or 2.700%.In China, the problem of traffic congestion seriously affects the development of urban environment. The more developed countries are, the more significant the impact of transportation on environmental quality will be, while for the less developed countries, energy structure is a more significant factor. According to the Energy Information Administration (EIA), transportation is the main factor affecting air quality in developed countries such as the United States, Canada and Australia, while energy structure is the dominant factor in developing countries such as India and South Africa. With the rapid development of regional economy, transportation has become an important factor affecting urban environment.Regional motor vehicle emission control policies have impacts on haze pollution. From 2003 to 2008, Beijing implemented stage 2, stage 3 and stage 4 of the national emission standard, which is the same as European standard. At present, the government encourages the public to use vehicles with stage 5 of national emission standard or clean energy and new energy vehicles. In 2009, Handan, Taiyuan and Jinan issued regulations on the control of motor vehicle exhaust. The qualification rate of environmental protection test for motor vehicles was more than 90% in Tianjin, Hengshui, Taiyuan, Changzhi, Jinan and Binzhou. Strict implementation of the national emission standard is conducive to the regional control of PM2.5.Traffic restrictions and purchase restrictions of motor vehicle have impacts on haze pollution. In 2014, Beijing, Tianjin and Shijiazhuang began to implement traffic restrictions. At present, Xingtai, Handan, Taiyuan, Yangquan, Jincheng and Changzhi have also implemented strict traffic restrictions. However, some studies showed that traffic restrictions have a positive effect on haze control in the short term, but the effect can be weakened in the long term. Beijing, Tianjin and Shijiazhuang have implemented purchase restrictions of motor vehicle, and the growth rate of total vehicle ownership has been controlled. However, in 2018, China’s auto market showed negative growth for the first time in 28 years, and the auto demand had fallen for three years. In contrast to the negative automobile consumption, large cities, such as Beijing, have been implementing traffic restrictions and purchase restrictions for a long time, resulting in a large amount of pent-up consumption demand.Based on the panel data of PM2.5 in Beijing-Tianjin-Hebei region and surrounding cities, this paper constructed the spatial autoregressive model and spatial error model to study the influence mechanism of PM2.5 from three aspects: industrial energy consumption structure, economic development and transportation.The study found that PM2.5 in Beijing-Tianjin-Hebei region and surrounding cities had significant spatial cluster, and the change of PM2.5 in one city formed spatial spillover effect on adjacent cities. Industrial energy consumption structure, economic development and transportation are the significant factors influencing PM2.5. The industrial energy structure was analyzed from two aspects of energy conservation and emission reduction, industrial transfer. Through EKC, the influence of economic development was further analyzed. The factor of traffic jam was analyzed from three aspects of vehicle emission control, traffic restrictions and purchase restrictions.To alleviate the haze pollution pressure in Beijing-Tianjin-Hebei region and surrounding cities, and promote the coordinated development of industrial energy structure, economic development, transportation and air quality, our suggestions and prospects are proposed as follows:First, restrict the non-essential industrial use of inferior coal and inferior oil. Power industry enterprises that contribute a lot to PM2.5 rush to introduce low-calorie coal with low price and mix it with high-quality coal for power generation, which greatly reduces the utilization rate of coal and increases the industrial energy structure, resulting in the rise of PM2.5. In 2015, Chinese government issued the Interim Measures for The Management of Commodity Coal Quality. Accordingly, Beijing, Tianjin, Hebei, Shandong and Shanxi issued relevant measures to control the use of inferior coal. However, China’s relevant laws and regulations have not yet been improved, resulting in difficulties in implementation. Switzerland, Belgium and Norway have already been running the coal-free system for many years. In 2017, the UK achieved a full 24 h without burning coal for electricity, with half of its electricity coming from natural gas, a quarter from nuclear plants and the rest from wind, biomass and foreign imports. In the long run, developing clean energy technologies and coal efficiency technologies and increasing the consumption of renewable energy are far-reaching measures.Second, abide by the standards for industrial transfer and formulate long-term development plans. Considering the spatial spillover effect of PM2.5, the short-term industrial transfer has only a short term effect on haze pollution control. Under the background that industrial transfer has become the inevitable requirement of industrial structure adjustment in Beijing-Tianjin-Hebei region, rational industrial planning is more necessary. As shown in Figure 3, from 2000 to 2017, Taiyuan and Yangquan in Shanxi Province always maintained a low-low spatial clusters of PM2.5, which was inseparable from the rational planning of industrial transfer and undertaking in Shanxi Province. As a major source of coal resources in China, Shanxi Province is also the main force to undertake industrial transfer in the eastern region. Taiyuan and Yangquan are major destinations for steel and non-ferrous metal industries. However, they can maintain low PM2.5 levels under the pressure of developing their own coal industry and undertaking the transfer of heavy industry from outside, which is closely related to the strict time-limited transformation or withdrawal mechanism of high-emission projects in Shanxi Province. Shanxi’s experience in dealing with emissions from the coal industry is worth learning from elsewhere, including Hebei.Third, encourage local governments to support carless families in purchasing their first new energy vehicles. Traffic restrictions and purchase restrictions of motor vehicle can improve urban air quality in the short term. However, it reduces consumption power and has a negative impact on the development of relevant industries in the region, such as the automobile industry. It cannot be regarded as a long-term and fundamental measure of urban traffic management. Efforts should be made on the supply side to deal with traffic congestion. More sustainable governance policies include encouraging regions where conditions permit to give preferential treatment to new-energy vehicles in terms of parking fees, exploring the establishment of zero-emission zones and alleviating traffic congestion through appropriate administrative measures on traffic restrictions and purchase restrictions.Conceptualization, M.L.; Data curation, M.L. and C.M.; Formal analysis, M.L.; Funding acquisition, C.M.; Methodology, M.L.; Resources, C.M.; Software, M.L.; Supervision, C.M.; Writing—original draft, M.L.; Writing—review & editing, M.L. All authors have read and agreed to the published version of the manuscript.This research was funded by China Scholarship Council, Fundamental Research Funds of Central Universities of China (No.2018B713X14) and Postgraduate Research & Practice Innovation Program of Jiangsu Province (No.KYCX18_0518).The authors declare no conflict of interest.The location of Beijing-Tianjin-Hebei region and surrounding areas.Moran scatter plot of PM2.5 in Beijing-Tianjin-Hebei region and surrounding cities in 2000, 2008 and 2017.Cluster map of PM2.5 in Beijing-Tianjin-Hebei region and surrounding cities in 2000, 2008 and 2017.The relationship between lnGDP and lnPM2.5 in Beijing-Tianjin-Hebei region and surrounding cities.Global Moran’s I for PM2.5 in Beijing-Tianjin-Hebei region and surrounding cities.Note: E(I) is −1/(n-1), which is the expected value of I; Sd (I) represents the variance of the I; Z is the z test value of I, and p-value is its adjoint probability, which was obtained by Monte Carlo simulation for 999 permutations.Results of spatial autoregressive model and spatial error model.Note: *, **, and *** respectively indicate that the estimated coefficient is significant at p < 0.05, p < 0.01 and p < 0.001. The value in parentheses is the t value of the coefficient.

Med-MDPI/ijerph_5/ijerph-17-15-05612.txt

@@ -0,0 +1 @@
+Emotional and cognitive-behavioral factors influence people’s adaptability to change. Based on this premise, the objective of this study was to develop, evaluate and validate the Adaptation to Change Questionnaire (ADAPTA-10) for identifying those who show poor adaptability to adverse situations, such as those caused by COVID-19. This study was carried out in a sample of 1160 adults and produced a 10-item instrument with good reliability and validity indices. It is an effective tool useful in research and in clinical practice. Calculation tables are provided for the general Spanish population and by sex to evaluate adaptability to change. The two-dimensional structure proposed in the original model was confirmed. This instrument will enable the needs for adaptation to the new reality associated with COVID-19 to be detected and also other situations in which the subject becomes immersed which demand adaptation strategies in the new situation lived in.The disease caused by the coronavirus, SARS-CoV-2, called COVID-19 [1], was declared a pandemic by the World Health Organization in March 2020 [2] because of its rapid spread and high death toll [3].The risk of this pandemic is not only a healthcare problem, but also has severe socioeconomic and psychological implications [4]. Therefore, the approach to the COVID-19 public healthcare emergency should try to minimize both the negative physical and psychological impacts of the virus [5]. Ignoring the immediate psychological effects of this global situation would have a disproportionate mid- to long-term impact [6]. Its severity and persistence are still unknown, and therefore, how long public restrictions and measures should be maintained is also unknown [7]. What is clear is that the transition to the new normality will be a process that will put the adaptability of every individual to the test.In situations of adversity, some individuals are at greater risk of developing psychological alterations, such as posttraumatic stress or biopsychosocial disorders, while others resist and adapt well [8]. In psychology, the concept of adaptation refers to functional change in response to environmental stimuli, whether in terms of sensorial, behavioral, cognitive or emotional functioning. These changes must provide benefits to the subject, improving adjustment to the current or future environment [9]. Positive adaptation to adversity is not a completely innate trait, so this ability can be learned and developed by actively reformulating life’s challenges [10]. From a perspective of virtue, based on committed action, the individual’s practical wisdom and courage, resilience and adaptability may be understood as the transformation of adversity into opportunity [11]. In research, psychological adaptation has often been analyzed in the scope of natural disasters due to the direct impact and high losses undergone by a high percentage of the population in such situations. These losses may be permanent or temporary, total or partial, depending on the capacity of strength of the individual and the number of stressors to cope with [12,13]. Women seem to adapt the worst to highly stressful events [14,15]. High financial losses and the death of loved ones are the two factors which have the strongest long-term repercussions on adaptation and subjective wellbeing of individuals [14,16]. Thus, psychological adaptability enables adjustment to or acceptance of difficult situations and is of great value in learning to struggle with the limitations of daily life [17]. On the contrary, absence of psychological adaptability has been linked to the presence of internalization (somatic complaints, anxiety or depression) and externalization (behavioral problems) symptoms [18]. Therefore, after a significant event, the wellbeing of an individual may not return to baseline or may take many years to do so, and the adaptation responses of each subject may differ enormously [19].The hedonic adaptation model by Graham and Oswald [20] states that individuals tend to remain at a certain stable level of endogenous wellbeing, recovering from harmful events and becoming used to the good ones. Thus, when exogenous events threaten their wellbeing, people may recover, safeguarding their adjustment if they can control the situation to a certain extent through the flow of psychological resources for coping with it. Based on integration of individuals in their setting, this could generate different styles and strategies for approaching the situation [21]. According to the Threat Appraisal and Coping Theory [22], people who are exposed to stressful situations may respond with adaptive behavior which provides them with immediate and long-term wellbeing, or, with maladaptive coping, which distracts them or relieves them, making them feel good temporarily, but generating psychological distress later. According to Black and Hendy [23], the choice in many cases is related to the perceived ability to do something about the situation. Thus, although exposure to stressful factors may not always be avoidable, if one perceives that something can be done to change the situation, a more adaptive coping strategy, mainly related to proactive efforts to change the situation or its meaning, will be chosen [24,25]. Thus, staying in control of events goes beyond the resources for coping. The sense of control may be internal or external. Beliefs about stronger internal control buffer the effects of stressors, so people who feel more able to control stressful events adapt better, as they are more able to face them, even though they may be more strongly exposed to this type of event [22,26]. However, other authors note that although at first sight it might be thought that the internal locus of control is related directly with wellbeing, it is not always that way. When people face completely uncontrollable situations, maintaining a high perception of internal control of events could be a negative strategy for adaptation, generating emotional distress [27].Another of the factors involved in an individual’s adaptability is tolerance to uncertainty. Uncertainty is present in daily life (e.g., Will it rain today?), at existential moments and important decisions (personal and professional), in relations with the world (e.g., the future and unemployment during an economic crisis) at significant times (e.g., illness of a family member) and in nature (e.g., fear of natural disasters in a prone geographic area). So, the way in which people perceive and cope with uncertainty is relevant to their adaptability [28]. Tolerance to uncertainty has been defined as the way in which people understand and process information in uncertain situations and how they respond with cognitive-behavioral and emotional reactions [29]. Repetitive, expected events do not usually awaken fear, but apprehension. Worry and uncertainty usually appear when the causes of an event cannot be explained [30]. Similarly, situations of psychological uncertainty are usually coupled with anxiety symptoms due to the agitation from anticipation of threat, and with stress, which refers to persistent irritability, impatience and tension; therefore, its management is important in order to lower psychological stress and ensure adjustment [31,32]. Along with stress management, the capacity for regulating emotions is fundamental to adapting later to indispensable situations [33,34]. Thus, people who show better ability to regulate their emotions have a stronger capacity for adapting and responding to a changing environment [35].Depression is another emotional component to be kept in mind in an individual’s adaptation. Perception of the inability to cope with demands is linked to dysphoric feelings and depressive symptoms [36]. It has also been associated with cognitive inflexibility, which is transformed into problems for adapting reactions to new situations [37]. Thus, cognitive flexibility is another component to be considered in adaptation to change [38]. This refers to the ability to modify cognitive and behavioral strategies in response to changes in environmental demands [39]. Increase in psychological flexibility has been shown to diminish stress and anxiety that handicap effective response and provide benefits for wellbeing [40]. This capacity in turn depends on the ability to detect characteristics and changes in situations [38]. Monitoring conflicts is linked to the capacity for cognitive control, which facilitates assimilation and accommodation of conflict, and in turn, orientation toward specific objectives and resolution of potentially problematic or incongruent situations [41]. In a stressful situation, this could involve concentrating on information related to the threat and the one that leads to eliminating stressors, distancing oneself from nonessential information [38]. Thus, the mechanisms of control and conscious awareness enable detection and adaptation to situations in which information is conflictive [42].The effects of awareness on the adaptability to change are mediated by perceived social support, which favors redefinition of stressful situations so they are not perceived as such or supply resources that enable the severity of such events to be reduced [43,44]. Thus, counting on strong perceived social support provides material sustenance and emotional comfort to people, in addition to helping them to reduce the negative evaluation of events, enabling them to alleviate distress and improve adaptation [45]. Along this line, the study by Koffer et al. [26] found that beliefs about control in stressful situations increase with age, postulating that this result could be due to the decrease in availability and efficacy of psychosocial resources.Interest in knowing the capacity of individuals to adapt to change has led to studies seeking to establish the cognitive and emotional dimensions giving rise to this variable. However, analysis of the factors that enable success in new situations and unexpected changes in one’s environment have focused mainly on the job context [46,47]. Hedonic adaptation to important life events has also been analyzed [48], but not to changes in environmental demands with less transcendence in life than the birth of a child or the death of a family member. Thus, to date, the factors determining adaptability to everyday events and demands have not been established. Therefore, the following model was hypothesized as a starting point for the design and validation of an evaluation scale for adaptability to change. The factors included on it are those mentioned above, differentiating between those that pertain to the emotional dimension because of their repercussion on feelings experienced during adaptation (social support, anxiety, stress and tolerance to uncertainty) and those pertaining to the cognitive-behavioral dimension because of its involvement in management, control and action on it (that is, stress management, locus of control, state of alertness, coping, emotional management, cognitive flexibility and tolerance to uncertainty) (Figure 1). The latter (tolerance to uncertainty) is included in both dimensions because it includes both emotional and cognitive responses [29].There are many gaps in our knowledge of control, treatment or even socioeconomic effects derived from the COVID-19 pandemic [49]. Along with the strong perception of uncertainty and threat caused by the pandemic and by the new measures that must be adopted in daily life [50,51,52], they can affect behavioral efficacy and the capacity for management and coping [53]. Adaptability to change is fundamental to avoid psychological alterations linked to the accumulation of stressors [15], however, there is no instrument that specifically evaluates this capacity in the individual. Therefore, based on the model conceptualized above, and in view of its relevance for ensuring adjustment to change in daily scenarios, the objective of this study was to develop, evaluate and validate the Adaptation to Change Questionnaire.The sample was made up of 1168 adult Spaniards. The questionnaire included control questions for detecting random or incongruent answers, which led to the elimination of eight subjects, so that the final sample was comprised of 1160 people. The mean age of the sample was 38.29 years (standard deviation (SD) = 13.71) in a range of 18 to 82. Of these, 69.9% (n = 811) were women and 30.1% (n = 349) were men, with mean ages of 37.05 (SD = 13.34) and 41.16 years (SD = 14.14), respectively.The sociodemographic data were collected in an ad hoc questionnaire, which included items on age, sex, marital status and education.The General Health Questionnaire-28 (GHQ-28) [54,55] was used for evaluating general health and related functional symptoms. This questionnaire consists of 28 items grouped in four subscales with seven items each: Subscale A (somatic symptoms), Subscale B (anxiety and insomnia), Subscale C (social dysfunction) and Subscale D (severe depression). Each question has four gradually worsening answer choices. The subject must mark the answer chosen based on recent weeks.The Adaptation to Change Questionnaire (ADAPTA-10) was designed to evaluate an individual’s adaptability to the demands of novel situations. This questionnaire is made up of 17 items related to the individual’s disposition to achieve successful adjustment to unknown situations or events. It includes items linked to emotions of distress and anxiety that could appear when faced with changes or others related to the capacity for controlling, managing and acting in different situations, that is: social support, anxiety experienced, depression, stress management, awareness and state of alertness, coping concentrated on the problem, tolerance to uncertainty, emotion management, mental flexibility and locus of control. The answers are rated on a five-point Likert-type scale (from “not at all” to “very much”).This cross-sectional study was done with snowball sampling carried out on social networks and instant messaging during the seventh and eighth week of confinement of the Spanish population, specifically from 1 to 12 May 2020. The participants filled out the tests individually in a time estimated at 5–10 min.The stages that led to the conceptualization and development of the ADAPTA-10 Questionnaire for evaluating adaptability to change are described below. The study was approved by the University of Almeria Ethics Committee (UALBIO2020/032, 06-25-2020). All the subjects in the study participated voluntarily and gave their written informed consent prior to filling out the questionnaire, after being informed of the objectives of the research and the anonymous nature of their answers. The data were collected and processed respecting all of the rights and guarantees as provided for in EU Regulation 2016/679 and Organic Law 3/2018 of 5 December on Protection of Personal Information and guarantee of digital rights.The questionnaire was implemented as a CAWI (Computer Aided Web Interviewing) interview, in which the participants expressly gave their consent by marking a box for the purpose before going to the questionnaire screen.The first step was an analysis of the scientific literature on the subject of adaptability to change in an adult population. Search machines were used to collect studies that could contribute to the development of the items on the questionnaire.After the review of the literature on the subject, experts were consulted to evaluate a first proposal of possible constructs for the final repertoire of indicators. The result of this stage was a list of constructs which we took as the starting point to develop the content of the items: social support, stress management, alertness, coping, tolerance to uncertainty, emotion management, locus of control, cognitive flexibility, anxiety and depression. Following this, a specific search was made on measurement of each of the proposals.The next step was to write the items, which were in first person because it was to be a self-informed questionnaire. To check the intelligibility and clarity of this first set of items, a pilot questionnaire was drafted and distributed to a sample of 30 subjects selected by snowball sampling, all of them adults over 18 years of age. Then, the content and wording of the items were reviewed considering their observations, making minor modifications to reduce the answer bias or misunderstanding.The questionnaire was comprised of 17 items and the answers for each item were rated on a five-point Likert-type scale (1 = not at all, 2 = a little, 3 = somewhat, 4 = quite a lot, 5 = very much).Finally, the questionnaire was validated by administering it to a representative sample of adults (see sample characteristics in the section on Participants). Although the scale was designed with several theoretical constructs as the basis, we could not determine any latent factor models until the measurement structure was proven statistically based on the original theoretical model proposed.Data were analyzed in two stages following the validation steps recommended by Pérez-Fuentes et al. [56]. The first stage dealt with the study of the structure according to the original theoretical basis of the Adaptation to Change Scale. To approach this objective, the sample was first divided at random into two homogeneous independent subsamples. The first sample was used for calibration (n = 578) in the exploratory (EFA) and confirmatory factor analyses (CFA) of the proposed theoretical Adaptation to Change model. The confirmatory factor analysis was done for the original model taking the following indices of fit as measures: χ2/df (Degrees of freedom), Comparative Fit Index (CFI), Tucker-Lewis index (TLI) and Root Mean Square Error of Approximation (RMSEA), with their confidence interval (CI) at 90%. Values below five were considered acceptable for the χ2/df index [57] for the CFI and TLI over or near 0.90, and for the RMSEA, below or very near 0.08 [58]. As a general rule, fit of the model is considered to be good when the χ2/df ≤ 3, TLI > 0.90, CFI > 0.95 and RMSEA ≤ 0.05. The appropriate re-specifications were made of the model, which showed good indices of fit, considering theoretical and statistical criteria (change index, error of estimation, standardized error of measurement), but the model was not improved. The Akaike Information Criteria [59] was used for model selection. Then, the re-specified model was validated based on the second subsample (n = 583), used as the validation sample. The Cronbach’s Alpha [60], Spearman-Brown and intraclass correlation coefficient were used for the reliability analysis of the new scale.Finally, in the second stage, an analysis was done that supports the invariance of the factor structure proposed across sex (men/women). First, the goodness of fit of these structures was tested in both subsamples separately (Model M0a—Men and Model M0b—Women). The result was four nested models which were evaluated: (A) Model 1: both samples together simultaneously with free estimation of the parameters, (B) Model 2: metric invariance shown, (C) Model 3: scalar invariance shown, (D) Model 4: strict invariance. With no criterion of consensus to determine the criteria to be used to evaluate the difference in fit between the nested models [61], for evaluation of fit, this study used the ΔCFI. Thus, the model is completely invariant if the ΔCFI is below 0.01 [62]. Similarly, the validity of the construct was evaluated by analyzing the correlation of the items and factors with other instruments that measure related aspects.The analyses were performed using the e SPSS Statistical Package, version 23.0, for Windows (IBM, Armonk, NY, USA) and the AMOS 22 Program (IBM, Chicago, IL, USA).In the first place, the data show that the items in the original ADAPTA-17GF (general factor) model have a distribution within the limits of normality according to the criteria of Finney and DiStefano [63], for whom 2 and 7 are the maximums permitted for skewness and kurtosis, which in our case were 1.28 and 2.74, respectively (Table 1).Table 2 shows the fit of the various questionnaire models according to the original ADAPTA 17 model (with a general adaptation factor and two other factors: emotional and cognitive-behavioral). This model was re-specified considering theoretical and statistical criteria (indices of change, errors of estimation, standardized errors of measurement).It may be observed that both the original 17-item model and the 12-item model show values that could be improved. The two-factor model with a general adaptation factor and 10 items is the best one after analysis. Thus, the ADAPTA-10GF Model showed much better fit in the calibration sample. There is also a smaller difference between the AIC default model = 141,996 and the AIC Saturated model = 110,000, showing that it is probably the best model according to the Akaike model selection criteria.The Principal Components Analysis revealed the existence of two components with eigenvalues over 1. The Scree Test showed the presence of two factors (Figure 2). Thus, we see in that in Table 3, there are two components corresponding to the Emotional Component and the Cognitive-Behavioral Component in the original model, with five items each, all with weights over 0.65, and explaining 59.55% of the variance (Table 3).Reliability of the model was analyzed with the Spearman-Brown coefficient p = 0.73 and the Cronbach’s Alpha, which for the whole scale was α = 0.84. The intraclass correlation coefficient (ICC) and its confidence interval (CI) were used for the analysis of temporal stability, with the following results for adaptation to change: 0.84 (CI = 0.82–0.86).Confirmatory Factor Analysis data for the model proposed (Figure 3) with the validation sample (n = 583) showed the following measures of fit: χ2/df = 3.21, CFI = 0.970, TLI = 0.956 and RMSEA= 0.062 (0.048–0.076), which were all adequate.The values in Table 4 for the six different models in the analysis of variance across sex show that in all cases, the ΔCFI is less than 0.01, and therefore, configural, metric, strict and strong invariance are accepted.With regard to construct validity, Figure 4 shows that the correlations in the direct scores on the GHQ-28 health questionnaires and the ADAPTA-10 questionnaire are significant (p < 0.01) and negative in all cases, backing the validity of the ADAPTA-10 construct. A higher score on the GHQ-28 shows more problems in each of the health dimensions.Adaptation to change is an important concept in psychology, as it depends on the adjustment of functioning and the responses with which one copes with the diversity of environmental demands [9,17]. Its absence has been related to psychological alterations [18]. Given the speed with which daily scenarios vary and the number of novel demands which must be coped with in short periods of time, knowing the effects of the capacity of adaptation to change of the population may be beneficial to both immediate and long-term psychological health [6]. In this respect, models have been proposed to establish the dimensions and factors that intervene in the process of adaptation to change, but linked to transcendental life events (such as the appearance of a disability, birth of a child or death of a spouse) or employment demands [46,47]. This study proposed construction of a model of adaptation to change in everyday events and circumstances, which would enable a scale to be designed for evaluating the response to challenges and changing circumstances.In the validation of the Adaptation to Change Questionnaire, ADAPTA-10, in the general population, the exploratory and confirmatory factor analyses showed the existence of the two dimensions previously found in the original model: emotional (linked to feelings that arise during adaptation) and cognitive-behavioral (related to cognitive management and behaviors for that purpose). Based on the factors analyzed and according to Bjorklund [9], it seems that the capacity for adapting to change includes both types of response, which would be in line with the model that showed the best fit. However, although the index of this two-factor model was adequate, after performing the corresponding re-specifications following theoretical and statistical criteria, seven items were eliminated from different factors in the original model. Specifically, items pertaining to the social support, stress management, locus of control and flexibility items were eliminated. In social support, the item eliminated may have been related, as mentioned by Kim et al. [43], with support being a mediator in the individual adaptation process, positively promoting one’s resources to cope with challenges, but not as a factor directly involved in this capacity. With regard to the stress management item, it may not be part of the validation process, since in situations in which the response must be rapidly modified or adjusted, a certain level of stress can eliminate lethargy or paralysis and generate the drive necessary to make the appropriate modifications. Furthermore, as items referring to emotional management were entered, negative thoughts and feelings that could arise with the appearance of stress (such as anxiety, irritability and so forth) and diminish the capacity for adaptation, could have been covered by that factor. Concerning the locus of control, even though items related to internal and external control which could diminish the capacity for adaptation were included, they did not form part of the final model. This may have been due to the perception of control and cause of events, although generating a stronger feeling of capacity for managing situations [22,26] may not be directly related to one’s possibility to adapt. That is, the capacity for adjusting to daily situations may be independent of the control that one feels one has over them. Finally, the item referring to cognitive flexibility was also eliminated from the two-factor, ten-item model. This result may be due to its being a relevant factor or variable with a heavy weight which acts as a mediator in the adaptation process but does not affect it directly. However, due to the wide presence in the scientific literature of this factor as the one which provides the most possibility of modifying strategies and reactions to meet changing demands to ensure adjustment [37,38,39,40], this relationship must be reexamined in the future.Therefore, based on the results, two dimensions were extracted from the ADAPTA-10 questionnaire. The first of these, the emotional dimension, would be linked to feelings experienced during adaptation. The five items that form part of this dimension pertain to the anxiety, depression and tolerance to uncertainty factors. Studies have shown the presence of anxiety symptoms, such as agitation from the need for adaptation to new demands [30]. Feelings of dysphoria and depressive symptoms are also present when the challenges one is faced with put the capacity to cope effectively with them to the test [36]. The cognitive-behavioral dimension refers to competence for managing and undertaking action to respond appropriately to daily situations that can be challenging. State of alertness enables conflictive situations to be detected and directs one’s attention toward specific objectives that must be met or problems that must be solved [41]. Coping concentrated on the problem means that efforts made are directly related to modifying the situation or its meaning, enabling its functional and adaptive management [24,25]. The emotional management factor may facilitate the regulation of negative feelings that appear because of the uncertainty, threat or perceived inability [35]. In the end, tolerance to uncertainty, as mentioned, formed part of the two dimensions through two different items. These referred to the emotional reaction and behavior in situations where one does not have all the information [29].Thus, the Adaptation to Change Questionnaire, ADAPTA-10, is a short instrument, easily applied, which enables finding out the individual’s ability to adjust the best way possible to new demands based on two dimensions. Even so, there are some limitations. It should be mentioned that most of the sample was made up of women, although the questionnaire showed good invariance across sex, and could be reflecting populational characteristics in Spain. Another limitation derived from the way data were collected, as the mean age was low with respect to the reality of the Spanish population, since fewer older people use the new technology tools with which the questionnaire was publicized and data were collected. In future, when the health situation so permits, these age groups should be approached to include more such subjects, although as observed in the section on participants, older people also answered correctly. Future research could validate our findings even more through the use of a more general sample. Another of the limitations is derived from the study design, because, as a cross-sectional study, there were variables which could not be controlled. The performance of a longitudinal study would solve this limitation by evaluating longitudinal invariance of the questionnaire.Although this is not a tool specific to COVID-19, it is a contextualized tool, so it would be necessary to analyze it again when the special situation of the health emergency ends. Meanwhile, its use along with other instruments evaluating psychological variables in the context of the COVID-19 pandemic can have very useful clinical applications. Evaluation of threat [52] or perceived risk from COVID-19 [64,65,66], combined with the capacity for adaptation to change, can help develop risk profiles and mental health protection measures in the mid- to long-term.The Adaptation to Change Questionnaire, ADAPTA-10, for the general population possesses favorable psychometric properties. The internal consistency of both the total scale and the two factors (emotional and cognitive-behavioral) is adequate, and therefore, the general fit is acceptable. However, it is recommended that the goodness and fit of the model for testing the psychometric properties of the instrument continue to be analyzed in other specific collectives or contexts. The construction of this scale can contribute to the analysis of the consequences associated with the presence of low adaptability to change. The analysis of this construct emerged during the pandemic from the SARS-CoV-2 coronavirus, which has been mentioned by various authors as both a physical and psychological health emergency due to the high impact of the illness on people’s daily lives. This is because, to a greater or lesser extent, everyone must adapt to a highly changing environment. The absence of the capacity to recover one’s previous state of wellbeing in transcendental life circumstances has shown to have long-term psychological effects. This scale can provide further knowledge of this ability and its repercussions in uncertain everyday situations, not necessarily linked to such events. It can also be valid for establishing the level of this variable in individuals, enabling development of intervention programs to strengthen adaptability, and thereby, promote better adjustment to demands. Therefore, the psychometric indicators, both for the factors and the global scale, reveal that it is a reliable, valid measurement instrument for use in research. Likewise, it is thought that it can be of maximum usefulness for the prevention and early diagnosis of problems related to mental health (such as depression, anxiety, development of health-risk behaviors or use of maladaptive coping strategies) in the general population derived from poor adaptation to adverse situations, similar to those triggered by the COVID-19 pandemic.M.d.C.P.-F., M.d.M.M.J. and Á.M.M. contributed to the concept, design, analysis and interpretation of the data. M.d.M.S.M. contributed to the technical details and manuscript preparation. E.F.-M., R.F.V., I.H.-P., D.J.-R., I.M.M. and A.S.G. contributed to collecting the data. J.J.G.L. contributed to critically revising the manuscript for important intellectual content and the final approval of the version to be published. All authors accept and agree that the work is original, any methods and data presented are described accurately and honestly and any relevant interests have been disclosed. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The present study was undertaken in collaboration with Excma, Diputación Provincial de Almería.The authors declare no conflict of interest.Theoretical explanatory model of the adaptability to change construct.Scree plot of factor analysis for the ADAPTA-10 GF Model.Confirmatory factor analysis ADAPTA-10GF Model (N = 583).ADAPTA-10 questionnaire correlations and General Health Questionnaire-28.Descriptive statistics. Calibration sample (n = 578). SD: standard deviation, Std. Error: standard error.Fit indices for the models proposed (calibration sample n = 578).CFI = Comparative fit index; TLI = Tucker-Lewis index; RMR = Root mean square residual; RMSEA = Root Mean Square Error of Approximation; CI = Confidence Interval; df = Degrees of freedom; Est. = Estimation.Factor structure, communalities (h2) eigenvalues, Cronbach’s alpha and percentage of explained variance (n = 583). Extraction method: Factoring of principal components.Multigroup analysis of invariance across sex (men/women).FS = Factor saturations, Int = Intercepts, Err = Errors.

Med-MDPI/ijerph_5/ijerph-17-15-05613.txt

@@ -0,0 +1,13 @@
+Commissioning workers at nuclear power plants have long been ignored in previous studies, although their performance is closely related to the overall safety of plants. This study aimed to explain and predict three types of behavior, i.e., errors, violations, and safety participation, of commissioning workers, under the general framework of the theory of planned behavior (TPB) and by considering organization and planning factors. The validity of the model was evaluated with a sample of 167 commissioning workers who completed a self-reported questionnaire. The results showed that perceived behavioral control, along with organization and planning, significantly affected all types of behavior. It was also found that violations and errors were a direct result of attitude. Besides, errors were predicted by subjective norm; unexpectedly, this occurred in a positive way. These findings revealed the underlying mechanisms for the development of errors, violations, and safety participation among commissioning workers and provided practical implications for safety improvement at the commissioning workplace.The burning of fossil fuels emits large quantities of air pollutants that are harmful to the environment and public health [1,2]. To cope with these adverse impacts, many countries have tried to reduce their reliance on fossil fuels by developing clean and renewable energy, such as nuclear, wind, solar, and geothermal energy. Among these emerging energy sources, nuclear energy has gained much attention for its relatively low cost and stable production rate [3]. By 2017, there were about 450 nuclear power reactors worldwide, providing over 10% of the world’s electricity [4]. These numbers will keep increasing, as many new nuclear power plants are on order or planned [4], notably in Asian countries such as China. China recently has restarted the construction of nuclear power plants after a two-year hiatus [5], with about 15 under construction and more about to start construction. With the intense development, commissioning, one of the key phases prior to the start of nuclear power plants’ commercial operation, has raised much attention recently.According to the International Atomic Energy Agency [6], commissioning is “the process by means of which systems and components of facilities and activities, having been constructed, are made operational and verified to be in accordance with the design and to have met the required performance criteria”. Commissioning is a safety-critical phase in the nuclear power plant lifecycle because it aims at noticing and fixing all deficiencies and possible errors before the plant commences operation. Great challenges lie in commissioning phase, such as the pressure of completing work within a limited time, the need to test new features of the plants, and the possible unexpected emergencies with regard to the fuel loading stage [7]. Incidents or accidents during commissioning can lead to consequences ranging from minor (e.g., malfunction of devices) to severe (e.g., prolonged completion time and even excessive radiation leakage). Much evidence in occupational safety literature showed that the aberrant behavior of workers, such as errors and violations, was one important cause of accidents [8,9], while workers’ safety participation behavior significantly reduced accident risk [10,11]. These findings indicate that the behavioral performance of the commissioning workers is of great importance to the safety and efficiency of commissioning phase and to the future safety of nuclear power plants when they are in operation.Numerous studies have been conducted and many theories have been proposed to explain individuals’ aberrant and safety participation behavior [12]. The investigated worker groups include construction workers [13], aviation maintainers [12], and oilrig workers [14]. Workers at nuclear power plants have also attracted some research attention, but the focus has been mostly on control room operators [15,16]. No research has focused on the behavior of commissioning workers. There is a great difference between commissioning workers and control room operators in terms of core tasks and ways of organization. For instance, the main tasks at control room are to monitor the status of nuclear power units and handle abnormities based on specific regulations. The main commissioning tasks, in contrast, are to start up the nuclear power units and conduct experiments to examine if the units can run normally, and such tasks are filled with uncertainties [7]. Another significant difference is that the control room usually has stable team members, while commissioning tends to build flexible teams based on specific tasks [17]. The complex and challenging features of commissioning may make the behavioral characteristics of commissioning workers quite different from those of control room operators. Therefore, studies focusing on commissioning workers are needed to understand the mechanisms behind behavior during commissioning.Another research gap is that most of the available studies mentioned above emphasized only the aberrant behaviors and ignored workers’ intent and enthusiasm for proactively engaging in safety-promoting behavior. Such voluntary behavior, often referred to as safety participation, extends beyond compliance with safety regulations and is not formally rewarded by the organization but can be beneficial to the overall safety of the organization [18,19]. To fill these gaps, this study aimed to propose models to predict and explain errors, violations, and safety participation behavior of commissioning workers. Related findings not only shed theoretical light on understanding antecedents of commissioning workers’ behavior, but also provide guidelines to policymakers when designing countermeasures to advocate or reduce certain types of behavior. The remainder of the introduction describes the theoretical foundation underlying this study and the model and hypotheses proposed.Aberrant behavior, which refers to a straying from the path [20], could be classified into two types based on whether the behavior is intentional or not. Instances of intentional aberrant behavior are traditionally referred to as violations, which involve deliberate deviation from rules or practices that are important in maintaining the safety of a particular task or job [12]. This is opposed to errors, which refer to unintended outcomes caused by slips, lapses, and mistakes made by individuals [12,20]. Both types of behavior are shaped by cognitive, psychological, and social factors. However, the effect sizes of these factors on violations and errors are different; the former are more closely associated with social and psychological factors such as attitude, while the latter are more strongly affected by deficiencies in cognitive abilities such as information processing efficiency and organization skills [21]. They also differ in their associations with demographic variables such as age and gender [22] and in their contributions to accidents [23,24].While compliance with safety rules and regulations is important in lowering the risk of accidents, organizations also need individuals’ proactive participation in safety [18]. Safety participation refers to an employee’s voluntary participation in safety activities that is beyond the employee’s formal role but does contribute to the development of a supportive safety environment [25]. Examples of safety participation include promoting safety programs within the workplace, helping coworkers, and raising safety concerns [26,27]. Safety participation has been reported to be a significant predictor of occupational accidents [10,28], the effect size of which is even greater than that of aberrant behavior [27].Many theories have been proposed and applied to explain human behavior [29]. For instance, theory of self-efficacy posits that two types of expectancies, i.e., outcome expectancy and self-efficacy expectancy, exert powerful influence on individual behavior [30]. Self-determination theory, by contrast, suggests that motivations are the strongest determinants in shaping who we are and how we behave. In particular, individuals are motivated to engage in certain activities when their needs for competence, connection, and autonomy are fulfilled [31]. Another well-recognized theory that tries to explain human behavior from a social cognitive perspective is the theory of planned behavior (TPB) proposed by Ajzen [32]. It is derived from the original theory of reasoned action (TRA) [33]. TRA was developed based on the premise that individuals make reasoned decisions to engage in specific behavior by evaluating the information available to them. It proposes that two factors, attitude and subjective norm, directly determine an individual’s behavioral intention. Attitude refers to the degree to which a person has a favorable or unfavorable evaluation or appraisal of the behavior in question, while subjective norm refers to the perceived social pressure to perform or not to perform the behavior [33]. The formed intention further decides whether the individual would actually do that behavior.Although intention is a strong predictor of actual behavior, there are times when individuals do not execute an action, despite having the intention to act in a certain manner, because of external factors that fall outside their control. As a result, TRA is limited to only predicting behavior of people who have complete volitional control [34]. To extend the explanatory scope, TPB was proposed; this theory incorporates perceived behavioral control, defined as the perceived ease or difficulty of performing the behavior [32], into the model. TPB posits that attitude, subjective norm, and perceived behavioral control together shape one’s intention to engage in certain behavior, and this intention, together with perceived behavioral control, determine the probability that an actual action will be taken. As a general rule, the more favorable the attitude and subjective norm with respect to the behavior are, the greater the perceived behavioral control is, and the stronger an individual’s intention to perform the behavior should be [32]. Recent evidence suggested that perceived behavioral control and subjective norm could also impact behavior indirectly by influencing attitude [12,35,36].As a type of intentional or planned behavior, violations at workplace should follow the framework in TPB. This has been confirmed by some empirical studies that reported the effectiveness of TPB in explaining violations in aviation maintenance [12], construction worksites [37], and road transportation [38,39,40]. For instance, Fogarty and Shaw [12] reported that a positive attitude towards violations and the perception that other people in work commit violations (i.e., subjective norm) significantly increased workers’ intention to violate and the actual occurrence of violations. Similarly, Wang et al. [40] found that all factors of TPB showed significant effects on violations in lane changing behavior. The other type of aberrant behavior, errors, seem to fall outside of TPB’s explanatory scope given that they are unintentional or unplanned behavior. However, some factors of TPB have been reported as significant predictors of errors in safety literature. For instance, Victoir et al. [41] found that perceived behavioral control was a dominant determinant that explained 33% of variance in driving errors. Lucidi et al. [42] and Mallia et al. [43] found that drivers possessing more a positive safety attitude reported fewer errors, suggesting that attitude towards traffic safety rules was a significant predictor of driving errors. Paletz et al. [44] proposed taking subjective norm into consideration when analyzing errors. Based on the above studies, it was hypothesized that TPB should have some explanatory power in regard to errors. Finally, although the motivating mechanism of safety participation has not been intensively investigated, there has been evidence suggesting that it falls under the explanatory power of TPB. Fugas et al. [39] applied TPB as the framework to predict safety participation behavior at a transportation organization. They found that subjective norm was the most significant predictor of safety participation behavior of transportation organization workers, followed by safety attitude. However, in their study, perceived behavioral control was not identified as a significant determinant. A more recent study [45] investigating safety participation behavior at construction sites showed that all three factors of TPB were significant predictors.Taking all evidence together, this study applied TPB as the theoretical framework to predict errors, violations, and safety participation behavior of commissioning workers. It should be noted though, that when predicting a specific type of behavior, most TPB-related studies have measured the perception or attitude towards this specific behavior. For instance, attitudes towards violations were measured when using TPB to predict violations [12]. However, instead of measuring the specific perception towards each type of behavior, this study used a general perception towards safety regulations to predict the three types of behavior. That is, we have measured the attitude, subjective norm, and perceived behavioral control towards safety regulations.Based on the above evidence, it was hypothesized that:
+            Attitude would have a direct negative effect on errors and violations and a direct positive effect on safety participation.
+          
+            Subjective norm would have a direct negative effect on errors and violations and a direct positive effect on safety participation.
+          
+            Perceived behavioral control would have a direct negative effect on errors and violations and a direct positive effect on safety participation.
+          With regard to the relations among the three predictors, TPB assumes that they are correlated. However, many empirical studies have demonstrated that subjective norm and perceived behavioral control are predictors of attitude [12,35]; therefore, it was further hypothesized that:
+            Subjective norm would have a direct positive effect on attitude.
+          
+            Perceived behavioral control would have a positive effect attitude.
+          TPB offers theoretical explanations in terms of why certain behavior would be planned and whether the planned behavior would be carried out or not. However, whether a behavior can be executed the way it is planned also depends on individuals’ executive ability. For instance, there are situations where individuals intend to comply with rules but end up failing to do so due to a low level of executive function [34,46,47,48]. Executive function includes many aspects relevant to successful task completion, such as organization and planning, inhibiting responses, thinking abstractly, and reallocating mental resources [47]. In the context of commissioning, the aspect of executive function most relevant to task completion is probably organization and planning. Organization and planning refer to the ability to think ahead and to carry out organized behavior through functions like multitasking, sequencing, and holding information in mind to make decisions [49]. As mentioned above, commissioning is characterized by collaborations within and across teams, in addition to a tight schedule, all of which require organization and planning ability to coordinate different parties and manage the projects [7]. Some empirical studies found that organization and planning ability could reduce errors associated with ineffective communication or planning and reduce regulation violations due to inappropriate organization of tasks [46]. Although little attention has been paid to the effect of organization and planning ability on safety participation, it was expected that better organization and planning ability could help workers improve work efficiency, therefore giving them more time and energy to engage in voluntary safety promotion programs. Based on the above evidence, it is therefore hypothesized that: 
+            Organization and planning would have a positive effect on errors and violations and a negative effect on safety participation.
+          The proposed model is shown in Figure 1. Utilizing the TPB framework, it is hypothesized that attitude, perceived behavioral control, and subjective norm would directly impact the occurrence of errors, violations, and safety participation behavior. Besides, subjective norm and perceived behavioral control could indirectly affect workers’ behavior through the attitude factor. Moreover, organization and planning ability is a direct determinant of commissioning workers’ behavior.Data used in this study were extracted from a large questionnaire survey investigating the relationship between personality, cognitive abilities, and performance of commissioning workers. The questionnaire was designed after an extensive review of the literature and revised based on the feedback from three commissioning experts to improve its clarity and readability. Three experts on human factors were also consulted to guarantee the quality of the questionnaire. Questions related to this study are introduced below and listed in detail in Table A1 in Appendix A.Basic information inventory: This part was designed to collect demographic information including age, gender, educational level, marriage status, working experience, and working division (i.e., technique management, electrical commissioning, conventional island commissioning, instrumentation and control commissioning, nuclear island commissioning, or other).Behavior: This part required respondents to rate their frequency of errors, violations, and active safety participation behavior at work on a 5-point Likert scale ranging from 1 = “never” to 5 = “always”. Seven items from Rao et al. [50] were adopted to measure errors (e.g., “I promised to return to someone with information but forgot to do so.”), and three items from the same source as above were used to measure violations (e.g., “sometimes I do not use the correct safety procedures for carrying out my job”). Based on its definition and discussion with commissioning experts, three items were developed to measure safety participation (e.g., “I would alert my co-workers if potential safety concerns exist”).TPB factors: This section was used to collect respondents’ perceptions and attitudes towards safety regulations at work. In particular, subjective norm was measured by adjusting the three items from Venkatesh and Davis [51] (e.g., “people who influence me a lot think I should follow rules and procedures at work”); perceived behavioral control was measured by adapting the five items from Taylor and Todd [52] (e.g., “I am able to follow rules and procedures at work”); and attitude was measured using three items from Iversen [53] (e.g., “many rules must be ignored to ensure work flow”). Items were also rated with a 5-point Likert scale ranging from 1 = “totally disagree” to 5 = “totally agree”.Organization and planning: This factor was measured using the twelve items from the Executive Function Index (EFI) developed by Spinella [49]. Respondents were required to indicate the extent to which they agreed with descriptions of themselves in daily life (e.g., “organized person”). Similarly, a 5-point Likert scale ranging from 1 = “totally disagree” to 5 = “totally agree” was used by respondents to rate their answers.Face-to-face invitation to participate in the survey was given by visiting the commissioning offices at Yangjiang and Taishan nuclear power plants in July of 2019. These two nuclear power plants are located in Guangdong Province in China, and both of them belong to the General Nuclear Power Corporation. Therefore, they are comparable in organizational structures, work regulations, and work environments. The ethics approval is included as parts of the official contract of the collaboration project between Shenzhen University and the State Key Laboratory of Nuclear Power Safety Monitoring Technology and Equipment of China. The ethic approval code is 007-EC-B-2019-C83-P.S.20-01122. Participants were informed that they could quit the survey at any time and all data collected would be handled anonymously and confidentially and only be used for research purposes. A total of 187 questionnaires were distributed, and 179 were returned within the required time period. Of the returned questionnaires, 12 were excluded from further analysis because of the high proportion of unanswered questions (>20%). Data from the remaining 167 questionnaires were used in this study.Of the valid responses, 130 (77.8%) were from Yangjiang nuclear power plant, and 37 (22.2%) were from Taishan nuclear power plant. Almost all respondents (n = 165, 98.8%) were males. On average, they were 32.5 years old (SD = 4.8), and they had an average working experience of 7.7 years (SD = 2.4). The majority of the respondents were married (n = 123, 73.6%). With regard to educational level, the majority (n = 146, 88.0%) had a bachelor’s degree, a few had a master’s degree (n = 13, 7.8%), and the rest had a college degree or below (n = 7, 4.2%). Given that the distribution of gender, marriage status, and educational level variables were highly skewed, their possible confounding effects on the proposed model were not investigated. The workers were distributed in different divisions: technique management (n = 21, 12.5%), electrical commissioning (n = 30, 18.0%), conventional island commissioning (n = 39, 23.3%), instrumentation and control commissioning (n = 36, 21.6%), nuclear island commissioning (n = 34, 20.4%), and other (n = 7, 4.2%).A two-step procedure for conducting structural equation modelling (SEM), as proposed by Anderson and Gerbing [54], was adopted. In the first step, a series of tests were carried out to evaluate the reliability and validity of the measurement model. In particular, internal consistency was evaluated with Cronbach’s α. A value of Cronbach’s α greater than 0.6 indicates acceptable internal consistency [55]. Confirmatory factor analysis (CFA) was used to examine whether multiple items of the same factor were in agreement, i.e., the convergent validity. Convergent validity is achieved when the model shows a good fit and the factor loading of an item on its posted underlying factor is significant and larger than 0.5 [56]. The model was considered to have a good fit when the following criteria were fulfilled: ratio of chi-square value to degree of freedom (χ2/df) < 3, comparative fit index (CFI) ≥ 0.90, Tucker–Lewis index (TLI) ≥ 0.90, incremental fit index (IFI) ≥ 0.90, standardized root-mean-square residual (SRMR) < 0.08, and root-mean-square error of approximation (RMSEA) < 0.06 [57,58,59,60]. Discriminant validity reflects the extent to which the factors differ from one another empirically [56]. According to the Fornell and Larcker criterion, discriminant validity is achieved if the square root of average variance extracted (SAVE) for each of the factors is greater than any of the bivariate correlations involving the factor in the model [61].In the second step, SEM was used to assess the goodness of fit of the proposed model and to investigate the strength of the relationships among factors in the model (i.e., hypothesis testing). The same goodness-of-fit criteria (i.e., χ2/df < 3, CFI ≥ 0.90, TLI ≥ 0.90, IFI ≥ 0.90, SRMR < 0.08, and RMSEA < 0.06) were applied to evaluate the fit of the proposed model. All analyses were performed in R software (R Foundation for Statistical Computing, Vienna, Austria) (R Core Team. (2016). R: A Language and Environment for Statistical Computing. Vienna, Austria. Retrieved from http://www.R-project.org/).The CFA results showed that the factor loadings of five items from the organization and planning factors (marked in the Table A1) were smaller than the required minimum value of 0.5 and therefore were removed from further analysis. The CFA was then reconducted. The revised model showed a satisfactory goodness of fit (χ2/df = 1.6, CFI = 0.91, TLI = 0.90, IFI = 0.91, SRMR = 0.072, and RMSEA = 0.059; see Table 1). All items in the revised model had factor loadings larger than 0.5, suggesting an acceptable convergent validity of the factors. All Cronbach’s α values were greater than 0.6 (Table 1), suggesting a good internal consistency of these factors. The SAVE (shown on the diagonal in Table 1) of each factor was greater than the associated interfactor correlations (shown off the diagonal in Table 2), indicating acceptable discriminant validity. In addition, all the interfactor correlations are smaller than 0.85, suggesting that multicollinearity is not a problem [62]. In summary, the measurement model showed a satisfactory reliability and validity and was appropriate for the analysis of the structure model. Furthermore, given that age and working experience showed no significant effects on variables in the proposed model, their effects were not controlled in the following SEM analysis.The SEM results suggested that our model had a satisfactory goodness of fit, with χ2/df = 1.6, CFI = 0.91, TLI = 0.90, IFI = 0.91, SRMR = 0.072, and RMSEA = 0.059. The estimated standardized path coefficients of the significant relationships in the model and the proportion of explained variance (R2) are shown in Figure 2. For clarity, models of the three kinds of behavior (i.e., errors, violations, and safety participation) are presented separately. For relationships within the three factors from TPB, attitude was significantly predicted by perceived behavioral control (β = 0.338, p < 0.001), with a higher level of perceived behavioral control over following safety rules leading to better attitude. The effect of subjective norm on attitude did not reach a significant level.Of the three types of behavior, errors were directly determined by attitude (β = −0.203, p = 0.036), subjective norm (β = 0.227, p = 0.009), and organization and planning (β = −0.368, p = 0.012). As hypothesized, workers with better attitude and a higher organization and planning ability had fewer errors. Perceived behavioral control did not have a significant direct effect on errors (β = 0.042, p = 0.736). However, it showed a significant effect on attitude, which further significantly predicted errors. Therefore, perceived behavioral control could indirectly influence the occurrence of errors by affecting attitude. Unexpectedly, workers who perceived a higher norm that they should follow rules and procedures were more likely to make errors. A total of 20.5% of the variance in errors was explained by the proposed model.With regard to violations, the path coefficients of attitude (β = −0.191, p = 0.036) and organization and planning (β = −0.503, p = 0.001) were significant. Again, perceived behavioral control did not show a direct effect on violations (β = −0.049, p = 0.694) and could only indirectly influence it through attitude. Subjective norm was not a significant predictor of violations either (β = −0.005, p = 0.953). A total of 38.0% of the variance in violations was explained.For safety participation, the most significant predictor was perceived behavioral control (β = 0.407, p = 0.002), followed by organization and planning (β = 0.352, p = 0.012). These two factors explained as high as 49.1% of the variance in safety participation. Attitude (β = 0.055, p = 0.549) and subjective norm (β = −0.093, p = 0.252) did not show any significant effects in shaping workers’ safety participation behavior. Table 3 summarizes the results of the hypothesis tests.Commissioning workers have long been ignored in previous studies although their performance is closely related to the overall safety of nuclear power plants. This study aimed to investigate the determinants of three types of behavior (i.e., errors, violations, and safety participation) of commissioning workers, under the general framework of TPB and by considering organization and planning factors. Our work demonstrated the usefulness of TPB in explaining commissioning workers’ behavior and revealed the working mechanisms under the three types of investigated behavior.All three types of behavior were significantly predicted by workers’ organization and planning ability. First, commissioning workers with a better organization and planning ability reported fewer errors. This is probably because those good at multitasking, sequencing, and planning are better prepared to deal with the pressure of time and the challenge of technological complexity during commissioning work, which have been reported in literature as major contributors to errors [12,63]. Another possible explanation is that those with a better organization and planning ability are less likely to engage in distracting activities [64] and therefore may commit fewer errors associated with distraction. Second, consistent with findings in the context of driving [46,48], our results show that violations were negatively associated with organization and planning. According to Lund and Rundmo [65], some violations are a result of underestimation of possible harmful consequences associated with such behavior. Organized persons, who carry out careful and comprehensive evaluation before taking action, are more likely to realize the negative consequences of violations and therefore less likely to violate rules. Finally, active safety participation was positively associated with organization and planning ability, probably because those with clear plans for the future would devote greater efforts to activities that are beyond their formal role but can contribute to realizing their future goals, like safety participation activities. Together, our results demonstrate that organization and planning affect all kinds of behavior and are critical to the overall safety of commissioning work.A favorable attitude towards rules and procedures resulted in fewer errors and violations but did not promote voluntary safety participation behavior. This result agrees with previous evidence that violations are deliberate and planned behavior and such behavior is a direct result of attitude [12,37,40]. Errors, which are supposed to be unintended or unplanned behavior, were also partially determined by attitude. This is probably because those showing a positive attitude towards safety rules and procedures pay more attention to work, which helps reduce the probabilities of temporary failures of concentration, memory, or judgement that may cause errors. Surprisingly, contrary to previous results [39,66], better attitude did not promote safety participation, but better perceived behavioral control and organization and planning did. These results in combination suggest that active involvement in safety promotion depends on organizational and individual support, not on individual preference.The three types of behavior showed different relationships with subjective norm. On one hand, behavior with clear intention, i.e., violation and safety participation, was not predicted by subjective norm in this study. This finding is inconsistent with previous studies which have found that behavior at workplaces like those of transportation [39,40] and construction [67,68] was related to subjective norm. This unexpected finding is probably a result of the specialization of commissioning work. Commissioning work requires specified knowledge and professional training, and workers might believe that others, especially those not working in commissioning, know very little about their work. Therefore, commissioning workers may not take his family members’ or friends’ opinions into consideration when planning intended behavior, i.e., violations and safety participation. On the other hand, subjective norm positively impacted the occurrence of errors. A possible explanation is that while perceived norm did not impact intended decisions, it is a kind of pressure that might lead to unintentional lapses and slips [69]. Finally, perceived behavioral control was a critical factor of commissioning safety, as it affected all three types of behavior. Consistent with much evidence in literature [70], those who perceived themselves as more capable of following rules and procedures at work had a more positive attitude towards rules and further reported fewer errors and violations. More importantly, our results showed that commissioning workers who had higher perceived behavioral control were more likely to actively participate in safety promotion activities. This suggests that, besides the widely recognized factors such as supervisor leadership [71,72] and safety climate [73,74], perceived behavioral control should also be considered when developing interventions to promote workers’ safety participation.This study is one of the first studies to investigate the behavior of commissioning workers at nuclear power plants. The main findings include that (1) all three types of investigated behavior (i.e., errors, violations, and safety participation) are significantly predicted by perceived behavioral control and organization and planning; (2) violations and errors are a direct result of attitude towards safety rules and procedures, while safety participation is not related to such attitude; and (3) errors are further positively predicted by subjective norm. These findings revealed the underlying mechanisms for the development of errors, violations, and safety participation among commissioning workers and provided practical implications for safety improvement at the commissioning workplace.Findings of this study have several practical implications for safety improvement at the commissioning workplace. Given the important and positive role of organization and planning in shaping commissioning workers’ behavior, it is recommended that organization and planning ability could be used as a reference when hiring commissioning workers or be considered as content for training in worker improvement programs. In addition, ways to improve workers’ feelings of being in control of their behavior, such as making safety equipment easier to use or offering enough personnel trainings, are of great importance for reducing violations and errors and promoting safety participation. Another remediation of errors is to reduce the level of perceived subjective norm, given that this factor was identified as a significant positive predictor of errors. Practically, this can be achieved by avoiding the emphasis of work regulations by friends and family members.Conceptualization, T.Z., D.T., X.Q., and Z.L.; Formal analysis, T.Z. and S.Z.; Funding acquisition, X.Q. and Z.L.; Investigation, T.Z., S.Z., D.T., and X.Q.; Methodology, T.Z. and D.T.; Project administration, T.Z. and D.T.; Supervision, X.Q. and D.T; Validation, D.T. and T.Z.; Writing—original draft, T.Z. and S.Z.; Writing—review & editing, T.Z., D.T., and X.Q. All authors have read and agreed to the published version of the manuscript.This work was supported by the National Natural Science Foundation of China (Grant No. 71801156), the Startup Foundation of SZU (Grant No. 000002110284) and the Natural Science Foundation of Guangdong Province, China (Grant No. 2019A1515010863).The authors declare no conflict of interest.List of items used in this study.*: Items reverse coded; #: Items excluded from the final model.Proposed model for explaining errors, violations, and safety participation behavior of commissioning workers.The final model and the significant standardized path coefficients, (a) result of SEM analysis for errors, (b) result of SEM analysis for violations, (c) result of SEM analysis for safety participation. * p < 0.05; ** p < 0.01; *** p < 0.001; dotted line represents a nonsignificant path parameter.Fit indices for the tested models.Mean, standard deviation (SD), and zero-order correlation between factors.Note: The numbers in parentheses are the SAVEs. The off-diagonal elements are the correlations between the factors. ** p < 0.01; *** p < 0.001.Summary of the hypothesis test results.

Med-MDPI/ijerph_5/ijerph-17-15-05614.txt

@@ -0,0 +1,15 @@
+The availability of sufficient and trustworthy energy services at the reasonable cost in a securely and environmentally friendly manner, and conventionality with economic and social development requirements, is an important factor of sustainable development (SD). Energy plays a significant role in eliminating poverty and increasing living standards. However, most of the present energy forms of energy supply and consumption are unsustainable. This paper analyzes the association between economic growth (EG), energy consumption (EC), and sustainable development (SD) among other economic factors. The sample of 14 developed and developing member states of the Union for the Mediterranean (UFM) was selected. To deal with the endogeneity issue, the system- generalized method of moment (GMM) model was employed. Moreover, panel co-integration, Granger causality tests, and robustness tests were employed to examine the long-run and short-run causality among variables of interest. The results confirmed the short-run dynamic association from sustainable development (SD) to energy consumption (EC), and economic growth (EG) to sustainable development (SD). Moreover, the results validated the presence of long-run equilibrium association in the equations of EC and sustainable development (SD). The findings of this study will be supportive for the policymakers to formulate sustainable energy policies to stimulate the economic growth (EG) in the way of sustainable development (SD) in the UFM countries.This study aimed at investigating the dynamic association between economic growth (EG), energy consumption (EC), and sustainable development (SD) among other economic factors. We selected a sample of 14 developed and developing member states of the Union for the Mediterranean (UFM) and used data for the period 1995 to 2014.The main motivation of this study is the lack of definite policies to achieve sustainable development in the countries. Earlier growth models of developed countries are based on exhaustive use of natural resources, but these growth models and policies are being questioned today. The major flaw of those models is to ignore the effect of environmental issues on the development of countries. To attain sustainable development, it is essential to adopt a balanced approach with economic, social, and environmental aspects [1]. On the other side, developing countries are also in a similar way of development as adopted by developed countries, but the environment of these countries could be severally affected. In a sustainable development strategy, the energy consumption (EC) has great importance, therefore, many countries are adopting policies to save energy, and these policies are based on the association between EC and EG [2].The Mediterranean countries have energy benefits due to the extreme diversity in their energy resources, while the share of energy production of these countries is 11.4% as compared to the world energy production capacity. These countries cooperate related to energy production and distribution through the Maghreb Electricity Committee (Comelec). The other characteristics of Mediterranean countries are the disparity of energy consumption in the north-side and south-side. Energy consumption (EC) was reported double in the north side as compared to the south side of the Mediterranean region in 2009 [2].The Mediterranean region has great potential for utilizing wind and solar energy sources to fulfill the energy demand. The energy mix in this region is dominated by fossil fuel, while renewable energy sources are not exploited well. Mediterranean countries have recently taken actions for implementing the strategies such as Mediterranean Solar Plant (MSP) and the Mediterranean Strategy for Sustainable Development (MSSD) to cope with energy and environmental challenges. The development of renewable energy projects in this region can give a lot of benefits, such as fulfilling the energy demand at a lower cost, attaining the sustainable economic growth, generating new employment opportunities, increasing the environmental quality, and increasing the cooperation between Mediterranean countries and European Union (EU) [3].Because of the diversity of energy resources and its consumption level, social and environmental aspects, there is a need to formulate common and comprehensive policies on different energy issues and appropriate infrastructure. The energy issue is the major challenge for the Mediterranean countries to attain sustainable development. In 2008, the Union for the Mediterranean (UFM) countries was formed to increase the cooperation between the countries of the Mediterranean region and the EU. The purpose of this cooperation was to deal with energy and environmental challenges [2].Previous studies have analyzed the connection between EC and EG both in developed and developing countries and found inconclusive findings. For example, Oh and Lee [4], Bowden and Payne [5], Karanfil [6] and Lise and Van Montfort [7] found unidirectional associations either from EG to EC or EC to EG. On the contrary, Belloumi [8] and Erdal, et al. [9] demonstrated a bidirectional association between EG and EC. This study focused on the UFM countries with the fact that these countries were given little attention in the literature. The literature on the causality between energy consumption (EC), economic growth (EG), sustainable development (SD), and other variables of UFM countries is quite limited as compared to other countries. However, to the best of our knowledge, no previous study has empirically examined the nexus of EC, EG, and sustainable development (SD), and this is the first systematic quantitative study that dealt with it.To complete this study, we used different analysis techniques. Initially, we applied system GMM to overcome the endogeneity issue. Moreover, we tested the stationary of each variable by employing the unit root tests. After finding co-integration in models by using a panel co-integration test, we employed vector error correction model (VECM) and vector autoregression (VAR) to examine the causation between EC, EG, and sustainable development (SD) among other economic factors both in the long-run and short-run. To find more robust results, we employed impulse response analysis and variance decomposition analysis.The following outcomes of this study are highlighted: EG and EC have shown bidirectional causality, while the sustainable development (SD) and EG have also shown bidirectional causality. Moreover, results have shown that sustainable development (SD) induces to EC in the short-run, while EG induces to sustainable development (SD) in the short-run. Furthermore, results demonstrated the existence of long-run equilibrium in the equations of EC and sustainable development (SD). These outcomes contribute to the literature by expanding the significance of EC and EG in the way of sustainable development (SD) in the UFM countries. This study could be helpful for these countries to reshape the policies for energy consumption, and other economic policies to find ways to increase economic growth and attain sustainable development (SD) in these countries.This paper is organized as follows: the Section 2 reviews the previous literature. The Section 3 describes the synopsis of UFM countries. The Section 4 provides details of data collection, sample selection, and econometric techniques used. The Section 5 discusses the results of the study. The last section concludes the study.Energy resources are generally considered compulsory for the development of society, while the sustainable development (SD) of society needs the supply of energy resources that are easily available for the long term at a reasonable cost and without harmful effects on society [10]. Energy helps in eradicating poverty and increasing the human welfare and living standards of people in society. However, the present forms of energy usage and its supply are considered unsustainable [11].Many regions of the world do not have sustainable and secure energy resources which bound the economic expansion, while in the other regions, environmental pollution from the energy consumption constrains sustainable development [12]. Some energy resources such as fossil fuels and uranium are considered limited, while energy resources such as water, wind, and sunlight are sustainable for a long time. Moreover, wastes and biomass fuels are also considered as sources of sustainable energy resources [10,13].Environmental concerns are the main issues that have to deal with countries in achieving sustainable development. Generally, the effect of environmental degrading activities is not sustainable for a long time, for instance, the cumulative effect of such activities on the environment creates problems related to health, ecological, and others. The major part of the environmental effects is connected with the consumption of energy resources. Preferably, societies want sustainable development with the consumption of energy resources that do not generate harmful environmental effects, but, in reality, all energy resources generate some harmful environmental effects. However, the harmful environmental effects can be overcome by increasing energy efficiency, because energy efficiency is strongly associated with environmental effects [14].Many social concerns are also related to energy consumption, which includes demographic transition, education, poverty, indoor pollution, quality of life, and gender and age-related implications. The social aspect of sustainable development related to energy is the availability of basic energy services in the shape of commercial energy to people all over the world at an affordable cost. Energy indicators of social aspects have more significance for the developing countries that still have substantial portions of the population deprived of modern energy services [12].The accessible and secure energy resources are essential for fortifying the economic growth. All sectors of the economy such as agricultural, residential, service, and others depend upon sustainable energy resources. Many economic activities including industrial growth, job opportunities, rural and urban development are strongly influenced by the energy contribution. The availability of electricity is most important in many production activities, distribution of information, and other industries. The energy indicators in the economic aspect reflect two themes: the ways of usage and production, and secure supply. The first theme of usage and production comprises of the sub-issues such as supply efficiency, usage, production, energy mix, and prices. The second theme of secure supply consists of reliance on supply and energy stocks [12].The discussion about the association between energy and EG is continued among economists for a long time, but they could not find conclusive evidence about it yet. On one side, neoclassical economists believe that energy is not an important factor that originates the EG, they argue that energy affects EG only in certain ways [15,16]. On the other side, ecological economists consider energy as an imperative factor of production in line with the Laws of Thermodynamics and proposed a model for it [17]. Afterward, many other researchers also endorsed their findings of the association between economic production and energy [18,19].The debate about the association between EC and EG revolves around four hypotheses: (1) Growth hypothesis, (2) Conservation hypothesis, (3) Neutral hypothesis, and (4) Feedback hypothesis.Growth Hypothesis: this hypothesis refers to the unidirectional causality between EC and EG, causality running from EC to EG. It infers that a decrease of EC may lead to a decline in EG, while an increase of EC can promote EG [20]. Besides the labor and capital, energy is an essential factor of output, while insufficient energy supply and energy supply shocks can restrict economic growth. From the empirical point of view, Soytas and Sari [21] found that Turkey has a one-way Granger causality from power consumption to manufacturing growth. Gurgul and Lach [22] used quarterly data from 2000 to 2009 to analyze the association between total EC and EG in Poland and found one-way causality from EC to EG in Poland. Chang, et al. [23] found one-way causation from EC to output in Taiwan Province of China by using the VECM model.Conservation Hypothesis: The proponents of the conservation hypothesis argue in the favor of unidirectional causation from EG to EC [24]. If EG causes EC, it shows that EG of the country does not depend on energy, so energy conservation policies will not negatively affect EG. From the empirical perspective, Ghosh [25] used the annual data of India from 1950 to 1997 and found a causal connection between EG and EC (electricity), but the reverse relationship does not exist. Ghali and El-Sakka [26] used data from Canada and established one-way causation between output growth and EC. Ang [27] used the data of Malaysia from 1971 to 1999 and confirmed one-way causation from EG to EC.Neutral Hypothesis: This hypothesis asserts that there is negligible or no effect of EC on EG [28]. If EC and EG do not cause each other, it infers that energy conservation or energy conservation policies will not affect EG, and the acceleration or deceleration of EG will not have a relevant effect on EC. From the empirical point of view, Ferguson, et al. [29] could not find causation between EC and EG by using the data of seven countries. Altinay and Karagol [30] used Hsiao’s causality test and data of Turkey from 1950 to 2000 and found no causality between EC and EG. Fatai, et al. [31] used Toda and Yamamoto tests to examine the data of New Zealand from 1960 to 1999, and findings of their studies supported the independent relationship between EC and EG.Feedback Hypothesis: This hypothesis postulates that EC and EG have bi-directional causality with each other [20]. If EC and EG bi-directionally cause each other, it infers that EC and EG are mutually affected, and any change in one aspect will cause corresponding changes in the other. From the empirical point of view, Glasure and Lee [32] found two-way causality between EC and EG through multiple VAR models. Yang [33] examined the association between EG and EC and found that there was a two-way causality between EC and EG by using data of India. Erdal, Erdal and Esengün [9] found two-way interactions between EC and EG by using data of Turkey from 1970 to 2006.Theoretically, the association between EC and EG has been explored in the literature based on different theories. For instance, Xiang and Diqing [34]] examined the intrinsic relationship between natural resources, environmental pollution, and EG based on the endogenous growth theory. Xiaobo [35] studied different energy factors based on the theory of Copeland and Taylor [36]. Xepapadeas [37] explored the association between resources, the environment, and EG based on the Solow model, and pointed out that sustainable growth and environmental protection can be achieved simultaneously under certain conditions. Zuo and Ai [38] investigated the association among EC, environment, human capital, technological innovation, and EG based on dynamic optimization theory.Empirically, researchers have used different empirical analysis techniques to examine the association between EC and EG. For instance, Lazzaretto and Toffolo [39] and Tashimo and Matsui [40], studied the 3E system composed of environment, EC and EG by using data envelopment analysis methods. Hawdon and Pearson [41], and Oliveira and Antunes [42] analyzed the interaction between environmental pollution, EC, and EG by using the input-output method. Zhao [43] studied the association between EC, EG, and environmental pollution by using system coordination and fuzzy mathematics. Cui and Wang [44] and Xia and Xu [45] examined the association between EC and EG by applying different measurement models such as VAR, co-integration, and VECM.By European standards, Albania is a relatively poor and economically backward country. Albania is steadily transitioning to a more modern open-market economy. In 2018, Albania had a real growth rate of GDP about 3.5%, and the per capita GNP was US $13,274. The sector-wise GDP was distributed among the agriculture (21.6%), industry (14.9%) and service (63.5%) sectors. The main export industries include textiles, footwear, asphalt, metal, non-metal minerals, crude oil, vegetables, fruits, and tobacco, etc. At present, agriculture accounts for about one-fifth of the GDP, while service industries such as tourism account for more than half of the GDP. It is important to recognize that in 2003 and 2004, the domestic economy of Albania grew strongly, while the country had a lot of oil and gas resources, and there was no inflation problem in the country.Bulgaria is an agricultural country, with roses, yogurt, and wine enjoying a great reputation in the international market. At present, food processing and textile industries are the main industries, while the tourism industry has developed in the past few years. Bulgaria is a member of the China-EU free trade agreement. The strongest sectors of the economy comprised of energy, mining, metallurgy, machinery manufacturing, agriculture, and tourism. The main industrial exporting products include clothing, steel, machinery, and refined fuel. The economy of Bulgaria has grown rapidly in recent years, and the per capita GDP of Bulgaria was US $20,116 in 2016 [46].The economy of Croatia is dominated by the tertiary industry. Tourism, construction, shipbuilding, pharmaceutical, and other industries are highly developed in this country. This country has a lot of forest and water resources, with a national forest area of 2,232,000 hectares. Besides, Croatia has oil, natural gas, aluminum, and other resources. The main industries include chemical, plastic, mechanical parts, metal, electronic parts, crude steel, aluminum, paper, wood, building materials, textiles, shipbuilding, oil, tourism, food and beverage, and the main export industries comprise of vehicles, machinery, textiles, chemicals, food, and fuel.The Czech Republic was listed as a developed country by the World Bank in 2006. By the end of 2019, the GDP of the Czech Republic was reported more than US $22,000. Machinery manufacturing, chemical industry, metallurgy, and other industrial sectors are highly developed in the Czech Republic. Foreign trade, tourism, and government financing are the main economic pillars of the country, and these are the main driving forces for the stable growth of the domestic economy. The important fact is that the Czech Republic is no longer a developing country but is steadily included in the list of the 30 most developed countries. The Czech Republic has plentiful resources of lignite, hard coal, and uranium, while it has also mineral resources such as manganese, aluminum, zinc, fluorite, graphite, and kaolin.The Egyptian economy is among the highly diversified economies of the Middle East. Various important industries contribute to the economy almost equally. Egypt is also considered to be an influencing power of Islamic faith in the Mediterranean and the Middle East areas. The economy of Egypt is mainly dependent on agriculture, oil exports, tourism, and labor exports. Oil is a very important part of the Egyptian minerals. The origin of the oil-producing area is on the west coast of the Red Sea, but the production of this area has been gradually reduced. In 2018, the gross domestic product (PPP) of Egypt was the US $1105.039 billion, with an average per capita of US $13,759 [47].Before 2006, Estonia had a strong economy with an annual growth rate of 10%. Estonia has been pursuing a free economic policy, vigorously implementing privatization, and free trade policy. Estonia is ranked at 1st in the EU Member States with rapid economic development and its annual economic growth rate. Estonia is almost energy independent country, with more than 90% of the electricity demand provided by locally mined oil shale. The main mineral resources consist of oil shale, peat, phosphate rock, limestone, etc. Estonia imports oil products from Western Europe and Russia.Hungary has a high-income mixed economy of the OECD, with an output value of US $265.307 billion, measured by purchasing power parity. Hungary has an export-oriented economy and focused on international trade, therefore, Hungary is the 36th largest export-oriented country in the world. Hungary has a private economy of more than 80% with an overall tax rate of 39.1%. The main industries include pharmaceutical, motor vehicles, chemical, metallurgy, electrical appliances, and tourism.In 2016, the GDP in purchasing power parity (PPP) of Lithuania was estimated at US $85.435 billion, with a per capita value of US $29,716 [48]. Agriculture is dominated by high-level animal husbandry, which accounts for more than 90% of the output value of agricultural products. The major crops produced by Lithuania include flax, potato, beet, and various vegetables. Lithuania is rich in amber, with a small amount of clay, sandstone, lime, gypsum, peat, iron ore, apatite, and oil, and it imports oil and natural gas from other countries. A small amount of oil and gas resources have been found in the western coastal areas of Lithuania, but the reserves have not yet been proven. The major industries of Lithuania comprise of mining and quarrying industry, processing and manufacturing industry and energy industry. Some industries are relatively developed, mainly including food, wood processing, textile, chemical industry, etc., while, machinery manufacturing, chemical, petrochemical, electronic, and metal processing industries are developing rapidly.In 2018, the GDP (PPP) of Morocco was the US $315.441 billion with US $8959 per capita GDP [47]. Major economic sectors of Morocco consist of tourism, fisheries, and phosphate minerals. Morocco has plenty of phosphate reserves about 110 billion tons and ranked 1st in the world. The agriculture and animal husbandry industries are greatly affected by climate change. The economy of Morocco relies on external financing in many ways, while France and Spain are the largest donors to help the economy of Morocco. The mining industry of Morocco has a good development momentum, mainly due to the increasing demand for phosphate in the international market. Chemical, automobile, aviation, electronics, and other industries have become the major helping hand for the development of the manufacturing industry. However, due to the poor performance of the textile, and leather manufacturing industries, the overall growth of the manufacturing industry was slow.According to the information released by the Central Statistical Office of Poland, the per capita GDP of Poland was US $13,414 with an annual growth rate of 4.6% in 2017. In 2019, the real economic growth rate of Poland was 4.1%, and the nominal GDP was US $589.847 billion. Poland is rich in mineral resources such as coal, shale gas, sulfur, copper, zinc, lead, aluminum, and silver. Poland is the largest producer of hard coal in central Europe, and its output can meet 10% of the EU’s total demand. Poland is ranked 9th in Europe in terms of lignite extraction, but only 15% of its reserves are developed. Poland has also specific hydrocarbon fuels. By the end of 2017, 9.513 million hectares of forest (green space) was covered, with a forest coverage rate of 30.4 percent. The main industrial products of Poland include coal, steel, cars, cement, and so on.The economy of Romania is considered among the top economies of Central and Eastern Europe. The economy of Romania was rapidly growing, and its overall performance was excellent in 2015. Romania made a lot of gratifying progress, not only in terms of macroeconomic indicators but also in terms of microeconomic level. In 2014, many policies were transformed into specific actions in favor of entrepreneurs and businessmen, especially in terms of employment promotion. In 2017, the economic growth of Romania was reached 7%, even it exceeded the Chinese economic growth (6.9%) [47].The Slovak Republic was an agricultural country, and there was no basic industry in the country in the early years. The Czechoslovak Communist Party gradually established steel, food processing, and military industries in Slovakia during its administration, and narrowed the economic gap with the Czech Republic. The economy of the Slovak Republic was declined in 2009 due to the international financial crisis, but it recovered growth in 2010 and beyond. The automobile industry is the industrial pillar of the Slovak national economy. The Slovak Republic is not a rich country in terms of oil and gas resources, and it mostly has small oil fields, which are scattered in the Carpathian Mountains and the eastern region.Since the government of Ben Ali was overthrown in 2011, the Tunisian economy was severely affected. In 2014, the economic growth of Tunisia was only 1%. By 2015, the unemployment rate in Tunisia had risen to nearly 30%, which was twice the unemployment rate during the administration of Ben Ali. Until 2016, the Tunisian economy was recovered strongly. In 2016, the total GDP of Tunisia was the US $130.77 billion, with a per capita GDP of US $11,651 [48]. Tunisia is rich in olive oil, it is known as the “olive oil garden of the world” and “the country of olives”. Renewable energy plays a secondary role in the energy supply, while solar energy is widely used in Tunisia. Photovoltaic power generation, wind power generation, etc. bring great impetus to the development of the Tunisian national economy.The economic status of Turkey is so important in the world, and it is a founding member of the Organization for Economic Cooperation and Development (OECD). According to the World Bank, the per capita GDP of Turkey was reached $766.5 billion in 2018, ranked at 19th in the world. At the same time, Turkey is rich in natural and mineral resources. Turkey has more than 60 kinds of mineral resources. According to the statistics of mineral diversity, Turkey is ranked 10th in the world in terms of minerals. Turkey has plentiful reserves of boron salt, account for 72% of the world. In addition to boron salt reserves, Turkey is also rich in coal, iron, copper, and chromium.We selected the sample of 14 developed and developing member states of the Union for the Mediterranean (UFM) (a list of all countries which are members of the Union for Mediterranean (UFM) countries is available here [49]), whose GDP per capita does not exceed $25,000 during the time duration from 1995 to 2014, by following the sampling method adopted by Comes, et al. [50]. The countries include Albania, Bulgaria, Croatia, Czech Republic, Egypt, Estonia, Hungary, Lithuania, Morocco, Poland, Romania, Slovak Republic, Tunisia, and Turkey. We used these countries based on Hierarchical Cluster Analysis (HCA), shown in Figure 1. The figure is drawn by using Ward-hierarchical grouping method [51] and hclust-clustering package [52] from R (R Core Team, Vienna, Austria) [53]. We collected annual data on EG, EC, sustainable development, FDI, international trade, labor force, capital stock, financial development, population and inflation from the World Bank database and International Monetary Fund (IMF), International Financial Statistics (IFS) for the period 1995 to 2014.To find the association among EC, EG and sustainable development (SD) in the UFM countries, in line with prior studies [2,54,55], this study considers the following equations for estimations:(1)LnEGi,t=αi+β1LnECi,t+β2LnSDi,t+β3LnCFi,t+β4LnLFi,t+β5LnFDIi,t+β6LnINFi,t+μi,t
+        (2)LnECi,t=α0+β1LnEGi,t+β2LnSDi,t+β3LnCFi,t+β4LnLFi,t+β5LnPOPi,t+β6LnFDi,t+λi,t
+        (3)LnSDi,t=ε0+β1LnECi,t+β2LnEGi,t+β3LnCFi,t+β4LnLFi,t+β5LnFDIi,t+β6LnTRi,t+νi,tFrom Equations (1) to (3), αi, α0 and ε0 are the intercept terms; LnEGi,t is the logarithm of economic growth; LnECi,t is the logarithm of energy consumption; LnSDi,t is the logarithm of sustainable development; LnCFi,t is the capital formation; LnLFi,t is the logarithm of labor force; LnFDIi,t is the logarithm of foreign direct investment; LnPOPi,t is the logarithm of total population of the country; LnFDi,t is the logarithm of financial development; LnTRi,t is the logarithm of international trade; LnINFi,t is the logarithm of inflation; μi,t, λi,t, and νi,t are error terms. Subscript i and t denote countries and periods respectively. The three way association among economic variables is broadly analyzed by Equations (1)–(3). The Equation (1) posits that sustainable development (SD), capital formation (CF), labor force (LF), FDI, and international trade (TR) are the potential determining factors of EG as suggested by Esseghir and Khouni [2] and Boţa-Avram, et al. [56]. Equation (2) postulates that economic growth (EG), sustainable development (SD), CF, LF, POP and financial development (FD) are the potential determining factors of energy consumption (EC) as recommended by [Anwar and Sun [57], and Lee [58]]. Equation (3) posits that EC, EG, capital formation (CF), labor force, FDI and international trade (TR) are the potential determining factors of sustainable development (SD) as recommended by [Boţa-Avram, Groşanu, Răchişan and Gavriletea [56]]. We assume that all variables are stationary, the Equation (1) entails the panel co-integration with both panel-VECM and panel-VAR as follows:For VECM:(4)∆lnEGit=φ1ecmi,t−1+δ11∆lnECit+δ21∆lnSDit+δ31∆lnCFit+δ41∆lnLFit+δ51∆lnFDIit+δ61∆lnINFit+∈1it
+        (5)∆lnECit=φ2ecmi,t−1+δ12∆lnEGit+δ22∆lnSDit+δ32∆lnCFit+δ42∆lnLFit+δ52∆lnFDIit+δ62∆lnINFit+∈2it
+        (6)∆lnSDit=φ3ecmi,t−1+δ13∆lnEGit+δ23∆lnECit+δ33∆lnCFit+δ43∆lnLFit+δ53∆lnFDIit+δ63∆lnINFit+∈3it
+        (7)∆lnCFit=φ4ecmi,t−1+δ14∆lnEGit+δ24∆lnECit+δ34∆lnSDit+δ44∆lnLFit+δ54∆lnFDIit+δ64∆lnINFit+∈4it
+        (8)∆lnLFit=φ5ecmi,t−1+δ15∆lnEGit+δ25∆lnECit+δ35∆lnSDit+δ45∆lnCFit+δ55∆lnFDIit+δ65∆lnINFit+∈5it
+        (9)∆lnFDIit=φ6ecmi,t−1+δ16∆lnEGit+δ26∆lnECit+δ36∆lnSDit+δ46∆lnCFit+δ56∆lnLFit+δ66∆lnINFit+∈6it
+        (10)∆lnINFit=φ7ecmi,t−1+δ17∆lnEGit+δ27∆lnECit+δ37∆lnSDit+δ47∆lnCFit+δ57∆lnLFit+δ67∆lnFDIit+∈7itFor the VAR Model:(11)lnEGit=η0+η1lnEGi,t−k+η2lnECi,t−k+η3lnSDi,t−k+η4lnCFi,t−k+η5lnLFi,t−k+η6lnFDIi,t−k+η7lnINFi,t−k+ψ1it
+        (12)lnECit=θ0+θ1lnECi,t−k+θ2lnEGi,t−k+θ3lnSDi,t−k+θ4lnCFi,t−k+θ5lnLFi,t−k+θ6lnFDIi,t−k+θ7lnINFi,t−k+ψ2it
+        (13)lnSDit=κ0+κ1lnSDi,t−k+κ2lnEGi,t−k+κ3lnECi,t−k+κ4lnCFi,t−k+κ5lnLFi,t−k+κ6lnFDIi,t−k+κ7lnINFi,t−k+ψ3it
+        (14)lnCFit=υ0+υ1lnCFi,t−k+υ2lnEGi,t−k+υ3lnECi,t−k+υ4lnSDi,t−k+υ5lnLFi,t−k+υ6lnFDIi,t−k+υ7lnINFi,t−k+ψ4it
+        (15)lnLFit=ρ0+ρ1lnLFi,t−k+ρ2lnEGit−k+ρ3lnECi,t−k+ρ4lnSDi,t−k+ρ5lnCFi,t−k+ρ6lnFDIi,t−k+ρ7lnINFi,t−k+ψ5it
+        (16)lnFDIit=ω0+ω1lnFDIi,t−k+ω2lnEGi,t−k+ω3lnECi,t−k+ω4lnSDi,t−k+ω5lnCFi,t−k+ω6lnLFi,t−k+ω7lnINFi,t−k+ψ6it
+        (17)lnINFit=ϑ0+ϑ1lnINFi,t−k+ϑ2lnEGi,t−k+ϑ3lnECi,t−k+ϑ4lnSDi,t−k+ϑ5lnCFi,t−k+ϑ6lnLFi,t−k+ϑ7lnFDIi,t−k+ψ7itFrom the Equations (4) to (10), ecmi,t−1 are the error correction term; φ1, φ2, φ3, φ4, φ5, and φ6 are for capturing the long-run connection among variables. ∆s represents the difference operators. φ1<0, φ2<0, φ3<0, φ4<0, φ5<0, and φ6<0 assume that long run connection does not obstruct fluctuations in EG, EC, sustainable development (SD), capital formation (CF), labor force, FDI, and inflation, whereas the greater sign exhibits opposite meaning of it. ∈1t, ∈2t, ∈3t, and ∈4t denote the error terms. From the Equations (11) to (17), η0, θ0, κ0, υ0, ρ0, ω0, ϑ0 are the intercept terms; k is the number of lags; ψ1it, ψ2it, ψ3it, ψ4it, ψ5it, and ψ6it denote the error terms.Generally, the problem of endogeneity occurs in the panel dataset as a result of the mutual correlation between endogenous variables and stochastic error terms. Consequently, we used a Durbin–Wu–Hausman test to check the endogeneity bias in our dataset. To deal with endogeneity bias, we employed system-GMM instrumental variables estimation technique. Moreover, we tested the stationary of each variable by employing the unit root test. After finding co-integration in the models by using a panel co-integration test, we employed VECM and VAR models to find the causation among EC, EG, and sustainable development among other economic factors. To find more robust results, we employed impulse response analysis (IRA) and variance decomposition analysis (VDA). All variables are described with measures and data sources in Table 1.Table 2 describes the summary statistics of EG, EC, sustainable development (SD), and other control variables used in this study.Table 3 exhibits the results of panel unit root tests such as Levin-Lin (LL), Im-Pearson-Shin (IPS), Fisher Augmented-Dickey Fuller (ADF) test, and Fisher Philips-Perron (PP). The results of all unit root tests indicate that all variables including EG, EC, sustainable development (SD), and others are found to be non-stationary at level and converted into stationery with the help of first difference of these variable. These results encourage us to employ a panel co-integration test to find the association between these variables.Table 4 displays the results of the system GMM to investigate the dynamic association between EG, EC, and sustainable development (SD) among other factors. The results of Model 1 indicate that EC has a significant positive association with EG at 1% level. The economic significance implies that a rise in EC by 1% can increase the EG by 0.3111%. Moreover, SD has shown a significant negative association with EG at 1% level. The estimated coefficient of this result specifies that a 1% rise in the SD can decrease the EG by 0.0822%. From the controlling variables, CF has shown a significant positive association with EG, while LF has a significant negative association with EG.The results of Model 2 demonstrate that EG has a significant positive effect on EC at 1% level. The estimated coefficient specifies that a 1% rise in EG can increase EC by 0.9640%. From the controlling variables, CF and POP have demonstrated a significant negative association with EC, while LF and FD have shown a significant positive association with EC. The results of Model 3 postulate that EG has a significant negative association with SD at 1% level. This result demonstrates that 1% rise in EG decreases the SD by 0.9633%. From the controlling variables, CF has shown a significant positive association with SD, while LF has shown a significant negative association with SD.Before using VECM, we employed a panel cointegration test with different methods including Pedroni and Kao to find the cointegration between EG, EC, sustainable development (SD), and other factors. The results of the panel cointegration test suggest that all these variables are cointegrated with each other in all models (1–3), results are shown in Table 5. This finding encourages us to employ VECM to find the causality among variables.Table 6 demonstrates the results of VECM used to inspect the direction of causality between variables used in the model (1) of EC, EG, and sustainable development (SD). The results of Model (4), the coefficient of error correction term (ECT) is exposed to be insignificant, while the labor force induces short-run dynamic association with EG at 5% level. It suggests that the growth in the labor force can stimulate the EG. The results of Model (5), the coefficient of ECT is exposed to be significant at 1% level with a negative sign (i.e., ECT is −0.0238). It suggests that the speed of adjustment (SOA) of energy consumption (EC) is 2.38% towards long-run equilibrium. Additionally, the short-run dynamic association is revealed from sustainable development (SD) to EC at 1% level.The results of Model (6), the coefficient of ECT is exposed to be significant at 1% level with a negative sign (i.e., ECT is −0.1489). It postulates that SOA of sustainable development (SD) is 14.89% towards the long-run equilibrium. Besides, the dynamic association is revealed in the short run from economic growth (EG) and capital formation (CF) to sustainable development (SD) at 5% and 10% levels respectively. The results of Model (7), the coefficient of ECT is exposed to be insignificant, while the energy consumption (EC) induces the short-run dynamic association with capital formation (CF) at 1% level.The results of Model (8), the coefficient of ECT is exposed to be significant at 10% level with a negative sign (i.e., ECT is −0.0104). It suggests that the SOA of the labor force (LF) is 1.04% towards long-run equilibrium. Additionally, the short-run dynamic association is revealed from EG, EC, sustainable development (SD), and inflation (INF) to the labor force (LF) at 5%, 1%, 10%, and 5% levels respectively. The results of Model (9), the coefficient of ECT is exposed to be significant at 5% level with a positive sign (i.e., ECT is 0.3549). It postulates that SOA of FDI is 35.49% towards long-run equilibrium. Moreover, EC induces the short-run dynamic association with FDI at 1% level. The results of Model (10), the coefficient of ECT is exposed to be significant at 10% level with a negative sign (i.e., ECT is 0.8983). It suggests that SOA of inflation (INF) is 89.83% towards long-run equilibrium.Table 7 shows the results of Models (11) to (17) estimated by employing a panel VAR model. The results of Model (11) indicate that economic growth (EG) lagged by 1 period significantly positively influences EG at 1% level. It suggests that an increase in the EG lagged by 1 period can stimulate the EG by 92.18%. Moreover, energy consumption (EC) lagged by 1 period significantly positively influences EG at 5% level. It implies that an additional input in the EC lagged by 1 period can increase the EG by 5.27%.The results of Model (12) specify that EC lagged by 1 period significantly positively influences the EC at 1% level. It postulates that an increase in the EC lagged 1 period can increase the EC by 98.59%. The results of Model (13) show that EC lagged by 1 period significantly negatively influences sustainable development (SD) at 1% level. It suggests that an additional input in the EC lagged by 1 period can decrease the SD by 17.71%. Furthermore, SD lagged by 1 period significantly positively influences the SD at 1% level. It infers that a rise in the SD lagged by 1 period can increase the SD by 65.71%. Additionally, inflation (INF) lagged by 1 period significantly negatively influences the SD at 10% level. It postulates that the growth in inflation (INF) lagged by 1 period can reduce the SD by 4.51%.The results of Model (14) imply that SD lagged by 1 period significantly positively influences capital formation (CF) at 10% level. It implies that an increase in the SD lagged by 1 period can stimulate the CF by 3.55%. Moreover, CF lagged by 1 period significantly positively influences the CF at 1% level. It suggests that an additional input in CF lagged by 1 period can increase capital formation by 86.18%. Furthermore, the labor force (LF) lagged by 1 period significantly positively influences CF at 5% level. It implies that growth in LF lagged by 1 period can stimulate CF by 14.35%. Additionally, FDI lagged by 1 period significantly positively influences CF at 5% level. It suggests that a rise in FDI lagged by 1 period can intensify CF by 3.33%.The results of Model (15) indicate that EG lagged by 1 period significantly negatively influences the LF at 5% level. It suggests that growth in EG lagged by 1 period can reduce the LF by 1.48%. Furthermore, EC lagged by 1 period significantly negatively influences the LF at 5% level. It implies that an additional input in the EC can reduce the LF by 0.75%. Moreover, CF lagged by 1 period significantly positively influences the LF at 1% level. It postulates that growth in CF lagged by 1 period can increase the LF by 1.65%. Additionally, LF lagged by 1 period significantly positively influences the LF at 1% level. It suggests that growth in LF lagged by 1 period can increase the labor force by 98.83%. Moreover, FDI lagged by 1 period significantly influences the LF at 10% level. It implies that a rise in FDI lagged by 1 period can increase the LF by 0.22%.The results of Model (16) specify that EC lagged by 1 period significantly positively influences FDI at 1% level. It postulates that growth in EC lagged by 1 period can raise the level of FDI by 51.44%. Moreover, FDI lagged by 1 period significantly positively influences FDI at 1% level. It implies that a rise in FDI lagged by 1 period can increase the level of FDI by 60.92%. Furthermore, inflation (INF) lagged by 1 period significantly negatively influences the FDI at 5% level. It suggests that an increase in the INF lagged by 1 period can decrease the level of FDI by 11.93%.The results of Model (17) indicate that EG lagged by 1 period significantly negatively influences inflation (INF) at 5% level. It suggests that a rise in the EG lagged by 1 period can reduce the INF by 40.53%. Moreover, sustainable development (SD) lagged by 1 period significantly negatively influences INF at 5% level. It postulates that an additional input in the SD lagged by 1 period can reduce the INF by 14.32%. Furthermore, INF lagged by 1 period significantly positively influences INF at 1% level. It implies that an increase in the INF lagged by 1 period can escalate the INF by 68.13%.Table 8 demonstrates the results estimated by VDA for the Model (1) related to economic growth (EG). Results show that EG has a short-run self-explanatory effect, however, the long-run effect is decreased to 89.15%. Energy consumption (EC), sustainable development (SD), CF, LF, FDI, and INF did not highlight shocks in the short run. Besides, 1.3% of EC is expounded by shocks to EG in the long run. Furthermore, EG is influenced by 1.29% shocks of SD in the long run. Likewise, 0.23% of CF is explicated by shocks to EG in the long run. Additionally, 0.10% of LF is expounded by shocks to EG in the long run. Besides, EG is influenced by 2.40% shocks of FDI in the long run. Moreover, 5.53% of inflation (INF) is explicated by shocks to EG in the long run.Figure 2 depicts the estimated shocks of variables by IRA which highlights the influence of one variable on the one unit of impulse in other variables. IRA is employed by keeping in view the order of variables with ten years period. Figure 2 shows that shocks in energy consumption (EC), sustainable development (SD), capital formation (CF), the labor force (LF) and FDI demonstrated positive connection with the EG over the period, whereas the shocks in the inflation (INF) exhibited the negative connection with EG over the period.Figure 2 illustrates that the shocks in EG, SD, LF, and FDI highlighted positive connection with EC over the period, whereas the shocks in INF and CF demonstrated a negative connection with EC over the period. Moreover, the shocks in the EG, LF, and FDI showed a positive connection with SD over the period, whereas the shocks in CF and INF exposed negative connection with SD over the period. While, the shocks in EC demonstrated a positive effect on SD in four years at the starting period, while these shocks showed a negative effect on SD in the rest of the period.Figure 2 shows the shocks in all concerned variables except inflation (INF) exhibited a positive influence on CF over the period. Furthermore, shocks in EG, SD, CF, and FDI exposed positive effects on LF over the period, whereas shocks in EC and INF demonstrated a negative influence on LF over the period. Moreover, shocks in EG, EC, CF, and LF showed a positive association with FDI, whereas shocks in SD demonstrated a negative association with FDI at the starting years (1–5), but later these shocks showed a positive association with FDI for the rest of period. While the shocks in INF demonstrated a negative connection with FDI over the period. Furthermore, shocks in EC and CF demonstrated a positive influence on the INF over the period, whereas the shocks in other variables highlighted a negative influence on INF over the period.This paper examines the association between EC, EG, and sustainable development (SD) among other economic factors for the UFM countries by using the system-GMM model, VECM, VAR model, VDA, and IRA for the period of 1995 to 2014. The economic factors include capital formation (CF), FDI, the labor force (LF), inflation (INF), population (POP), international trade (TR), and financial development (FD).The estimated results by GMM confirmed the bidirectional causality between EG and EC, no causality between EC and SD, and bidirectional causality between EG and SD. The empirical results of the co-integration test confirmed the co-integration between variables. The VECM results confirmed the long-run equilibrium association in the equations of energy consumption (EC), sustainable development (SD), the labor force (LF), FDI, and inflation (INF). Moreover, the results validated the short-run dynamic association from sustainable development (SD) to EC, EG to sustainable development (SD), EC to CF, energy consumption (EC) to the labor force (LF), sustainable development (SD) to the labor force (LF), and energy consumption (EC) to FDI. Moreover, results also validated the bidirectional short-run causality between EG and the labor force (LF).Furthermore, the results of VAR model validated the short-run causality from EC to EG, inflation (INF) to EC, EC and inflation (INF) to sustainable development (SD), sustainable development (SD), the labor force (LF) and FDI to the capital formation (CF), EC, capital formation (CF) and FDI to the labor force (LF), energy consumption (EC) and inflation (INF) to FDI, and sustainable development (SD) to inflation (INF).Based on these results, we can conclude that there is a strong association between EC and EG, and economic growth (EG) is also strongly connected with sustainable development (SD) of the UFM countries. It implies that the higher level of EC can stimulate the growth of the economy, which could help to attain sustainable development (SD) in this region. The key policy implication of this conclusion is that the energy policies should give full attention not only a causal association between EC and EG but also whether it is temporal or permanent. Consequently, policymakers should formulate policy actions. Policymakers should also consider the possible effects of energy consumption on health, society, and the environment. They should take into consideration the current status of economic and energy sustainability. Consequently, they should highlight deficiencies, and formulate ways to improve the situation. Thus, policymakers need to know about the significances of energy, environmental and economic plans, and their possible effects on the shaping of development and the viability of converting this development into sustainable development (SD).Conceptualization, Data Curation, Writing—Original Draft Preparation, R.L.; Project Administration, Funding Acquisition, Supervision, Y.K.; Formal Analysis, Review and Editing, Y.P., and Validation, and Review and Editing, S.A.J. All authors have read and agreed to the published version of the manuscript.This study was funded by the National Natural Science Foundation of China (no. 71973054).The authors would like to acknowledge the comments and suggestions given by anonymous reviewers that have significantly improved the quality of our work. We further acknowledge the overall support and guidance of Usman Sattar, from the College of Law and Political Science, Zhejiang Normal University, China to complete this study.The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.Hierarchical Cluster Analysis (HCA) dendrogram analysis.Impulse Response Analysis (IRA).Description of Variables.1 World Development Indicators (WDI) Database; International Monetary Fund (IMF); International Financial Statistics (IFS).Descriptive Statistics.Panel Unit Root Tests.Significance Level: * p < 0.1, ** p < 0.05, *** p < 0.01. LL is Levin-Lin; IPS is Im-Pearson-Shin; ADF is Fisher Augmented-Dickey Fuller; PP is Fisher Philips-Perron.Results of System-Generalized Method of Moment (GMM).1 L1 is the lag of one period for the said variables; AR(1) is the first order autoregressive process; AR(2) is the second order autoregressive process; Significance Level * p < 0.1, ** p < 0.05, *** p < 0.01.Panel Co-integration Test.Significance Level ** p < 0.05, *** p < 0.01.Panel Vector Error Correction Model (VECM) for the Model (1).Significance Level. * p < 0.1, ** p < 0.05, *** p < 0.01.Panel Vector Autoregression (PVAR) for the Model (1).Significance Level: * p < 0.1, ** p < 0.05, *** p < 0.01.Variance Decomposition Analysis for the Model 1 (LNEG).

Med-MDPI/ijerph_5/ijerph-17-15-05615.txt

@@ -0,0 +1 @@
+Mental illness is not uncommon among young adults, but negative attitudes towards mental disorders and lack of parental support might be associated with hesitancy in seeking professional help. This study aimed to examine the relationships of parental support, beliefs about mental illness, and mental help-seeking among young adults in Saudi Arabia. This quantitative cross-sectional study included a convenience sample of 236 young adults (ages 18–25) with the majority of the total (86.4%) being female. Data were collected via three self-administered questionnaires: The Perceived Parental Support Scale, Beliefs toward Mental Illness scale, and Mental Help Seeking Attitude Scale. Results indicated that the participants had a moderately negative attitude toward mental illness, a moderately positive attitude toward parental support, and a highly positive attitude toward mental help-seeking. No significant relationships were found among the study variables. The study highlights that despite young adults’ positive attitude toward mental help-seeking and parental support, they have negative views toward people with mental illnesses. Educational programs in schools and media are needed to improve attitudes and enhance readiness to interact with people with mental illness.Mental illness is a neglected issue in Saudi Arabia, as evidenced by the lack of reported data on the prevalence of mental illness in the general population. Further, negative attitudes toward mental illness among the Saudi population, as well as hesitancy toward seeking mental help, have been reported in the literature [1]. For example, reports indicate that treatments for mental illnesses might be seen as useless, time-wasting, and expensive [1].People diagnosed with mental illness are often rejected by the community and considered a burden to their families; thus, they, and their relatives, often hide the disorder from others because they fear public disapproval. Various studies have found that most people isolate themselves from individuals with mental illness, considering them dangerous and incapable of friendship [1,2]. Negative attitudes toward mental illness promote public discrimination against and stigmatization of patients with mental disorders and may cause them to withdraw from social interactions, which is likely to aggravate their mental illness condition [3].Parental support is essential to a child’s recovery from mental illness. Research indicates that children diagnosed with mental disorders thrive in an environment where they are accepted and encouraged [4]. Such situations allow youths to build up their self-confidence, social skills, and emotional control abilities [4]. Parents are commonly involved in the mental help-seeking progression of youths suffering from mental health disorders [5]. In addition, parents have a strong influence on encouraging their children to seek help and finding the right path for managing their children’s psychological problems, and parents who have stable marriages and shared intimacy tend to have better attitudes toward mental health disorders [6].There is a growing concern about the burden and effect of mental disorders on people’s health, their social lives, and economies worldwide [7]. It was found that among 160 patients who attended primary healthcare centers in the capital of Saudi Arabia during three months in 2017, 28.5% reported cases of mental health disorders [8]. Further, the Saudi National Health & Stress Survey reported that 2 in 5 Saudi youth meet the criteria for having a mental health condition sometime in their life and only 5% of Saudis seek professional treatment for their mental illness [9]. It is important to emphasize that supportive attitudes toward mental disorders are vital to minimizing the stigma around mental illnesses and encouraging people to seek help from the appropriate sources [10]. Nevertheless, a study of 650 Saudi adults (age 18 years and older) revealed that most of the participants (87.5%) indicated a lack of knowledge about mental illness. Moreover, 66.5% of the participants reported negative attitudes toward people with mental illness in relation to successful treatment, maintaining a job, and getting married, as well as toward their obtaining professional help [1]. Thus, it is important to draw attention to attitudes toward mental illness and mental help-seeking among young adults in Saudi Arabia. It is also critical to explore attitudes towards parental support. Young adults aged 18 to 25 usually live with their parents in the same house in the culture of Saudi Arabia, which allows parents to be involved in a variety of aspects of their children’s lives, including their psychological problems. Parents’ behaviors and attitudes have a powerful impact on the mental, psychological, and emotional well-being of their children [11]. Parents should be actively involved in their children’s lives by offering support and encouragement when required. Such an approach will help children to manage and defeat mental disorders [4]. However, in 2019, the general divorce rate in Saudi Arabia reached 3.20 divorces per 1000 in the Saudi married population [12]. It was discovered that separation among parents is significantly related to depression, anxiety, and stress among young Saudi females [13].Negative attitudes toward mental disorders appear to have an adverse impact on patients with mental disorders. Abolfotouh et al. found that 66.5% of people reported negative attitudes toward people with mental illness, as discussed above [1]. Another study conducted with 557 undergraduate college students showed that half of the students (52.5%) held negative attitudes towards persons with mental illness, believing them incapable of friendships [14]. A further study reported that one-third of 3464 Saudi participants had a hesitant attitude toward mental illness and persons with mental illness [15]. However, another recent study discovered that 90% of Saudi adults sampled believed that patients with mental illness deserve respect [16]. Nevertheless, 35.8% of people in the same study believed that patients with mental illness leaned toward dangerous and aggressive behavior, and 33.2% thought that society would be more secure if patients with mental illness were hospitalized [16]. Researchers indicate that religion, culture, and urbanization influence parents’ beliefs on psychological disorders. For example, James et al. argue that conservative cultural and religious tendencies in the nation contributed to those beliefs [7]. Studies also indicate that, compared to people living in urban centers, those in rural areas are more tolerant of people with mental disorders [17]. Negative attitudes could also be due to a lack of knowledge about mental disorders. In contrast, individuals who possess high levels of education are more knowledgeable about mental illnesses [3]. While there are studies exploring attitudes about mental illness, there are scarce data on attitudes toward mental help-seeking among the Saudi population. A study that aimed to explore the attitudes towards mental health help-seeking in 650 Saudi adults aged > 18 years showed that 54.5% of participants reported negative attitudes to help-seeking behaviors, 40.5% reported neutral attitudes, and only 5% reported positive attitudes [1]. Religious, supernatural, and social beliefs can affect how people perceive causes of mental illness and how to deal with it. It was found that, in Saudi Arabia, people strongly believe in demonic possession and the evil eye, which may lead them to seek unprofessional help or help from religious healers before going to mental health specialists [2]. Therefore, there is a need to investigate the attitude toward seeking help from official psychological and psychiatric specialists among Saudi young adults.Parental support has been defined as “parental behaviors toward the child, such as praising, encouraging and giving physical affection, which indicate to the child that he or she is accepted and loved” [18] (p.176). Parents’ behaviors and attitudes have a powerful impact on the mental, psychological, and emotional wellbeing of an individual [11]. In adulthood, parents continue to be fundamental supporters in helping young adults with mental illness needs [19]. Divorce and separation among parents are problems that may interfere with the psychological wellbeing of their children. A study was conducted in Riyadh on 327 females, aged 12 to 16 years, who disclosed that the prevalence of parental divorce was 13.3% (n = 40) while that of parents living separately but not legally divorced was 10.9% (n = 33). Taken together, the prevalence of marital discord (divorced or separated) was 24.6% (n = 73 out of n = 296). The results revealed that separation among parents was significantly related to depression, anxiety, and stress among young Saudi females [13].The association of mental help-seeking behaviors with parental support has been borne out by the literature [4,11,20,21]; however, this has not been researched among Saudi young adults. One study showed that when young people had positive past experiences and received social support and encouragement from others, such as parents, they had a more positive attitude than others in this age group toward seeking mental help [20]. In addition, better parent–child relationships were associated with greater mental help-seeking intentions [21]. Another study conducted among 1482 students (mean age 17 years) disclosed that positive parenting had an association with greater intentions of seeking mental help from professional sources; 88% of parents indicated they had influenced their young children to obtain mental health services [4]. Wahlin and Deane examined the mental help-seeking process among 256 young people, reporting that 90% of parents of young people stated that parents are more influential than any other source regarding seeking mental help from professional services [11]. Additionally, one study that examined parental support and mental help-seeking attitude in adolescents showed a small but significant positive relationship between parental support and help-seeking intentions among 1482 Australian students aged 16 to 18 years [4]. Another study exploring parents’ attitudes toward their children’s mental disorders and help-seeking behaviors among 400 Iranian parents revealed a significant relationship between seeking help from official sources and fathers’ higher level of education [6].Much of the literature indicates that attitudes toward mental illness and mental help-seeking are correlated concepts, and the connection between them seems logical; when young adults live with supportive parents who have accepting attitudes toward mental illness, their attitude toward mental help-seeking will be positive as well [4,6]. However, no studies were found to support, with empirical evidence, the possible existence of the relationships among the variables for young adults in Saudi Arabia. In order to better understand and improve the mental health of Saudi young adults, it is crucial to assess their attitudes toward parental support, mental illness, and mental help-seeking. At present, there is little research regarding the role of parenting and parental support with mental health illnesses in Saudi Arabia. Additionally, there is a lack of empirical evidence to support the hypothesis that if parenting support and beliefs about mental illness are positive, attitudes toward mental help-seeking will be positive as well. Therefore, it worth exploring the attitudes of, and possible relationships between, parental support, beliefs about mental illness, and mental help-seeking attitudes among young adults in Saudi Arabia.In order to examine the potential relationships between these attitudes, three specific objectives were developed: (1) to assess the attitudes of young adults about parental support, mental illness and mental help-seeking, (2) to examine the relationships between parental support, beliefs toward mental illness, and mental help-seeking attitude, and (3) to observe the difference in the mean scores of these three measures across the characteristics (age groups, gender, occupation, educational status and parents’ status) of study subjects.A cross-sectional design was used among young adults (aged 18–25 years) in Saudi Arabia. Participants were young adults recruited from the general public. Inclusion criteria were (a) young adults aged from 18 to 25 who were (b) able to speak and read English. There were no exclusion criteria for those who met the inclusion criteria.A convenience sample of 236 young adults was enrolled in this study. The relationships between parental support, beliefs toward mental illness, and mental help-seeking attitudes have not been previously studied; hence, a relevant published effect size is not available. Therefore, the estimation for a medium effect size was considered. The estimated medium effect size is 0.30 for correlations with two-tailed tests at 0.05 level of significance and power of 0.80 [22], so the required sample size for correlation was 84 in this study, as calculated by G*Power 3. 1. The questionnaire in this study was comprised of four measures: demographic data, the Perceived Parental Support (PPS) Scale, the Mental Help Seeking Attitude Scale (MHSAS), and the Beliefs Toward Mental Illness Scale (BMI). Demographic data collected included gender, age, country of residence, occupation, educational level, marital status, and parents’ relationship/marriage status.The Perceived Parental Support (PPS) Scale [23], a valid and reliable tool used with multiple populations, demonstrated a Cronbach’s alpha between 0.77 and 0.87. The PPS consists of a five-item scale to measure parental general support [23]. The PPS starts with a question, “How easy or hard is it for you to receive the following from your parents?” followed by five items (a) “Caring and warmth,” (b) “Discussions about personal affairs,” (c) “Advice about studies,” (d) “Advice about other issues (projects) of yours,” and (e) “Assistance with other things.” The responses are rated on a 4-point Likert scale from very difficult (1) to very easy (4). The possible score values range from 5 to 20 with higher mean scores indicating greater levels of perceived parental support.The Mental Help Seeking Attitude Scale (MHSAS) [24] is a 9-item scale designed to measure respondents’ overall evaluation of their attitude toward seeking help from a mental health professional if they were to find themselves dealing with a mental health concern. The resulting mean score can range from a low of 1 to a high of 7. The possible score values range from 9 to 63. A higher score indicates a more positive attitude toward seeking help [24]. The HSAS has demonstrated initial evidence of reliability with a Cronbach’s alpha of 0.93 [24].The Beliefs Toward Mental Illness Scale (BMI) is a 21-item self-report measure of negative stereotypical views of mental illness [25]. The measure holds promising evidence of validity and reliability with a Cronbach’s alpha of 0.82 [25]. The BMI involves three subscales: “dangerousness,” “poor social and interpersonal skills,” and “incurability.” Items are rated on a six-point Likert scale ranging from completely disagree (0) to completely agree (5). The possible score values range from 0 to 105. Higher scores indicate greater levels of negative belief toward mental illness. Primary data collection was done using a questionnaire. The participants were approached using online social media platforms including WhatsApp, Facebook, and Twitter. The survey was distributed using an online tool and was composed of 35 items consisting of the demographics, PPS, MHSAS, and BMI. The questionnaire for data collection remained available online from March 2020 to April 2020. Data entry and analysis were performed using the latest version of the Statistical Package for the Social Sciences (SPSS) 26.0 version (IBM Inc., Chicago, IL, USA). Descriptive statistics were used, including frequencies for all study variables. Means and standard deviations were reported for continuous variables, counts, and percentages for categorical variables. For inferential statistics, Pearson’s correlation coefficient was performed to test the significance and the strength of relationships between continuous variables. Student’s t-test and one-way analysis of variance were also used to compare the mean scores of the three instruments in relation to the categorical variables. The reliability of three instruments was assessed by calculating Cronbach’s alpha. A p-value of ≤0.05 and 95% confidence intervals were used to report the significance and precision of the results.Ethical approval for this study was obtained from the Ethics and Research committee at the Nursing Faculty, King Abdulaziz University. Study enrolment was voluntary and data collection was anonymous, as the participants were not asked to include their names. Detailed information about the study was included on the first page in the online survey tool. Hence, before the participant started the survey, they had the chance to read the information provided and decide whether they were interested in being enrolled in the study. Informed consent was implied by the participants completing and submitting the survey.Out of 270 total respondents, 34 were excluded because their age was not within the inclusion criteria. Most subjects (66.5%) were older than 20, and the majority of the total (86.4%) were female. About 84% of study subjects were students of a variety of disciplines, and remaining participants specified whether they were employed or unemployed. More than 50% had a bachelor’s degree, and 96.2% were single. The parent’s status was “together” in 78% of the study subjects; the remaining parents were separated, divorced, or “other” (see Table 1).The internal consistency of the three measurements was assessed and indicated acceptable reliability; Cronbach’s alpha of all three measurements was above 0.75. Descriptive statistics of the total scores of the three measurements are also given in Table 2.From the descriptive statistics, it can be observed that the study participants showed a moderately negative belief toward mental illness and indicated a moderately positive attitude toward parental support. Additionally, the participants’ mean scores of mental help-seeking indicate a highly positive attitude toward mental help-seeking. To observe the linear relationship among the three outcome variables (scores of three measurements), we used Pearson’s correlation analysis, which showed no significant relationship among the three variables. The comparison of mean scores of three measurements (parental support, beliefs towards mental illness, and metal help-seeking attitude) in relation to the study demographic variables (age groups, gender, occupation, educational status, and parents’ status) showed no statistically significant difference in the mean scores of any of these three measurements across the any of the five study variables. This bivariate comparison does not provide any evidence that the mean scores are statistically significantly different across the categories of the five study variables (see Table 3).In this study, participants showed a moderately negative attitude toward mental illness and people with mental illnesses overall. Specifically, about 40% of the participants reported a negative attitude toward the capability of a person with mental illness to be functional as a parent, make friends, and be a trustworthy person within a team. Participants’ attitude toward other aspects, such as people whether with mental disorders are likely to be criminals or have unpredictable behaviors [25], was mostly positive or favorable. Thus, the attitude of the participants could be considered mixed according to each item analysis in the BMI tool. The study results are consistent with a study that was conducted with a similarly aged population in Saudi Arabia [14]. Among 575 undergraduate students, it was found that they reported a mixed attitude toward people with mental illness, with 32% reporting that they could not maintain a friendship with a person who has a mental illness and 82.5% reporting that people deserve the same rights as anyone else [14]. Moreover, 66.5% of 650 Saudi adults aged > 18 years old reported negative attitudes toward people with mental illness concerning treatment, work, marriage, and recovery, and toward getting professional help [1]. This finding was incongruent with another study that was done with 232 undergraduate students who indicated positive attitudes towards all aspects of mental illness [26]. Based on the available data, it is evident that Saudi young adults have a varied attitude toward mental illness and people with a mental illness, which was proved by a previous study conducted with 3464 Saudi adults recruited from the public [15]. A possible explanation for our study’s result of a moderate negative attitude toward mental illness can be explained by the low mental health literacy among young adults in Saudi Arabia. A study on mental health literacy among Saudi youth uncovered that 575 Saudi young adults showed intermediate mental health literacy levels, viewing mental illness as God’s punishment, the evil eye, magic, or demonic possession [14]. A similar finding was noted in a qualitative study done in Saudi Arabia that found that mental illnesses among Muslims were perceived as possession by supernatural forces or punishment from God [27]. Moreover, stigma or shame of mental illness is not an uncommon finding among the Saudi community [2,27].In this study, the results indicated that young adults had a moderately positive attitude toward their parental support. A marked number of separated parents were noted, as 30% of participants reported having divorced parents. This might be a possible explanation for participants’ reporting a moderate rather than high positive attitude toward parental support. A separation or divorce is a highly stressful emotional experience, and long-term negative parental conflict can negatively impact children’s psychological wellbeing. Although no studies have examined the attitude of Saudi young adults about parental support, one study done in Saudi Arabia using Jeeluna® national survey data (12,121 observations) found that a poor relationship with parents was significantly associated with feeling sad or hopeless and worried [27].There is a scarcity of data on the attitude toward mental help-seeking among the young adult Saudi population. The participants in the current study reported a highly positive attitude toward mental help-seeking. However, conflicting evidence was found in the literature [1,14,28]. Researchers found that 45% of 575 Saudi young adults believed that religious healers could treat people with mental illness [14]. Similarly, it was found that spiritual approaches, such as exorcism, are prominent in the Saudi population [28]. The conflicting evidence might be due to the increasing amount of mental health service in the past few years in Saudi Arabia [29]. Further, people’s attitude could be changing with time because of heightened awareness. Further longitudinal studies with larger samples are needed to clarify the conflicting evidence.The data revealed no significant relationship between the three variables despite their logical associations. No studies have been done in Saudi Arabia previously for comparison; however, our results are congruent with a study that indicated no significant relationship between beliefs toward mental illness and help-seeking among young adults in the United States [30]. A contradicting result was found in a study that disclosed a small but significant positive relationship between parental support and help-seeking intentions among 1482 Australian students aged 16 to 18 years [4]. Overall, a possible explanation for our study’s finding might be young adults’ need for independence; too much parental involvement with their help-seeking services could feel pressuring or undesirable to young adults. A study conducted with young adults in China showed that those who had higher levels of parental involvement were more likely to withdraw from professional mental health services [31]. Future studies using advanced mixed-method approaches are recommended to reveal more reliable evidence.No association was found between the outcome variables and the demographic variables of the current study. Overall, it was difficult to link the current study findings with comparable studies, as there is a scarcity of studies examining the relationships of similar variables in the literature. The differences between means in the education variable were near significance (p = 0.06). However, this result was congruent with a study that has been done on 650 Saudi adults, which revealed that education level was not correlated with the attitude toward mental illness and mental help seeking. Nevertheless, the same study reported that employment and male gender were positively correlated with the attitude toward help-seeking [1]. It was not surprising that there was no significance detected of gender and employment with the outcome variables due to lack of variability in the current study sample, where the majority of the sample was females and students. Another study conducted among 1482 students disclosed that parental authoritativeness was associated with greater intentions of seeking mental help from professional sources [4]. However, in the current study, the examined variable was parental status whether separated or together, which might not have necessarily indicted a parenting style. Further studies are recommended to examine the correlation between the different parent variables and mental health variables. The current study limitations include use of self-reported questionnaires. Self-reported data cannot always be independently verified and present a potential source of bias such as recall bias, where participants tend to seek social desirability, hide errors, and exaggerate their knowledge. In addition, this study is limited because of the use of a convenience sample due to limited time and resources with the majority of the enrolled participants were females. However, in the culture of Saudi Arabia women have been understudied in the past [32]. Hence, including more women to make their voice heard is desirable. The use of a cross-sectional design could be a potential limitation as well, because this design examines variables at a single point in time, while some study variables may change over time. For example, the attitudes of participants would best be captured over time, because feelings are subject to change. Finally, the cross-sectional design may have limited detection of possible relationships among the study variables. A longitudinal design is suggested for future studies looking to reveal more reliable data.The perception of young adults toward mental illnesses and people with mental illnesses was moderately negative. This result highlights the possibility of low mental health literacy among this young adult population, which could increase the stigma and negative stereotypes among clients with mental illness. The study showed that young adults positively perceived their parents’ support; however, the divorce rate among the participants could not be ignored. Moreover, the results indicated that young adults would most likely seek professional mental help, evidenced by their highly positive attitude toward mental help-seeking. There was no relationship between parental support, beliefs toward mental illness, and mental help-seeking attitude. It is highly recommended that education about mental health and complications of stigma be encouraged and applied in schools, universities, awareness-raising events, and the media. Finally, making community mental health services easily available might help in reducing stigma and providing early detection and treatment of young adults with psychological problems.Based on the study findings, multiple recommendations are suggested. Young adults must be educated about mental illnesses at an early age to improve their understanding of mental illness, decrease negative attitudes and stereotypes against people with mental illness, and guide them to available professional mental help services. For example, it might be useful to implement mental health literacy events in schools and public places, such as shopping malls, and create trusted media platforms to raise awareness surrounding mental illness and acknowledge available resources. Moreover, customized workshops for parents to minimize the negative psychological impact of separation and divorce on their children are recommended. Finally, there is a need to conduct similar studies using mixed-methods and longitudinal approaches and recruit larger sample sizes to enhance the accuracy of the findings and enrich the literature on the progression of the attitudes toward and knowledge about mental illness in Saudi Arabia.Conceptualization, A.M., M.B., L.S., and N.A.; methodology, E.A., R.J., A.B., and S.A.; software, E.A., R.J., A.B., and S.A.; validation, A.M., M.B., L.S., and N.A.; formal analysis, A.M., E.A., R.J., A.B., and S.A.; investigation, M.B., L.S., and N.A.; data curation, E.A., R.J., A.B., and S.A.; writing—original draft preparation, A.M.; writing—review and editing, M.B., L.S., and N.A.; supervision, A.M.; project administration, A.M. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The authors declare no conflict of interest.Distribution of characteristics of study subjects (n = 236).Reliability (internal consistency) of three instruments and descriptive statistics of outcome variables: Parental Support, Beliefs Toward Mental Illness, and Mental Help-Seeking Attitude (n = 236).Comparison of mean values of three outcome variables in relation to study demographic variables.

Med-MDPI/ijerph_5/ijerph-17-15-05616.txt

@@ -0,0 +1 @@
+(1) Background: The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) began spreading across the globe in December and, as of 9 July 2020, had inflicted more than 550,000 deaths. Public health measures implemented to control the outbreak caused socio-economic havoc in many countries. The pandemic highlighted the quality of health care systems, responses of policymakers in harmony with the population, and socio-economic resilience factors. We suggest that different national strategies had an impact on mortality and case count. (2) Methods: We collected fatality data for 17 countries until 2 June 2020 from public data and associated these with implemented containment measures. (3) Results: The outcomes present the effectiveness of control mechanisms in mitigating the virus for selected countries and the UAE as a special case. Pre-existing conditions defined the needed public health strategies and fatality numbers. Other pre-existing conditions, such as temperature, humidity, median age, and low serum 25-hydroxyvitamin D (25(OH)D) concentrations played minor roles and may have had no direct impact on fatality rates. (4) Conclusions: Prevention, fast containment, adequate public health strategies, and importance of indoor environments were determining factors in mitigating the pandemic. Development of public health strategies adapted to pre-existing conditions for each country and community compliance with implemented policies ensure the successful control of pandemics.A novel coronavirus (CoV) was identified in December 2019 in Wuhan, China. This highly contagious and previously unknown pathogen caused complications in the respiratory system and was named severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). WHO first declared the outbreak of coronavirus disease 2019 (COVID-19) a six public-health emergency of international concern, before subsequently declaring it a pandemic [1]. Human coronaviruses (HCoVs) are enveloped, non-segmented, positive-sense, single-strand ribonucleic acid (RNA) viruses [2]. HCoVs can infect animals and humans, causing respiratory, hepatic, and neurologic diseases [3]. CoVs are divided into four genera: alpha-coronavirus (α), beta-coronavirus (β), gamma-coronavirus (γ), and delta-coronavirus (δ) [2,3,4,5]. Human coronaviruses were identified in late 1960 and are known to infect humans, other mammals, and birds. To date, six HCoVs have been categorized. In 2003, a virus was identified in Guangdong province in China causing severe acute respiratory syndrome (SARS). Later, the virus was confirmed as a member of the beta-coronavirus family and was named SARS-CoV [6]. A decade later in 2012, two Saudi Arabian nationals were identified to be infected with another coronavirus. This virus belongs to the beta-coronavirus (β) family and was termed the Middle East respiratory syndrome coronavirus (MERS-CoV). In 2019, a novel CoV caused the outbreak of a SARS-like illness in the animal market of the Chinese city of Wuhan and was termed severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) [7].The source of origination and transmission are relevant factors for the development of preventive strategies and treatment protocols. SARS coronaviruses may have circulated in humans before causing the outbreak in 2003 [8]. Rhinolophus bats were identified as having anti-SARS-CoV antibodies, thus suggesting the bats were a source of the virus [9]. The Middle East respiratory syndrome (MERS) coronavirus (2012) in Saudi Arabia is a beta-coronavirus originating from camels as the primary host [10]. In the case of SARS-CoV-2, only bats were identified as key reservoirs [11,12]. COVID-19 originated in Wuhan and was carried through migration by the population to different cities and other provinces around Hubei [12]. Infection rates increased rapidly and were controlled by a rapid and unprecedented lockdown of cities in the Hubei province followed by further control mechanisms for the whole country [12,13]. This unparalleled approach effectively prevented an exponential increase of infection on a national basis in China [12,13]. Similar measurements were implemented by many other countries once cases started appearing within their borders. Enforcement methods and periods of implementation varied from country to country and delivered mixed results. Factors influencing transmission, disease progression, susceptibility, and virulence are surfacing slowly [2]. Park et al. studied clinical and demographic characteristics of 44,672 COVID-19 cases in China and noted that everyone without age limitation is susceptible [14]. The fatality rate is highest in elderly, immunocompromised individuals with comorbidities [14]. The mean or median incubation time of COVID-19 is less than 13 days and should perhaps be considered to be lower than 4–6 days, similar to SARS-CoV and MERS-CoV median incubation times of 4.4 and 5.5 days, respectively [14]. SARS-CoV-2 is highly contagious and infection may occur by respiratory droplets carrying the virus or through contact with infected surfaces (formites) and subsequently touching mouth, nose, and eyes [15]. Liao et al. reported that particles smaller than 1 μm remain in the air for an unknown time and particles larger than 5 μm settle within one hour [16]. Particles with sizes larger than 10 μm, also called droplets, settle much more quickly and cause infections through fomites [16]. Thus, airborne transmission happens mainly via smaller particles. Fine particles smaller than 5 μm have the highest virulence factor because they remain in the air longer and can easily reach the lower respiratory tract [16]. Particles with a size larger than 5 μm can reach the upper respiratory tract [16]. Hou et al. showed that the nasal cavity is the first entry and proliferation point of airborne particles [17]. Accordingly, they argue, preventive measures such as wearing masks and using complementary therapeutic strategies on the nasal site may help control the spread of SARS-CoV-2 [17].In this work, we analyzed the effectiveness and impact of national strategies and their impact on fatality rates by evaluating data of 17 countries until 2 June. We also investigated the public health strategies in the UAE, which has not been previously represented in the literature. We examined the relationship between pre-existing conditions, climate, population size, mean age, and level of SARS-CoV-2 control by implemented measures.In this study, we analyze confirmed COVID-19 cases and fatalities in selected countries using the data on the worldwide geographical distribution of SARS-CoV-2 as of 2 June 2020 from the European Centers for Disease Control and Prevention (ECDC) website on 3 June 2020 [18]. For the first time, we highlight the situation in the United Arab Emirates (UAE) by discussing its national strategies and their impact on case count and mortality. Furthermore, we report the experiences of USA, Italy, and India as special cases. We compare several countries with similar population size in terms of COVID-19 case count and mortality. Sweden–Switzerland–Portugal, Germany–Turkey, and France–Spain are compared as examples of European countries. Latin America is represented by analysis of data from Brazil, Peru, Ecuador, and Mexico. South Korea and Japan were chosen as examples of Far Eastern countries with previous experience of SARS and efficient national strategies to combat COVID-19. We depict the results of the controversially discussed factors of average serum (25(OH)D) concentrations and median age, in addition to geographical factors of temperature and humidity. The outcomes of the analysis underline the importance of pre-existing conditions (health care system preparedness, median age, population size, population density, geographical conditions) and adequate public health strategies within a suitable time-frame. We comment on the effectiveness and problems of the implemented control mechanisms, including the general lockdown- and post-lockdown implications. Finally, we conclude with recommendations for the COVID-19 outbreak and future pandemics.Progression of CoV to severe pneumonia is directly related to advanced age, history of metabolic syndrome, smoking, and other chronic conditions such as cardiovascular disease [19]. Countries with a high percentage of advanced age and high population density in urban areas need to impose swiftly efficient measurements to reduce fatalities [19]. Primc et al. analyzed the public health measures of European countries during the pandemic [19]. They concluded that qualitative development of the health care system and improving protective health conditions such as serum (25(OH)D) concentrations may be important for the control of COVID-19 [19]. To investigate the importance of serum vitamin D levels, median age, temperature, and humidity we compare infection control measures and their impact on COVID-19-related fatalities in Portugal, Sweden, and Switzerland (Figure 1).Figure 1 shows the highest fatality numbers in Sweden, followed by Switzerland and then Portugal. Sweden is one of the few European countries refraining from nationwide closures and lockdown [19]. This, in combination with a lack of clear recommendations for the public in how to prevent viral spread, especially in indoor environments, may have aggravated SARS-CoV-2 spread in Sweden.Primc et al. reported a possible relationship between low 25-hydroxyvitamin D (25(OH)D) concentrations and susceptibility to COVID-19 [19]. Vitamin D deficiency is prevalent in the northern hemisphere due to reduced solar irradiation especially in the winter months [20,21,22,23]. The infection rates of COVID-19, which started in the month of December 2019, are now showing exponential growth in Northern Hemisphere countries, [19,24,25]. According to a recent review by Grant et al., 25-hydroxyvitamin D (25(OH)D) concentrations may play a relevant role in the progression of COVID-19 [20]. Sufficient plasma (25(OH)D) concentrations around 75 nmol/L ensure cellular immunity, decrease the risk of infections caused by microorganisms, and protect against metabolic syndrome and other chronic illnesses [20]. Severe vitamin D deficiency is defined as serum 25(OH) D levels lower than 30 nmol/L [21]. Factors such as screen-based entertainment, sedentary lifestyle, and less outdoor physical activity are directly related to poor endogenous vitamin D synthesis [21,22,23]. Higher susceptibility of advanced age individuals to COVID-19 may confirm this hypothesis. Commonly, serum 25-hydroxyvitamin D (25(OH)D) concentrations are less than 30 nmol/L in elderly individuals and geriatric patients due to indoor dwelling, physical inactivity, and increased pharmaceutical drugs intake, especially in the winter [20,21,26]. Sufficient serum vitamin D levels above 70 nmol/L may have a positive impact on recovery rate and survival of patients during illness [20,21,22,23]. Ilie et al. associated mean serum vitamin D levels in elderly populations of European countries with COVID-19 infection and mortality [27]. They compared the mean serum vitamin D levels for the populations in Portugal (39 nmol/L), Spain (42.5 nmol/L), Switzerland (46 nmol/L), UK (47.4 nmol/L), Italy (50 nmol/L), Germany (50.1 nmol/L), Turkey (51.8 nmol/L), France (60 nmol/L), Sweden (73.5 nmol/L), and other countries [27]. Portugal has the smallest, Switzerland represents the average, and Sweden the highest mean serum 25-hydroxyvitamin D (25(OH)D) levels and similar population sizes of 10,281,762, 8,516,543, and 10,183,175, respectively (Figure 1) [18,27]. The highest number of fatalities is in Sweden, followed by Switzerland, and then Portugal, although the opposite trend was expected due to mean serum vitamin D levels and population number. We can conclude that mean plasma 25-hydroxyvitamin D (25(OH)D) levels have no direct impact on fatality rates and remain controversial in their relation to COVID-19. The results of Figure 1 show that the highest COVID-19-related mortality was in Sweden, followed by Switzerland and Portugal with median ages of 41.1, 42.7, and 44.6, respectively [24]. Therefore, we also can conclude from our analysis that median age is not a sufficient variable to predict fatality.According to Scafetta, temperature and relative humidity have an impact on viral transmission [24]. Temperatures from 4 to 12 °C and relative humidity of 60–80% with gentle wind speeds increase the susceptibility to respiratory tract infections and the spread of SARS-CoV-2 [24]. Our analysis reveals that temperature and relative humidity are not the only factors with direct impact on fatality rates. The first fatal cases were announced in each of the three countries in March, while the first confirmed COVID-19 case was reported in Portugal, Switzerland, and Sweden on 3 March, 26 February, and 1 February 2020, respectively [18,24]. The average temperature in March 2020 was the highest in Portugal at 12.4 °C, followed by Switzerland with 2.2 °C, and the lowest was Sweden with −3.3 °C [24]. The lowest relative humidity was recorded in Portugal with 70%, Switzerland’s was 73%, and the highest was in Sweden with 88% [24]. According to these numbers, the spread of COVID-19 should be higher in Portugal, but again it was Sweden. An explanation for the higher fatality in Sweden is the cold weather itself. The virus cannot survive for long outside in the cold, but spreads easily because people remain indoors [24]. Indoor air is more suitable for the spread of SARS-CoV-2, especially due to close contact of people in unventilated, closed rooms without UV irradiation [24].A coordinated effort to educate the population about the means of preventing transmission and infection in indoor and outdoor environments during a pandemic is crucial to avoid uncontrolled spread of the virus. Indoor and outdoor climate are related to each other and aggravate viral transmission under given conditions. Indoor environments can proliferate SARS-CoV-2 infection by raised viral loads through droplets and direct contact on infected surfaces more than an outdoor climate. This also confirms the spread of COVID-19 in household contacts and the need for indoor social distancing. According to Luo et al., pre-symptomatic cases within the household may transmit the virus rapidly [28]. Especially vulnerable and high-risk individuals are at risk of being infected. Luo et al. concluded in their prospective cohort study that the risk of transmitting COVID-19 is highest in household contacts [28]. Public health strategies need to include habitancy and dwelling patterns in relation to climate in measures to control outbreaks. Including dwelling modes in virus containment and control is essential, especially if nationwide closure and lockdowns are not imposed. In addition, a mask wearing strategy needs to be enforced in indoor and outdoor settings, especially in countries with a high population density. Wearing masks can reduce the airborne transmission by droplets and smaller particles, especially in indoor environments [16,17]. A study compared community-wide mask compliance in relation to the number of confirmed SARS-CoV-2 cases/fatalities in Hong Kong, Singapore, and other countries [29]. Singapore has the second highest population density in the world, directly followed by Hong Kong [29]. Due to their previous experience with the SARS virus in 2003, 97% of the studied population in Hong Kong began using masks when the first COVID-19 confirmed case emerged [29]. This mask-wearing strategy combined with social distancing, personal hygiene, cancellation of social gatherings, use of the home office, and school closures resulted in the effective control of the SARS-CoV-2 transmission compared to other neighboring countries [29]. The authors conclude that a mask-wearing strategy irrespective of COVID-19 symptoms mitigates infection of susceptible individuals [29]. Another study from Cowling et al. analyzed data from 60 general outpatient clinics and three cross-sectional surveys in Hong Kong [30]. They findings indicate that rapidly implemented public health measures and the compliance of the population mitigated the SARS-CoV-2 outbreak in Hong Kong [30]. The population readily embraced the recommended prevention methods and social distancing, together with restrictions on the borders, while cases and their surrounding contacts were successfully tracked, isolated, and quarantined [30]. According to the authors, these measures controlled the SARS-CoV-2 outbreak more efficiently and reduced socio-economic tensions without strict nationwide closure and lockdown during the COVID-19 outbreak compared to other countries [30]. The measures also reduced the transmission of influenza in the month of February 2020 [30].High levels of fatality were recorded in advanced age patients with immune systems compromised by diabetes, hypertension, and cardiovascular disease during SARS-CoV, MERS-CoV, and COVID-19 [31]. Deng et al. compared the characteristics of the fatality group of patients with those of the recovered group [31]. Advanced age, dyspnea, comorbidities, low oxygen saturation, high white blood cell (WBC) count, low lymphocytes, and high levels of C-reactive protein (CRP) are significant indicators of mortality [31]. Giang et al. directly related people aged 65 and above with death rates [32]. They found that the measures of social distancing, lockdown, quarantine, and a high-quality health care system with appropriate numbers of health care workers and emergency rooms helped to decrease fatality rates [33]. The quality and preparedness of the health care system in a country defines the thin line between survival and death.In most COVID-19 confirmed cases, symptoms such as fever, dry cough, fatigue, runny nose, or other upper respiratory symptoms appeared in SARS-CoV and MERS-CoV [34,35,36,37,38]. Distinguishing COVID-19 from other known respiratory tract viruses clinically is a difficult task due to its non-specificity. A precise diagnosis can be carried out by reverse real-time quantitative polymerase chain reaction (RT-qPCR) testing and radiological study of the cases [35,39,40,41,42]. Fast and reliable methods of COVID-19 testing can be life-saving tools to eradicate the virus in Far Eastern countries. To validate this thesis, we compared the fatality numbers in South Korea and Japan during the SARS-CoV-2 pandemic (Figure 2).The first confirmed COVID-19 case in Japan was announced on 15 January 2020, while South Korea had a confirmed case five days later [18]. The first fatality in Japan occurred on 13 February 2020, while South Korea’s first COVID-19 fatality occurred on 21 February [18]. The total number of fatalities in Japan on 2 June 2020 was 894 within a population of more than 126 million, and South Korea’s was 24 within more than 5 million [18]. The low fatality rate in both countries indicates excellent pandemic control. Nonetheless, the factors relevant to this success should be discussed. The median age in Japan of 48.6 years is the highest globally, and could have led to the highest global fatality rates because of the higher susceptibility of elderly [28]. The number of deaths is low compared to Mexico with a similarly sized population (>126 million) and 10,167 fatalities until 2 June 2020 [18]. The median age in South Korea is 43.2 [28]. Both South Korean and Japan adopted early intervention strategies to control the viral spread with measures of social distancing and widespread COVID-19 testing, combined with strict follow-up of clusters of infections [43]. Both societies gained experience during the SARS outbreak, whereby they developed the habit of wearing masks, which prepared them for this pandemic. South Korea introduced a dynamic and quick response to the needs of the time. The measurements were implemented in harmony with the public through open information sharing and transparency, thus ensuring public support and collaboration [43]. These joint efforts orchestrated a supply of masks, their distribution, and smart use [43]. A drive-through screening system was used to support general widespread COVID-19 testing and all individuals were followed up via personal tracking [44]. As a result, the pandemic was controlled without a complete lockdown in both countries mainly by widespread COVID-19 testing, individual tracking, and follow-up of clusters. Public support was essential through social distancing, wearing of masks, personal hygiene, transparency during the pandemic, collaboration to meet the needs of the market in mask production, and, most importantly, widespread COVID-19 testing with strict follow up mechanisms for clusters.Many countries responded with different control mechanisms compared to the experiences of China, South Korea, Japan, and Hong Kong [29,30,44,45]. Highly industrialized, European countries were expected to best cope with the pandemic due to their level of sustainable development. To verify this thesis, we analyzed the data of selected European countries and grouped them according to their population size and/or number of fatalities [18]. Germany (population 2018: 82,927,922) and Turkey (population 2018: 82,319,724) faced their first reported COVID-19 case on 28 January 2020 and 12 March 2020, respectively [18]. Until 2 June 2020, the number of confirmed cases in Germany was 182,028, while Turkey had 164,769 cases. Germany had reported a total number of 8522 fatalities, while Turkey recorded 4563 fatal cases (Figure 3) [18].The COVID-19 fatality numbers in Germany are almost double the number of fatalities in Turkey. The reason may be due to the higher median age in Germany [24]. The median age of the German population is the second highest after Japan [24]. The arrival of the pandemic in Turkey was more than one month later than in Germany. Both countries reacted after the first COVID-19 fatality with a partial lockdown and social distancing measures, and closure of workplaces, schools, and universities [19]. The borders were closed for non-essential travel [19]. Citizens stranded in other countries were repatriated and quarantined for two weeks. All border crossings were controlled and checked. Authorities started with widespread testing and followed-up on clusters and asymptomatic cases. Turkey started producing masks, effective personal protection equipment (PPE), and respirators to cope with demand inside and outside the country. According to the curves in Figure 4, the peak of the pandemic is over for both countries. Both countries were well prepared to handle the pandemic due to well-developed health care systems and by taking the needed measures to control the spread of the virus. Italy and Spain have younger populations compared with Germany [24]. The median age in Germany is 47.8, while in Italy it is 46.6 and in Spain it is 43.9 [24]. Furthermore, the mean serum vitamin D levels for the populations in Italy and Spain are similar to those of Germany [27]. These factors should have resulted in lower SARS-CoV-2 susceptibility and better control of the outbreak. However, Italy and Spain were confronted with a massive outbreak of COVID-19 in 2020. Italy reported the first confirmed case in 31 January 2020 and the first fatality on 23 February 2020 [18]. Spain confirmed the first case on 1 February 2020 and the first fatality occurred almost one month later on 5 March 2020 [18]. Spain had recorded 239,638 cases and 27,940 fatal cases with a population of approximately 46,723,749 [18]. The number of COVID-19 cases in Italy was 233,197 with a population size of 60,431,283 and 33,475 fatalities until 2 June of 2020 (Figure 4) [18].The highest fatalities were recorded towards the end of March in Italy, while the number of confirmed cases started to decline 10 days earlier (Figure 4). Widespread COVID-19 testing was initiated in the beginning of March with a delay of one month after the first confirmed case. Both curves descended slowly towards 2 June 2020 which indicates that the SARS-CoV-2 outbreak was under control prior to the summer months (Figure 4). The delay of widespread COVID-19 testing for one month aggravated the pandemic in Italy. Due to this delay, the implemented state of emergency, nationwide closures, lockdown, and further measures did not effectively control the spread of the virus. In addition, lower median age and similar serum vitamin D levels compared to Germany did not have any positive impact on the COVID-19 disease course and did not reduce fatality numbers in Italy.France and the United Kingdom (UK) were also severely hit by the COVID-19 pandemic. Both countries have around 67 million inhabitants [18]. The total number of fatalities in France was 28,833 and in the UK 39,045 until 2 June 2020 (Figure 5) [18].France recorded 152,091 and UK reported 276,332 COVID-19 cases until the same date [18]. The highest number of fatalities were counted in France between 31 March and 20 April 2020. UK registered a steady decline of COVID-19 related fatalities after 20 April 2020. According to Ilie et al., the mean serum vitamin D level of the French population is 60 nmol/L, which is higher than that for the populations of Portugal, Spain, Switzerland, UK, Italy, Germany, and Turkey, with 39, 42.5, 46, 47.4, 50, 50.1 and 51.8, respectively [27]. As a result, low serum vitamin D levels have no direct impact on COVID-19 fatality rates. The COVID-19 case and mortality counts on the European continent reveal mixed results that do not depend on temperature, mean age, or mean serum 25-hydroxyvitamin D (25(OH)D) levels, but on the preparedness of the health care system and suitable public health strategies.The next epicenter evolved on the American continent. The first confirmed case of SARS-CoV-2 infection in the United States of America (USA) was reported on 21 January 2020 one day after that of South Korea [18]. The fatality rate on 10 April 2020 was 2% for South Korea and 3.6% in the USA [46]. South Korea, due to its previous SARS experience, reacted swiftly with sophisticated, new measures including transparent, community-supported policies and widespread COVID-19 testing with strict individual follow-up tracking mechanisms of clusters of transmission [43,44]. The USA had recorded 105,147 total COVID-19-related fatalities and 1,811,277 confirmed cases as at 2 June 2020 (Figure 6) [18]. Figure 6a shows a sharp increase in the confirmed COVID-19 cases after 17 March 2020.The fatality curve in Figure 6b shows a peak on 15 April 2020. Thereafter, a steady decline in fatality numbers was reported. This curve may indicate that the worst of the outbreak in the USA is over and the peak lasted for around one month after the first reported case of COVID-19 on 17 March 2020. The US started to test for COVID-19 one month after this first confirmed case [46]. This delay of testing caused a rapid viral spread through the population because the infected cases were not identified and followed-up [46]. The unexpected mass of severe cases strained the unprepared health care system. Other compounding factors, such as a lack of drugs, treatment protocols, or vaccines resulted in a health disaster for the USA (105,147) (fatalities in USA due to COVID-19 at the time of writing), in addition to countries such as the UK (39,045), Italy (33,475), Brazil (29,937), France (28,833), Spain (27,940), and Mexico (10,167), which had recorded mortalities in excess of 10,000 prior to 2 June 2020 [18]. Asymptomatic cases are a leading cause of the viral outbreak [46]. Widespread and random testing are effective methods to identify clusters of infections and mitigate the transmission patterns through follow-up and quarantine [46]. Re-testing cases during the isolation and post-quarantine is essential [46].The first COVID-19 case in Latin America was reported on 26 February 2020 [47]. Subsequently, SARS-CoV-2 spread at a rapid rate across the continent. We analyzed the data for selected countries in Latin America and compared their number of fatalities from the European Centers for Disease Control and Prevention (EDCD) (Figure 7) [18].Brazil has reported the highest number of fatalities with 29,937, followed by Mexico with 10,167, Peru with 4634, and Ecuador with 3394 [18]. At the present time, it appears that Brazil and Mexico have reached the peak of the pandemic according to Figure 4, while Ecuador and Peru may still be in the initial phase of the pandemic if the curve continues to rise [18]. The COVID-19 pandemic in Ecuador may also overlap with further viral outbreaks of dengue and/or zika, when the country already lacks a suitable number of intensive care unit beds [47]. The arrival of further epidemics such as measles and malaria may exacerbate the burden on the already depleted health care system in Latin America [47]. The situation in South America is far from resolved and could develop into the worst COVID-19 outbreak this year.China, South Korea, Singapore, Hong Kong, and Taiwan are among the countries that were able to control the COVID-19 outbreak by reducing infection numbers [41,48,49,50,51]. Nationwide closures, lockdowns, travel restrictions, personal hygiene, and social and workplace distancing may help to prevent infection. A modeling study from Singapore investigated the impact of control measures implemented to reduce the spread of COVID-19 [49]. The most effective method was a combination of isolating infected cases, strict quarantine, and closure of schools and workplaces [16,49]. The lockdown of educational establishments and offices (workplace distancing) paved the way for online solutions, which helped to education and work to continue from home. After the outbreak in Wuhan, China, the Korean government activated a 24/7 emergency response system to screen all travelers entering the country from that city [49]. In Taiwan, proactive and comprehensive health checks of passengers from Hubei province were established quickly from the beginning of January 2020 [50]. The Taiwanese Centers of Disease Control (CDC) tested 2105 cases by February 28 using multiplex PCR analysis with FilmArrayTM Respiratory Panel and confirmed 34 COVID-19 patients [50,51,52]. According to Hsih et al., COVID-19 is more contagious than seasonal respiratory pathogens but infected cases have common clinical and laboratory results [52]. Therefore, each suspected case needs close follow-up through individual tracking, gathering of mobility and contact data, isolation, and quarantine [52]. Hsih et al. underline the elevated pandemic level of SARS-CoV-2 manifested by prolonged viral shedding, and the large number of asymptomatic and mild illness cases that remain undiscovered and continue to spread COVID-19 [52]. They also note that some recovered patients still had detectable virus levels for almost two further weeks [52]. Furthermore, some patients showed negative COVID-19 test results from their naso-oropharyngial system but positive results in their sputum or fecal specimens [52]. The authors are concerned about possible SARS-CoV-2 transmission through the fecal–oral route [52].We investigated the SARS-CoV-2 pandemic in the United Arab Emirates (UAE) as an example of a highly populated, globally interconnected country with an equatorial hot climate and excellent control of the COVID-19 outbreak. In this work, we present the level of public health responses and control of SARS-CoV-2 in the UAE, and underline that publications about COVID-19 in the UAE are very rare. Figure 8a shows the confirmed cases and Figure 8b the fatalities in the UAE.The population in the UAE totals around 7 million according to the 2018 census [18]. The total number of confirmed cases was 35,192 and the number of fatalities was 266 on 2 June 2020 [18]. The first fatality occurred on 22 March, almost one month after the first reported case [18]. During the month of March, the median temperature was 23.6 °C and the relative humidity was 59%. This high temperature and low relative humidity were not conducive to SARS-CoV-2 proliferation [24]. Similar climatic conditions also exist in Hong Kong, Singapore, and other Gulf countries [24]. All of these countries are characterized by warm temperatures and low humidity in winter months, dense populations in their capital cities, well-connected international airports, and low median age (with the exception of Hong Kong in the case of the latter) [24]. These countries can be counted as examples of excellent pandemic control.The UAE authorities responded in a rapid, flexible, effective, and transparent way to curb the spread of SARS-CoV-2. Rapidly implemented public health measures, closure of borders, nationwide closures of educational institutions, complete lockdowns for certain periods, and work-from-home schemes reduced mobility and helped to efficiently control the viral outbreak in the UAE [53]. Via stepwise measures, on 25 March 2020 the UAE General Civil Aviation Authority suspended all flights into and from the country [53]. The “National Disinfection Programme” was implemented by the Ministry of Health and Prevention (MoHaP) and Ministry of Interior (MOI) to disinfect public facilities, public transport, and roads [53]. The disinfection campaign was initiated on 26 March 2020 and performed every night from 8 pm to 6 am [53]. Nationwide COVID-19 testing was launched by the Ministry of Health and Prevention, the Ministry of Interior, and the National Emergency Crisis and Disasters Management Authority at the end of March 2020 [18]. Drive-through COVID-19 testing facilities were opened in many locations throughout the UAE and continue to serve the population together with mobile laboratory units, home testing for people of disabled hospitals, and licensed medical centers [53].Recommendations for community-wide social distancing and a mask-wearing strategy complemented by gloves and personal hygiene measures were embraced by the population. The beginning of the fasting month of Ramadan eased the closure of restaurants, shops, and other facilities for entertainment and leisure activities. Prayers were suspended shortly before the month of Ramadan, and this suspension was extended on 16 March 2020 by the National Emergency Crisis and Disaster Management Authority and the General Authority of Islamic Affairs and Endowments until further notice. The Ministry of Education announced school and university closures on 8 March 2020 coinciding with the spring break and implemented a distance learning initiative on 22 March 2020 [53]. Education continued without disruption through online teaching after the spring break and is expected to remain distance-learning-based until the end of the educational year in summer 2020 [53]. Remote working was adopted by most institutions of the federal and local governments in the UAE to protect employees and customers. The pre-existing advanced technological infrastructure of the UAE enabled remote working and online meetings throughout the educational and public sectors [53]. Numerous apps, such as ALHOSN UAE, were created to track cases and advise the population [53]. Offices and malls briefly closed and reopened towards the end of May 2020 with strict control measures to prevent COVID-19 spread. The closure of all shopping malls, with the exception of pharmacies, food retail outlets, cooperative societies, grocery stores, and supermarkets, was announced on 25 March 2020 [53]. Children below 12 years and individuals above 60 years were not permitted to enter malls and outlets [53]. Everyone entering any facility is required to wear surgical masks and gloves, and is screened for symptoms of fever. Disinfectant dispensers are widely distributed in all buildings. Most institutions, such as Ajman University in Ajman, UAE, contributed unreservedly to the campaign by disseminating the recommendations and applying the necessary measures in a timely manner (Figure 9).The COVID-19 outbreak is currently under control in the UAE: the number fatalities is low and shows a downward trend (Figure 8). The health care system was rapidly supported and prepared for a mass outbreak. Steadily rising temperatures from April 2020 and high relative humidity did not prevent COVID-19 from spreading. The curve in Figure 8b peaks on 12 May 2020 in terms of the number of fatalities, while in Figure 8a the peak for confirmed cases appears around 10 days later [18]. Increasing UV irradiation, temperature, and relative humidity may have been supportive factors in controlling the SARS-CoV-2 outbreak in the UAE.The influence of hot weather and increasing humidity in the coming months should be questioned. The virus cannot be inactivated solely by high temperatures, and can survive in the human body [24]. The outdoor climate is highly interconnected with the indoor climate and human dwelling patterns. With the steady rise of temperatures above 45 °C and high humidity, indoor environments should be included in thinking about the virus. When the pandemic began at the end of February, the temperature was suitable for outdoor activities. In very hot climates, people gather in indoor environments. The summer months are marked by indoor dwelling, and spending time in air-conditioned closed apartments, offices, buildings, and shopping malls. The indoor environment in summer may cause aggravated SARS-CoV-2 transmission patterns and an increase in COVID-19 cases and fatalities. New waves of infection can only be prevented by continuing the community-wide, strict measures of social distancing, personal hygiene, wearing of surgical masks and gloves, nationwide testing, and individual tracking of clusters and their contacts. Equatorial countries may witness an increase in cases and/or fatalities during the coming months. The spread of SARS-CoV-2 in the summer months can be prevented by adding new strategies to the existing measures. Our recommendation for countries with hot climates in summer is to prevent an influx of cases from other countries and to control indoor environments. The latter can be achieved via sufficient ventilation several times a day, continued strict disinfection protocols using antimicrobial agents/disinfectants for surfaces, and use of humidifiers equipped with essential oils and/or antimicrobial plant extracts. Split air conditioning systems may be better than central systems. These air conditioning systems need to be regularly cleaned and maintained. Sterilization of indoor environments may be possible using UV lamps as an alternative. These measures may inactivate the virus and reduce the number of viral droplets in indoor air. The temperature in closed rooms may not reach 4–12 degrees Celsius, but virus-loaded droplets can easily infect anyone in the vicinity by entering the naso-pharyngial tract or the eyes if the viral density in the air is high, or through contact transmission by fomites [16,17,24]. However, airborne transmission is a controversial issue, which depends on climatic pre-existing conditions (indoor/outdoor environments) and needs a case-by-case approach when public health measures are to be implemented. According to Cheng et al., a community-wide mask-wearing strategy may decrease the shedding of respiratory droplets or saliva from symptomatic or asymptomatic COVID-19 cases [29]. Pecho-Silva et al. indicate that airborne transmission can occur due to floating viral droplets suspended in the air [54]. They suggest that droplet nuclei may be only produced by specific medical treatment protocols such as intubation and nebulization [54]. Stadnytskyi et al. reported that SARS-CoV-2-loaded droplet nuclei can be transmitted from an asymptomatic case while speaking in a closed room with stagnant air [55].As a result, transmission by fomites and direct inhalation of droplets in indoor environments is a serious problem in closed, unventilated, and crowded rooms with very poor air exchange. Although the issue remains controversial, strict guidelines should be followed to prevent viral infections until further research emerges. India implemented strict nationwide lockdowns to address the problem of viral transmission; however, by keeping the population indoors it may have created precisely these unventilated, crowded room scenarios that can lead to a high number of cases and fatalities.The Indian government reacted rapidly with very strict nationwide closures and a complete lockdown, and prevented an uncontrolled SARS-CoV-2 outbreak from March to June 2020 [56]. The first fatality in India was recorded on 13 March 2020 (Figure 10) [18].The first COVID-19 case appeared with a delay of more than one month on 30 January 2020 [18]. The total number of confirmed cases for India was 198,706 within a population of around 1.4 billion [18]. A total of 5598 fatal cases were reported until 2 June 2020 [18]. The COVID-19 outbreak was controlled firmly until beginning of April 2020, however, incidence thereafter began to increase. The curve currently shows an upward trend in fatalities and a peak is not yet visible in Figure 10. A lack of awareness and a relaxed attitude has resulted in an increase in COVID-19 cases and fatalities [56]. The situation in India is far from over and strict regulations should continue, paired with mask-wearing strategies, nationwide testing, and control of overcrowded indoor environments.We collected and analyzed the implemented public health measures during the SARS-CoV-2 pandemic. Different countries implemented various measures and experienced differing results during the COVID-19 pandemic (Table 1).A lack of preparedness for the sudden onset of COVID-19 by governments and health care systems, combined with inadequate public health strategies and deficiencies in diagnostic mechanisms, treatment options, and management protocols, aggravated the virulence of SARS-CoV-2 globally. There has been a direct impact of public health strategies on case count and mortality. Far Eastern countries with experience of the SARS outbreak controlled the pandemic with the lowest number of COVID-19 fatalities. COVID-19 case and mortality counts in Europe are not directly dependent on temperature, mean age, or mean serum 25-hydroxyvitamin D (25(OH)D) levels, but on preparedness of countries’ health care systems and adequate public health strategies.Infection control measures in indoor environments are essential in viral transmission prevention and need to be taken into consideration. Indoor environments are interconnected with climatic conditions. Indoor climates are crucial to the control of the SARS-CoV-2 outbreak. Development of health care systems by increasing the number of hospitals, critical care units, and health care personnel are also key factors, and are particularly needed for the protection of the vulnerable and elderly population. The preparedness of the health care system includes stockpiles of appropriate and effective personal protection equipment (PPE), fast and reliable testing methods, and cluster and individual tracking of cases and their contacts. Recommendations of social distancing, personal hygiene, avoiding of gatherings, and mask wearing must be strictly followed even in indoor environments and household settings, particularly for the protection of those aged over 65 years. Further COVID-19 outbreaks may be expected due to new waves and mutations of the virus. Rapidly implemented, transparent region-specific public health strategies and community-wide compliance are essential during the COVID pandemic and the post-lockdown period. Increased health literacy in the population can improve the management and control of further global pandemics.Conceptualization, Z.E., H.M.P. and S.H.B.; methodology, Z.E., H.M.P. and S.H.B.; software, Z.E. and H.M.P.; validation, Z.E., H.M.P. and S.H.B.; resources, Z.E.; data curation, Z.E.; writing—original draft preparation, H.M.P., Z.E. and A.A.S.; writing—review and editing, Z.E., S.H.B. and H.M.P.; visualization, A.A.S.; supervision, Z.E., H.M.P. and S.H.B.; project administration, Z.E., S.H.B. and H.M.P.; funding acquisition, S.H.B. All authors have read and agreed to the published version of the manuscript.This work was kindly supported by the Deanship of Graduate Studies and Research, AU, Ajman, United Arab Emirates.We are thankful to the Ajman University Environmental Health and Safety (EHS) department, Ajman UAE, for providing us with the figure “Follow Safety Measures on Campus” and Ms. Fariyal S. Shaikh from New Vision University, Tiblisi, Georgia with her contributions to this paper by preparing the graphical abstracts. We also like to thank Prof. Dr. Florian Stadler and Prof. Dr. Daniel Tan for their advices related to personal hygiene.The authors declare no conflict of interest.Fatalities during COVID-19 in Portugal, Switzerland, and Sweden until 2 June 2020 [18]. First fatality occurred in Portugal on 18 March, Switzerland on 6 March, and Sweden on 12 March 2020.Fatalities during COVID-19 in South Korea and Japan until 2 June 2020 [18]. First fatality in South Korea on 21 February 2020 and Japan on 13 February 2020.Fatalities during COVID-19 in Germany and Turkey until 2 June 2020 [18]. First fatality in Germany on 10 March 2020 and Turkey on 19 March 2020.Fatalities during COVID-19 in Italy until 2 June 2020 [18]. First confirmed case in Italy on 31 January 2020 and first two fatal cases on 23 February 2020.Fatalities during COVID-19 in France and UK until 2 June 2020 [18]. First fatality in France on 15 February 2020 and in UK on 7 March 2020.Confirmed COVID-19 cases/fatalities in the USA until 2 June 2020 [18]. First fatality in US on 1 March 2020. From left to right: (a) confirmed cases and fatalities in the USA; (b) fatalities in the USA.Fatalities during COVID-19 in Latin America until 2 June 2020 [18]. First fatality in Brazil on 18 March, Peru on 20 March, Ecuador on 14 March, and Mexico on 24 March 2020.Confirmed COVID-19 cases and fatalities in the UAE until 2 June 2020 [18]. First confirmed case on 27 January 2020, first fatality on 22 March 2020 in UAE. From left to right: (a) confirmed cases in the UAE; (b) fatalities in the UAE.Safety recommendations and measurements during the COVID-19 pandemic implemented by Ajman University (AU), Ajman, UAE. (Reproduced with agreement of the AU Department of Environmental Health and Safety (EHS) on 11 May 2020.).Confirmed COVID-19 fatalities in India until 2 June 2020 [18]. First fatality on 13 March 2020 in India.Countries, their markers, some public health measures, and outcomes during the COVID-19 pandemic.limited entry of customers to 30 percent of its capacitymaintained distance of at least two meters between customersno crowding allowedobserve physical/social distancingwear face masks and gloves outdoorsstay at homeFollow the guidelines on family visitsfollow medical advice issued by relevant authoritiesperform prayers at homedo not enter shopping malls and outlets if above 60 years or below 12 years

Med-MDPI/ijerph_5/ijerph-17-15-05617.txt

@@ -0,0 +1 @@
+Regular Nordic walking (NW) improves physical fitness, including the ability to maintain balance, in older adults. However, little is known about whether complementing the exercise programme with cognitive training (CT) contributes to increased effects. The aim of the study was to determine and compare the effect of regular NW and NW combined with CT on the ability to maintain static balance in older adults. The study examined 61 women aged 64 to 93 years living in adult day care centres. Twenty people participated in a three-month programme combining NW and CT (group NW + CT), 20 people participated only in NW classes (group NW), and 21 people were a control group (group C). The Romberg balance test, Fullerton Functional Fitness Test, and Attention and Perceptivity Test were used. After the programme, an increase in the time of maintaining the balance (with eyes open on the left and right legs) was observed in groups NW + CT and NW, with no such changes found in group C. This increase was greater in group NW + CT. Increased agility and strength of the hand were predictors of improving the ability to maintain balance. Regular NW improved the ability to maintain balance with eyes open in female residents of adult day care centres.During the ageing process, many unfavourable changes occur in the human body. Biological changes are initially imperceptible, but they lead to a gradual deterioration of all functions, including the locomotor system, internal organs, and in the psychological and social spheres [1]. The systems responsible for maintaining balance and cognitive function are weakened, and the body at this stage of ontogenesis is characterized by lower compensation and recovery capabilities [2]. Older adults experience a deterioration in all motor skills [3]. One of the major symptoms of ageing is the progressive loss of muscle weight and strength, which leads to numerous changes in body function, primarily the deterioration of physical fitness. Sarcopenia is particularly dangerous for older adults, leading to disability and mortality in this group of people [4,5]. In most older people, sarcopenia has a multifactorial background. It is a consequence of many different factors, including disturbed skeletal muscle homeostasis and, indirectly, the entire body. In the ageing body, there is then a relative advantage of catabolic over anabolic processes [6].With age, the number of motor units gradually decreases, consequently reducing muscle function. Between 70 and 80 years of age, the number of motor units in the tibialis anterior muscle is about 50% of the value observed in young people [7]. Lower limbs are more likely to lose motor units than upper limbs, which are more resistant to progressive ageing processes [8]. Between 30 and 40 years of age, humans reach the highest level of maximum strength. As a result of involutional changes, it drops to a level of about 60–70% of the maximum values at the age of 70 [9]. Between 30 and 80 years of age, the strength of isometric contraction also decreases by about 50% [10]. The gradual decrease in muscle mass is accompanied by a decrease in muscle flexibility and, consequently, a decline in muscle efficiency. Muscle power decreases even more than muscle strength, which results from a progressive decrease in the speed of shortening of muscles with age [11]. Consequently, reaction time becomes longer, as the rate of nerve impulse conduction is reduced. This leads to limitations in walking and maintaining a stable body posture. These changes cause disorders in neuromuscular coordination and reduced ability to perform fast and rhythmic movements [12]. As a result of ageing of the locomotor system, thoracic kyphosis is increased, whereas lumbar lordosis is decreased. As a consequence of these changes, the centre of gravity moves forward with simultaneous limitation in the mobility of the limbs and spine [13].Involutional changes in individual systems lead to substantial difficulties during walking. With larger body sway during walking and reduced step length and formation of a larger support surface (wider steps), walking becomes less and less economic, leading to greater fatigue, turning into an “old age gait” [14]. This slower gait is strongly connected with the weakening of cognitive function, and at the same time, it represents a predictor of cognitive dysfunction [15]. Deterioration of the ability to maintain balance, which is one of the coordination capacities, reduces movement control and consequently increases the risk of falling. Balance disorders and the related falls are associated with injury. They can cause fear of movement and lead to loss of motor independence [16].As the number of older adults is increasing every year, it is advisable to look for factors influencing the best possible psychophysical status of the person in the longest possible time [17]. It is becoming necessary to develop and implement special prevention programmes aimed at maintaining and even improving the fitness of the elderly, especially in terms of balance and muscle strength. The results of Nordic walking (NW) classes appear to be promising. The female participants of the Third Age University programme who attended regular NW classes for three months improved their results in functional fitness tests, especially in tests that measured physical capacity and strength [18]. Ossowski and Kortas [19], after completion of a six-month NW programme, documented improvements in balance and agility of older adults. In the study by Piotrowska and her team [20], older people participating in a 12-week programme of NW exercises and cognitive training (CT), improved their performance both in terms of balance and functional/cognitive fitness.The results of previous research indicate that regular exercise, such as NW, improves physical fitness, including the ability to maintain balance, in older adults. However, little is known about whether complementing the exercise programme with CT contributes to increased effects. There are more and more reports in the world literature that emphasize the role of cognitive functions as components of motor control. Studies confirmed that the speed of walking after adding a cognitive task correlates with the efficiency of executive functions [21]. The research confirmed that the walking speed of the elderly, which is a manifestation of the efficiency of executive functions, significantly decreased after adding the cognitive task.In the ageing process, executive functions are weakened, which affects the functional efficiency of seniors. Additionally, weakening of the muscular system function translates into impairment of body posture control function and an increased risk of falls [22,23]. Cognitive processes play an important role in the control of motor functions, and their improvement may cause an increase in the functional efficiency of seniors [24]. The combination of NW and CT was aimed at checking whether this combined intervention would have a greater impact on the ability of the elderly to maintain balance.The question also arises as to what the changes in balance depend on and which factors make it possible to anticipate them. The aim of the study was to determine and compare the effect of regular NW exercises and NW exercises combined with CT on the ability to maintain balance in older women. We assumed that the effects in group NW + CT would be greater than in group NW. Relationships between changes in ability to maintain balance and changes in physical and perceptual fitness were also examined.The examinations were performed using the experimental method in natural conditions. The participants were residents of three elderly day care centres in Warsaw. Three groups were formed: Two experimental groups—the group that participated in NW sessions and CT (group NW + CT) and the group participating only in NW sessions (group NW)—and the control group without interventions (group C). Each group included residents from different day care centres. Purposive sampling with “threes” was employed to ensure evenness of the groups while adopting the criterion of general health status and physical and cognitive fitness.NW exercises were aimed at improving aerobic capacity, joint stabilization, flexibility, increasing muscle strength and improving static and dynamic balance and motor coordination (improvement of functional fitness). The programme took three months, with training sessions held twice a week for 60 min. Classes included a 10-min warm up, exercises to improve muscle strength (15 min) and breathing and flexibility exercises, with a duration of 5 min. During the classes, participants walked for 30 min and covered an average distance of 2.5 km (min. 1.5 km, max. 3.2 km). During the first 5 min (5–8 weeks) and 10 min (9–12 weeks) of the walk, the participants from the groups NW and NW + CT performed additional tasks, such as solving arithmetic problems (e.g., subtraction every 3 from 197). During these exercises, the walking speed was definitely decreased by 2 km/h on average.During each class, attention was paid to maintain the correct technique of walking with poles, particularly the pole planting and push phase. The participants from both experimental groups NW + CT and NW covered a greater number of kilometres each month. The route was measured using dedicated equipment with an in-built GPS POLAR m400 module operated by a coach who supervised training.The author’s programme of exercises of selected cognitive functions was used in the CT in accordance with the multicomponent Baddeley’s model of working memory [25]. It included exercises aimed to improve coordinated functions of attention and working memory, training using the visuospatial sketchpad and exercises of management functions. The training was complemented by increasing the working memory capacity using the mnemotechnic chain method (associating words in specific order) and symbol method (consisting of associating words with a number having its equivalent in the form of a symbol; this method allows the person to memorize and then to retrieve the elements in any order). The programme took three months, with training sessions held twice a week for 60 min.The group without interventions participated in typical classes offered to the residents of the day care centres.A total of 88 residents aged over 60 years participated in the experiment. The inclusion criterion was at least average (for the appropriate age group) level of health, physical fitness and basic cognitive functions. The data on health condition of the respondents was obtained from the physician caring for residents. The participants were elderly people without some kinds of serious health problems, namely cardiovascular and locomotor system diseases, and neurological diseases. The data on physical fitness was obtained from caregivers in day care centres. Furthermore, the participant had to be able to walk 1.5 km without pain or breathing problems. Additionally, we applied the Katz scale (ADL—Activities of Daily Living) and the Lawton scale (IADL—Instrumental Activities of Daily Living) to determine whether the subjects could perform their daily activities independently, safely and without too much effort. Information on the cognitive possibilities was provided by a psychologist or a geriatrician. All the subjects were independent in terms of their motor skills and performed their daily activities independently. Furthermore, we used the Minimental Test—Mini Mental State Examination (MMSE). The participants had to achieve the maximum result in this test to be qualified to our study.People suffering from chronic diseases of the cardiovascular system, locomotor system and neurological diseases that could affect the results of the balance and physical fitness tests, and people with limited cognitive abilities (dementia) were excluded from the study.The laterality of the limbs was assessed by checking which hand the subject used while writing and asking her/him to kick a small ball. People with homogeneous right laterality participated in the study.An additional criterion of inclusion into analyses in the case of the experimental groups was participation in at least 85% of sessions. The final analyses involved 68 people who participated in all measurements of physical parameters and physical and cognitive fitness. Due to an insignificant number of men (n = 7) meeting the inclusion criteria, the study presents the data concerning only 61 women.The age of participants ranged from 64 to 93 years (M = 80.25; SD = 5.755). They were mostly widows (n = 54) and women living alone (n = 51). Women included in the study most often had secondary education (n = 47) and vocational education (n = 9). Four participants had primary education, and one participant had higher education. All the participants had been employed in the past and had performed physical work (n = 40) more often than mental (n = 11) or mixed (n = 10) work. The study participants evaluated their financial status mostly as average (n = 41) and less often as good (n = 11) and poor (n = 7). The majority of the women assessed their former lifestyles as medium active (n = 35) and less often as active (n = 17) and very active (n = 8); 32 participants were very rarely involved in physical activity, and 22 of them did this at any opportunity. At the time of the study, the respondents travelled mainly by public transportation (n = 46) and less frequently by car (n = 9) or on foot (n = 5). Furthermore, 37 respondents declared that, in the past, they did not pay attention to healthy diets, whereas the others answered in the affirmative.The groups NW + CT and NW had 20 people each, whereas group C consisted of 21 participants. The groups did not differ significantly in terms of age (F = 0.528; p = 0.592), education (chi squared = 4.723; p = 0.580), marital status (chi squared = 7.036; p = 0.318), living alone/with family (chi squared = 2.766; p = 0.598) and economic status (chi squared = 9.669; p = 0.139). The groups were evenly matched in terms of baseline physical fitness and the ability to maintain balance.Data on age, marital status, education, professional activity, financial status and physical activity in the past and currently were obtained by means of an author’s survey that was used in previous studies of the elderly [18,20].In order to evaluate the ability to maintain balance, the Romberg balance test was performed with hands along the body or crossed on the chest on the left and right leg with eyes open and closed [26].The Fullerton Fitness Test was performed to evaluate physical fitness. The test is composed of six consecutive trials that evaluate [27]:Arm Curl. Evaluation of upper body strength.30-s Chair Stand. Evaluation of lower body strength.Back Scratch Test. Examination of upper body mobility.Chair Sit and Reach Test. The aim of the test is to assess the flexibility of the lower body (especially the popliteus tendon).8-Foot Up-and-Go and 6-Min Walk tests. Evaluation of agility (dynamic balance) and long-distance aerobic endurance.2-Min Step in Place Test. Evaluation of aerobic endurance is performed if the 6-Min Walk test is impossible.Arm Curl. Evaluation of upper body strength.30-s Chair Stand. Evaluation of lower body strength.Back Scratch Test. Examination of upper body mobility.Chair Sit and Reach Test. The aim of the test is to assess the flexibility of the lower body (especially the popliteus tendon).8-Foot Up-and-Go and 6-Min Walk tests. Evaluation of agility (dynamic balance) and long-distance aerobic endurance.2-Min Step in Place Test. Evaluation of aerobic endurance is performed if the 6-Min Walk test is impossible.Furthermore, the hand grip test was performed using a dynamometer. Each of the elderly participants, after hearing the instructions, squeezed the dynamometer bar twice with both the right and left hands in a sitting position. The result of the second test was adopted as an indicator of hand strength.In order to determine the perceptual efficiency, the Attention and Perceptivity Test was used [28], which consists of crossing the indicated digits and provides indices of perceptual work rate (number of characters analysed), perception fallibility (number of mistakes) and attention fallibility (number of skipping instances). Standard conditions were used when performing the test (duration of 3 min). The tests were conducted in small groups of several people. Before the test began, it was verified whether each participant understood the instructions correctly.All measurements were carried out twice: Before and after the NW programme.The study was conducted according to the ethical guidelines and principles of the Declaration of Helsinki. All subjects gave their informed written consent to the experimental procedures, which were approved by the Senate Ethics Committee of Scientific Research AWF Warsaw (SKE 01-46/2016).Statistical analyses were performed using IBM SPSS version 25. (Armonk, NY, USA) In order to determine the normality of variable distributions, the Kolmogorov–Smirnov test was used. Repeated measures analysis of variance (group x measurement: 3 × 2) was used to determine intergroup differences and changes in time. One-way analysis of variance (ANOVA) and Tukey’s post-hoc tests were used to determine the differences between the groups. The study also examined the relationships of changes in the ability to maintain body balance with age and changes in functional and cognitive fitness. In order to evaluate these relationships, the first step was to compute the indices of changes in functional and cognitive fitness (perceptual work rate and perception and attention fallibility), then Pearson’s r linear correlation coefficients were calculated for the entire study group. The stepwise regression analysis was used to determine which indices of change made it possible to anticipate an improvement in the ability to maintain balance with eyes open and closed.Repeated measures ANOVA was used to determine intergroup differences and changes in time. The respective data are presented in Table 1 and Table 2. The main effect of the group was not significant in any case. The main effect of the measurement was significant for the results of the tests with eyes open, both for standing on the right and left legs. The results of the second measurement were better compared to the first. The interaction effect was significant in three cases (except for the test with eyes closed on the right leg). In the test with eyes open on the left and right legs, the improvement of results was found only in the experimental groups (groups NW and NW + CT). No significant intergroup differences were found for these tests (in the first or second measurement) (Table 1). In the test with eyes closed, the post-hoc tests indicated no intergroup differences in the first and second measurements and no differences between measurements in all groups (Table 2).The differences in the ability to maintain body balance were also computed by subtracting the first measurement from the result of the second measurement, where a positive value means an increase in the time of maintaining body balance. One-way ANOVA and Tukey’s post-hoc tests were used to determine the significance of differences between the groups (Table 3).The groups differed significantly in terms of changes in the ability to maintain body balance with eyes open and closed standing on the left leg. In group C, all change indices were negative, which indicates a reduction in execution time. In group NW, the difference was positive in the tests with eyes open, whereas during the test with eyes closed, this value was close to 0 for the left leg and negative in the test of standing on the right leg. In group NW + CT, the difference was positive in all tests. Improvement indices in tests with eyes open were the highest in group NW + CT and lowest in group C. In the test of standing on the left leg with eyes closed, the improvement was significantly greater in group NW + CT than in group C. No significant intergroup differences were found in the test of standing on the right leg with eyes closed (Table 3).The study also examined the relationships for changes in ability to maintain body balance with age and changes in physical and cognitive fitness. In order to evaluate these relationships, the first step was to compute the differences in physical and cognitive fitness (perceptual work rate and perception and attention fallibility), then Pearson’s r linear correlation coefficients were calculated for the entire study group.The improvement in the ability to maintain balance did not correlate significantly with the age of respondents. The correlation between the improvement in the test of balance on the right leg with eyes closed reached a trend level (r = 0.236; p = 0.067) and had a positive value (which may be surprising). Numerous positive relationships were also found with indices of improvement in physical and perceptual fitness (Table 4). Interestingly, most of them were found for the improvement in the ability to maintain balance on the left leg with eyes open. It was directly proportional to almost all indices of improvement of physical fitness (except for the improvement in mobility of the upper body in the left-hand test). Slightly fewer relationships were found in the case of improving the ability to maintain balance on the right leg with eyes open. Both indices correlated positively with all indices of improvement in perceptual efficiency.The improvement in balance on the left leg with eyes closed correlated significantly with the increase in the strength of the lower body, the strength of both hands and the increase in aerobic endurance. Correlations with the increase in agility and perception speed reached only a trend level. Interestingly, no significant correlations were found for the index of improvement in the ability to maintain balance on the right leg with eyes closed.The next step was to determine which indices of change made it possible to anticipate an improvement in the ability to maintain balance with eyes open and closed. The stepwise regression analysis was used for this purpose. The factors (explanatory variables) were changes in physical parameters and in physical and cognitive fitness, whereas the dependent (explained) variables were total indices of improvement in the ability to maintain balance with eyes open and closed. These indices were the sum of the indices of improvement in the ability to maintain balance in two trials (on the right and left leg). The results of the analyses are shown in Table 5. The predictors of the improvement in balance with eyes open were the increase in agility and increase in the strength of the right hand, which, in total, allowed for prediction of the improvement in the ability to maintain balance with eyes open in over 40%. The increase in left hand strength allowed for prediction of an improvement in balance with eyes closed only in 7% of participants.With the introduction of an additional factor (group affiliation), this was the only predictor of improvement in the ability to maintain balance with eyes closed (R2 = 0.478; F = 55.892; p < 0.001; beta = 0.697). In this case, none of the factors allowed for prediction of the improvement in the ability to maintain balance with closed eyes.Functional efficiency deteriorates with age. This is particularly important for people who do not engage in any physical activity for various reasons. Balance disturbances and problems with proper gait may result in serious injury and lead to fear of falling [29]. The results of our study indicate that systematic NW exercises improve the static balance of older women. We can expect that those participating in NW training will be at a lower risk of falling. This will reduce the fear of independent movement and risk of losing independence in everyday life. The World Health Organization (WHO) recommends that older people take up physical activity on a regular basis and, if possible, every day. NW is a safe form of physical activity for every human being, since it is based on cyclic and simple movements.However, the improvement of static balance took place in both experimental groups only in the test with the eyes open. A study by Melzer et al. [30] found that a much higher concentration is needed in older people, compared to young people, to perform balance tests if memory tasks are additionally performed. The response to body sway in order to maintain balance is limited and requires reorganisation in the system of sensorimotor control. During equilibrium tests with eyes open and closed, older people stiffen the muscles of the lower limbs. The contribution of visual control in maintaining static balance is smaller than in the case of young people. The most important element of the sensory system that determines postural stability is the function of sensory receptors [31]. The NW exercises activate both the vision analyser and proprioception. In the absence of vision, complete concentration on the proprioceptive information is observed. This may be the cause of no improvements in the results of the Romberg tests with eyes closed.Improvements in the ability to maintain balance with eyes open were found in both experimental groups. The subjects from group NW participated only in regular NW classes, performing simple cognitive exercises during the walk. People from group NW + CT performed additional memory training. However, they did not differ significantly in terms of the results of the balance test in the first and second measurements. The only significant difference was in the ability to maintain balance on the right leg with eyes open. A greater improvement in results was observed in group NW + CT. However, the groups differed significantly in terms of the indices of changes in the ability to maintain body balance with eyes opened. These results suggest that combining NW with simple exercises involving cognitive functions yields results only slightly worse than the programme composed of NW and cognitive (memory) training. Our hypothesis was therefore only partially confirmed.The degree of cognitive involvement in NW depends on additional stimuli occurring during the walk (e.g., on uneven and unstable ground). It requires performing sudden movements or re-establishing previously learned movement patterns, which may cause problems for older adults due to lower cognitive skills [32]. In our experiment, the study participants from groups NW and NW + CT performed simple exercises involving cognitive functions. Westlake et al. [33] found that performing a sufficiently difficult additional task often leads to a reduction in problems with balance. Focusing on a secondary task leads to increased stability through a hidden learning effect. It was also observed that during the performance of the so-called double task, the walking speed increased, whereas the risk of falls decreased [34]. This effect was also observed in the women analysed in our study. Therefore, it seems necessary to use not only physical exercises in programmes of activation for older adults but also to improve their cognitive functions.Grip strength was a predictor of the improvement in balance, both with the eyes open and closed. Garcia et al. [35] demonstrated a relationship between the grip force and walking speed. Beseler et al. [36] studied the relationship between the strength of lower limb muscles, especially those responsible for correct gait, and grip strength in older adults requiring hospitalization. They found a relationship between grip strength and mobility resulting from the strength of lower limb muscles. It was also shown that grip strength is related to cognitive disorders.In the study, older people from groups NW and NW + CT achieved improved results in the “8-Foot Up-and-Go” test (agility). It was a predictor of the improvement in maintaining balance with eyes open. Getting up from a chair depends on the level of strength of the lower limb muscles and the ability to maintain balance [37]. Repeating this activity increases the strength of the lower limb muscles and contributes to an increase in walking speed [38]. Elderly people had the greatest problems in maintaining balance during the shift from the sitting to standing position [39]. Therefore, there are mutual relationships between the ability to maintain balance and agility, with both abilities determining gait speed, which also increased in our study participants.During the 12-week NW programme, the participants in our experiment improved their walking by the fact that they covered more and more kilometres during their 30-min sessions. It is very important to use the correct walking technique, especially the grip technique, planting the poles in the ground and the push phase. This increases the involvement of the upper body and, consequently, the walking speed [40]. Proper use of poles helps shift the centre of gravity backward, thus allowing the person to restore the correct gait pattern and adopt the correct body posture. Such effects have been documented in studies of healthy adults [41,42,43]. It was shown that even a six-week NW training programme increased walking speed and step length in older people [44,45]. After five months of NW training, a greater range of mobility was observed in plantar- and dorsiflexion of the foot, which is of great importance for correct gait. Furthermore, the women studied had a greater range of mobility in their hip joints, a greater level of muscle strength in the elbow joints and improved dynamic balance [46].The presented research confirms the need for developing and implementing special programmes aimed at older people. Ostrowska et al. [47], after analysis of the gait of older adults, recommended that the programme for such people should include balance training and exercises improving mobility in the hip and knee joints and strengthening the erector spinae muscles, which increases joint mobility. Maintaining an adequate level of physical activity in older adults is important not only for their physical fitness but also for the associated cognitive functions related to reception and selection of information coming from the environment. Programming physical activity for older adults requires a comprehensive and individual approach depending on the level of physical and cognitive performance. Regular NW classes and CT can offer an excellent tool for the development of physical fitness and cognitive function in older adults. The exercise programme should be clearly defined in terms of intensity and frequency in order to improve adaptability to changes that occur with age. The attractiveness of the programme is also important for motivating older people. The NW programme adjusted for older adults and supervised by coaches can be an effective strategy to improve functional fitness and, consequently, to delay the period of disability. Motivating older adults for undertaking physical activity is particularly important in day care centres that help people with lower physical and cognitive fitness. In future studies, authors should determine and compare the effect of regular NW combined with CT on the ability to maintain static and also dynamic balance in older adults with some physical or/and mental diseases.The results of our study are not free of limitations. The first limitation is related to the experimental design used in our study. For organizational reasons, it was possible to create only three study groups. A full experimental model would require the creation of an additional group participating only in CT. This would better distinguish between the effects of NW training and CT.The second limitation results from the choice of the respondents. The groups were recruited from residents of different day care centres. This solution was chosen to avoid a situation where only some residents would participate in additional activities. We attempted to limit the effect of this factor by making the recruitment in “threes.” However, it cannot be excluded that differences in the social environment had, at least in part, an impact on the results obtained.With the insignificant percentage of men in the centres, we analysed only the results obtained for women, and the relationships found relate only to this sex. The possibility of generalization of the results is additionally limited by the fact that our participants were quite healthy and physically and cognitively fit for their age. Moreover, they lived in the capital city. This group is not representative for the entire population of older adults in Poland. In the capital city, more opportunities are offered for older adults spending their leisure time actively compared to rural areas.The authors are aware that the number of subjects did not provide adequate testing power. The initial calculations at the research planning stage assumed larger differences and slightly less variance of results. It should be noted, however, that the study was conducted in a specific population and the intervention lasted a relatively long time. Gathering 60 people and conducting research as intended was a challenge. Every effort was made to ensure that the study was conducted accurately. However, the small number of studied groups remains a limitation.Another limitation concerns the CT and cognitive function indices used in the study. Further research is needed to compare the effects of training focused on different cognitive functions. It is necessary to use tools that allow for their reliable and accurate measurement.Implementation of specialized programmes with the use of NW poles for Social Welfare Homes can be an effective strategy to maintain and improve the ability to maintain balance with eyes open in elderly people with at least average health condition as well as physical and cognitive fitness. It seems necessary to introduce an additional element, CT, to the programmes for healthy and fit elderly people. The inclusion of elements of CT increased, although slightly, benefits in the ability to maintain balance with eyes open, especially when standing on the right leg. In elderly people who did not participate in motor activities, there were no changes in balance in short-term observations.Conceptualization: J.P.; Methodology: J.P.; Software: M.G.; Validation: M.G.; Formal analysis: M.G.; Investigation: A.L., I.R.; Resources: A.L., I.R.; Data curation: M.G.; Writing-Review and editing: A.L., I.R.; Visualization: A.L., I.R.; Funding acquisition: M.G. All authors have read and agreed to the published version of the manuscript.The results presented in the study were obtained within the statutory project DS.256. The impact of Nordic Walking and mental exercises on the level of physical and cognitive fitness of elderly people” financed by the Ministry of Science and Higher Education.The authors declare no conflict of interest.The results of the Romberg test for measuring the ability to maintain balance (comparison by group and measurement).1—measured in seconds; the longer the time, the better the ability to maintain balance; M *—weighted averages; ns—not statistically significant.The results of the Romberg test for measuring the ability to maintain balance in subgroups (Tukey’s post-hoc test).1—measured in seconds; the longer the time, the better the ability to maintain balance; *—Tukey’s post-hoc test; ns—not statistically significant.Changes in the results of the Romberg test for measuring the ability to maintain balance in subgroups (ANOVA).1—measured in seconds; the longer the time, the better the ability to maintain balance; ns—not statistically significant.Correlations between the improvement in the ability to maintain balance and improvement in physical and cognitive fitness (Pearson’s r coefficients).1 6-Min Walk (m); 2 Chair Sit and Reach Test (cm); 3 Back Scratch Test (cm); 4 30-s Chair Stand (the number of stands); 5 8-Foot Up and Go (sec); 6 Arm Curl (the number of lifts); 7 number of characters analysed; 8 number of mistakes; ** p < 0.001; * p < 0.05; t p < 0.1.Predictors of improving the ability to maintain balance with eyes open and closed (stepwise regression analysis).1 (sec), a positive value indicates an improvement in the result; 2 8-Foot Up and Go (sec), a positive value indicates an improvement in the result; 3 dynamometer; a positive value indicates an improvement in the result.

Med-MDPI/ijerph_5/ijerph-17-15-05618.txt

@@ -0,0 +1 @@
+The use of advanced learning technologies in a learning management system (LMS) can greatly assist learning processes, especially when used in university environments, as they promote the development of Self-Regulated learning, which increases academic performance and student satisfaction towards personal learning. One of the most innovative resources that an LMS may have is an Intelligent Personal Assistant (IPA). We worked with a sample of 109 third-grade students following Health Sciences degrees. The aims were: (1) to verify whether there will be significant differences in student access to the LMS, depending on use versus non-use of an IPA. (2) To verify whether there will be significant differences in student learning outcomes depending on use versus non-use of an IPA. (3) To verify whether there will be significant differences for student satisfaction with teaching during the COVID-19 pandemic, depending on use versus non-use of an IPA. (4) To analyze student perceptions of the usefulness of an IPA in the LMS. We found greater functionality in access to the LMS and satisfaction with teaching, especially during the health crisis, in the group of students who had used an IPA. However, both the expansion of available information and the usability of the features embedded in an IPA are still challenging issues.The use of advanced learning technologies can be an excellent teaching aid for efficient learning processes, especially when adapted to Self-Regulated learning (SRL). The learner can use various technologies to interpret how to approach the resolution of learning tasks and, according to the needs that are detected, the use of a particular learning technology will help to guide the learner towards successful outcomes [1]. Advanced learning technologies are frequently applied within a Learning Management System (LMS). An LMS has many advantages, among which we can highlight personalized attention to the student [2], which makes it possible to increase motivation [3]. Likewise, an LMS can facilitate individual and group work [4], and the use of different resources such as virtual laboratories, computer graphics, flipped learning, and flipped classroom experiences, virtual workshops, and messaging, among others [5]. The use of advanced learning technologies can also facilitate collaborative work within an LMS, such as the application of the Project-Based learning (PBL) methodology [6,7]. Therefore, it can be used for the analysis of multimodal and multichannel data on SRL provided by LMS environments, in which different resources such as smart tutoring, serious games, multimedia resources, augmented reality resources, and virtual reality are applied. In addition, LMS with additional technologies can be used to record information related to eye tracking, physiological records, facial expressions, and speech analysis, among others [8,9,10,11]. Later, these records can be analyzed with statistical and data-mining techniques, through which the learning path of an individual student or a group of students may be consulted during the resolution of different tasks [12]. In addition, the use of LMS with additional technologies can facilitate the use of SRL in almost real time [8]. The collaborative methodology implemented through LMS with additional technologies will guide student learning processes and provide oriented feedback to processes [13]. This methodology is useful through automated and individualized resources, so that the help each student may need is available at any time [14]. Nevertheless, learning autonomy with an LMS is a related disadvantage, due to the need for continuous supervision of the LMS by the teacher. However, improvements to the usability of LMSs have been advanced in recent research by the introduction of intelligent multi-agents, currently found in many automated chat systems. Based on natural voice assistance, these systems can perform many internal and external actions based on the user search queries. The results show that the proposed system can have a positive impact on both students’ perceptions of the usability of an LMS, and student performance [15].The use of an Intelligent Personal Assistant (IPA) to improve learning is an emerging practice that, although not yet widespread, has an important future role. The implementation of IPAs through Voice User Interfaces (VUIs) (see Barcelos et al. [16] for an analysis of the taxonomy of voice assistants) means that these assistants can give immediate and intuitive responses to natural language stimuli, so that the user can develop voice interaction through the computer system. In addition, many of them include the possibility of creating applications at no cost for their development and use, such as the Amazon Echo or the Google Home assistants. The system begins from a stimulus (voice) and gives an answer to the query from the user. The characteristics of these IPAs are their functionality, immediate availability or inductiveness, and the empathy they generate with the user, and some are compatible with the Chatbot format [17]. Recent studies have indicated that IPAs can increase their efficiency, if they include the figure of an avatar as an agent of conversational interaction [12]. In educational contexts, IPAs are incorporated in LMSs, such as Moodle (Modular Object-Oriented Dynamic Learning Environment), specifically for the support of learning among students with some type of educational need [18], such as the visually impaired. The functionality offered by IPAs includes guidance for navigation on the platform or on the web [15], guidance for both reading and writing texts [19], and providing feedback on the tests results, for example, quizzes [18]. IPAs are also incorporated in Moodle modules, one example of which is the “Lesson”. Bearing IPA architecture in mind, it can be a tool to build scripts and learning scenarios [17]. This new IPA functionality is potentially effective in virtual labs and simulated environments, as well as when completing quizzes [20]. The incorporation in Moodle of a module called “voicerec” [21] also recently commenced, although this technology is still in an initial state of development and presents implementation difficulties. Its advantages for the user are that it favors coaching and helps the student to find and to access information that has been tested, filtered, and prepared by the teacher. It can also be used at all times, which favors the personalization of learning and, at the same time, promotes collaborative work (teacher-student, student-teacher, student-student, student-materials, etc.). All of these aspects increase student motivation for learning [22].Regarding the studies on the usability assessment of IPAs, users have indicated that interfaces must be adapted to the needs of each task [23,24]. IPAs that include holograms are under evaluation as learning aids [25,26]. In summary, the type of IPA and its objectives vary and although they are all implemented using natural language, the technology underlying each one is different [27]. In short, IPAs are increasingly finding their way into educational and health-related environments [28,29]. Their advantages are that they encourage personalization in learning [30,31] and in therapeutic intervention [15]. In addition, they can provide insight into patterns of interaction, on which basis students can be provided with personalized interventions [32,33]. Even so, this technology is very complex and is still in an initial state of development [28]. Research studies therefore have emphasized the need for extensive research in this area [34,35].As previously noted, current IPA technology incorporates Machine Learning techniques (deep learning and reinforcement learning) resources based on voice-recognition systems [36]. IPAs provide users with information on coursework, facilitating its planning [37,38]. Specifically, the recent use of this technology in university-learning contexts has been associated with very good results and levels of acceptance, specifically among students with special educational needs (visual, auditory, memory, etc.) [39,40]. Furthermore, computer security resources are also incorporated in IPAs, as users must log in before implementing them [41]. Their inclusion in blended learning university learning environments is also beginning to find acceptance, increasing their functional applications [42,43]. IPAs also generate high levels of student satisfaction, as students can access teaching at the most convenient time and place and can receive personalized feedback [44]. The use of this technology also provides a further channel for teachers and academic leaders to connect with students and to understand their main concerns [45]. IPAs can likewise be used to provide students with information on administrative aspects [45,46] and they are very useful for students with visual [47] and auditory needs [48]. These users particularly value the versatility of access to information searches [49]. However, recent studies have also indicated that each IPA needs to be adapted, in terms of both interface and functionality, to respond to the needs of each user [50,51,52]. This field of study still has a long road to travel down, as users currently value the effectiveness of IPAs at only 60% [53].In conclusion, the world is increasingly turning digital, which implies an urgent need for a series of changes to teaching methods for the inclusion of learning tools within higher education. Experts are calling for an intelligent university in which technology and pedagogy are implemented in teaching–learning environments [54,55]. These environments may be blended learning, or virtual, yet they are quite unlikely ever to be purely face-to-face again. In particular, the global pandemic caused by COVID-19 has quite suddenly underlined the value of telematic teaching tools, prompting governments and university leaders to urge both teachers and students to make good use of these technologies. Research must therefore be conducted to determine the effectiveness of these different resources such as IPAs in blended learning and e-learning spaces [5,56,57].Based on the research noted above, the research questions in this study are as follows: (RQ1) to verify whether there will be significant differences in student access to the LMS, depending on use versus non-use of an IPA; (RQ2) to verify whether there will be significant differences in student learning outcomes depending on use versus non-use of an IPA; (RQ3) to verify whether there will be significant differences for student satisfaction with teaching during the COVID-19 pandemic, depending on use versus non-use of an IPA; (RQ4) to contrast students’ perceptions of the usefulness of an LMS that incorporates an IPA.The convenience sampling process concluded with a sample of 109 third-grade students in Health Sciences degrees: 61 in Group 1 and 48 in Group 2. The sample included all students studying on the third year of a Health Sciences degree at the University of Burgos. In Table 1, the statistics on the two variables, age and sex, can be consulted in Table 1.(a) The Scale of learning strategies (ACRAr) by Román and Poggioli [58]: a widely tested instrument in investigations on learning strategies. It is used to identify 32 strategies at different times of processing information: acquisition information (α = 0.78); encoding information (α = 0.92); recovery information (α = 0.83); and metacognition strategies (α = 0.90). In this study, only the metacognitive strategies scale was used. The indicators of scale validity for the sample were metacognition strategies α = 0.90. ACRAr has been widely validated among secondary education and university students [59].(b) Alexa’s Computer application. “UBU(Universidad de Burgos) VoiceAssistant”: a specific application was developed for students to consult the key dates on the course (delivery of practices, completion of questionnaires, delivery of project, etc.) through a (mobile or computer) device. This application has a client-server within the Alexa service system integrated in the Amazon Web Service (AWS). An example of the interface and operation can be seen in Figure 1 and Figure 2, respectively. Students have first to accredit their identity to enable use of the “UBUVoiceAssistant” computer application. This process is achieved with the use of UBUVirtual, the learning platform (LMS) of the University of Burgos. The students must provide valid credentials in the accreditation of their identity to access the platform. After the successful validation of these credentials, the student is then allowed further access to the “UBUVoiceAssistant” Computer application. The connection is therefore secure, and the protection of personal data is guaranteed [60].(c) Scale of assessment of the development of the subject. The ad hoc development of the scale yielded 18 closed response questions, measured on a 5-point Likert-type scale, and 8 open-ended questions (4 of which refer to the development of teaching during the COVID-19 health crisis) [20]. The total reliability of the scale was α = 0.95 and, for each item, the values were within an interval of α = 0.94–α = 0.96. The scale can be found in the Supplementary Material, Table S1.(d) Questionnaire for assessing the functionality of the IPA “UBUVoiceAssistant”. This instrument consists of two closed-ended questions: a multiple-choice question (with 5 options) a no/yes question, and three open-ended questions. As it is fundamentally a qualitative opinion survey, no reliability analysis could be performed. The questionnaire can be consulted in the Supplementary Material, Table S2.(e) Learning Management System “UbuVirtual” based on Moodle 3.7: UBUVirtual was used in Moodle version 3.7 with a platform design based on a constructivist development designed for personalized learning and collaborative work on the platform [61,62,63].(f) eOrientation plugin: a Moodle plugin, now registered under patent No. BU-09-20, was funded through research project No. BU032G19 awarded for research, in 2019, by the Junta de Castilla y León [64]. The plugin is compatible with Moodle log analysis of student and teacher access to the platform, and interaction with it through the various available activities and resources [65]. This Moodle plugin and its associated graphics can be used to follow the progress of students, for more information see the research of Sáiz-Manzanares, Marticorena-Sánchez, and García-Osorio [65].(g) Pedagogical Model: in both groups, the same pedagogical model was used. The pedagogical model includes the following elements: development and defense of PBL, quiz-type questionnaires, and co-evaluation activities in evaluation processes throughout the teaching–learning process and flipped learning experiences. The effectiveness of this pedagogical model has been tested in various investigative studies [5,6,25,35,56,61,63,65].Before the study commenced, the authorization of the Bioethics Committee of the University of Burgos and the informed consent of all participants were obtained in writing (see point 2.5). The subject was designed with a blended learning methodology using flipped classroom experiences, which meant that teaching, although delivered in person, was through a Moodle-based LMS (UBUVirtual: learning platform of the University of Burgos), which contained hypermedia resources (videos in flipped classroom experiences and computer graphics). The pedagogical design of the subjects included the following elements: practices (weighted 20% of final grade), quizzes (weighted 30% of final grade), project work, and a defense of a project using practical assumptions drawn from PBL methodology (weighted 25% and 20% of final grade, respectively) and participation in co-evaluation (satisfaction and opinion surveys on the organization of the course) (weighted 5% of final grade). The difference between Group 1 and Group 2 was that, in the second group, an IPA based on the Alexa computer application and integrated into AWS was used from the beginning of the course. Students accessed the IPA using their UBUVirtual credentials. The voice assistant informed the students about events and test deliveries and evaluation procedures in relation to course planning. These events were also collected in a PDF calendar of processes and procedures with assignment dates, available to students from the beginning of the course (an example can be seen in Figure 3). The development of the teaching began on February 3 and ended on April 2 (9 weeks) of 2020. However, on March 12, the Spanish state declared a state of alarm over the COVID-19 health crisis and from that time onwards the teaching was imparted online, exclusively for both groups, over a total period of 4 weeks.A quasi-experimental design with an equivalent control and sample group was used for quantitative data analysis. With regard to statistical analyses, the non-parametric Mann–Whitney U test for independent samples was used to check homogeneity between groups before the intervention. Asymmetry and kurtosis analyses were also used to study the normality of the sample. In addition, to check research questions 1, 2 and 3, a single factor fixed-effects ANOVA and the eta-squared formula yielded their respective effect sizes. In addition, a descriptive multidimensional ideographic design was used for the qualitative analysis. The open-ended responses to research questions 3 and 4 were analyzed, first through a categorization of the responses, and then through a frequency and percentage analysis applied to their categorizations. The SPSS v.24 software has been used for data analysis [66].At the beginning of the project, approval was obtained from the Bioethics Committee of the University of Burgos (No. IR 30/2019). The informed written consent of all participants in the study was documented in accordance with the Declaration of Helsinki.Before the study began, an analysis of the homogeneity between groups was performed with the ACRAr Metacognitive Strategies Scale [58,59], using the non-parametric Mann–Whitney U test for independent samples, of the responses from the students before the instruction commenced. As can be seen in Appendix A, no significant differences are found in Table A1. Therefore, the groups can be considered similar.A normality analysis was then performed on the sample distribution on the ACRAr Metacognitive Strategies Scale [58]. Values over |2.00| are indicative of extreme asymmetry and the lower values that the sample follows are indicative of a normal distribution. Kurtosis values between |8| and |20| suggest extreme kurtosis. In this study, as can be seen in Appendix A, Table A2, no extreme values of asymmetry or kurtosis were detected, so it was concluded that the sample followed a normal distribution, and parametric statistics may be applied.A fixed-effect factor ANOVA was performed (IPA use vs. non-use) to test RQ1. As can be seen from Table 2, significant differences were found for: the number of accesses to the practice resources on the platform in favor of Group 2, for which IPAs returned a high effect size of 43% [F(1,107) = 81.97, p = 0.00, η2 = 0.43]; access to information on the quiz-tests [F(1,107) = 116.25, p = 0.00, η2 = 0.52] in favor of Group 1 that made no use of an IPA, with a high effect size of 52%; and, access to all information on the platform [F(1,107) = 21.81, p = 0.00, η2 = 0.17] in favor of Group 1 that had made no use of an IPA, with a low effect size of 17%.In relation to RQ2, significant differences were only found in favor of Group 1 for the learning outcomes obtained in the practices [F(1,107) = 6.02, p = 0.02, η2 = 0.06] with a very low effect size (Table 3).The tests performed on RQ3 may be checked in Appendix A Table A3. The results indicate that the degree of student satisfaction with the development of teaching in which a blended learning methodology was applied was high in both groups (Group 1: M = 4.90 out of 5, SD = 0.37; Group 2: M = 4.90 out of 5, SD = 0.34). However, significant differences were found in student perceptions of the following items: item 1 (degree of prior knowledge) [F(1,97) = 3.89, p = 0.05, η2 = 0.04]; item 2 (degree of knowledge after completion of teaching [F(1,97) = 4.38, p = 0.04, η2 = 0.04]: item 3 (clarity of the objectives of the course) [F(1,97) = 4.53, p = 0.04, η2 = 0.50], item 7 (facilitation of group work) [F(1,97) = 109.88, p = 0.00, η2 = 0.54], in this case with a high effect size. All of the results are in favor of the group in which the IPA had been applied. Although significant differences were also found in item 9 (possibilities that the development of the subject offers for future labor market insertion) [F(1,97) = 5.35, p = 0.02, η2 = 0.05], in this case in favor of the group in which no IPA had been used.The open-ended responses on the scale were then analyzed. First, a categorization of the responses given by both groups to the open-ended questions was performed. Secondly, a frequency and percentage analysis by category was applied. Both procedures were performed with the program ATLAS.ti v.8 (see Table A4, from Appendix A). The results indicate that for question 1 (“Do you think it is convenient to change anything in the subject? Why?”), the highest response percentage was found in Group 2 in the category “There is no need to change anything” (57.89%); in question 2 (“In your opinion, which units of the current subject should be expanded? In theoretical content or in practical content? Why?”), the highest percentages were found in Group 1 in the category “Nothing” (37.50%) and in Group 2 in the category “Nothing” (54.17%). With respect to question 3 (“In your opinion, which units of the current curriculum should be reduced? In theoretical content or in practical content? Why?”), the highest percentage was found in Group 2 in the “Nothing” category (70%). With regard to question 4 (“Please give any indications you consider appropriate for the improvement of the development of the subject”), the highest percentage was found in Group 2 in the category “There is no need to change anything” (66.67%).Regarding questions on teaching during the COVID-19 state of alert, it was found that in question 1 (“How has work on the platform been in the weeks following the outbreak of the COVID-19 pandemic alert?”), Group 1 had the highest percentages, in the category “Difficult” (25%) and Group 2 in the category “Very good” (75%); in question 2 (“After the COVID-19 pandemic alert, the resources of virtual meetings, email and platform support from the teacher have been.”), Group 1 had the highest percentages, in the categories “Increasing the explanations by videoconference” (16.67%) and “Very good” (16.67%) and Group 2 “Very good” (66.67%). In question 3 (“What would you have added as an aid to teaching during the state of alarm?”), Group 1 had the highest percentage in the categories “Nothing has been taught correctly” (33.3%) and “Nothing, everything has gone very well” (33.3%), and Group 2, in the category “Nothing, this type of methodology has facilitated the continuation of the course” (33.3%). In question 4 (“Would you include any other resources than those used by the teacher (virtual meetings, email and platform support, etc.) during the COVID-19 pandemic alert?”), Group 1 had the highest percentages in the “Nothing” category (33.33%) and, in Group 2, in the “Nothing” category (66.67%). Regarding question 5 (What would you have eliminated as a teaching aid during the state of alarm?), Group 1 and Group 2 had the highest percentages, both in the “Nothing” category (50%).The responses of the students in Group 2 were analyzed, in order to study RQ4, for which the Scale for Evaluating the Functionality of the IPA “UBUVoiceAssistant” was applied. Questions 1 and 2 were respectively answered, on a Likert-type scale and with a yes/no question. The response rate was 87.75%. Regarding the first closed-ended question (“To consult the events of the subject (dates of delivery of practices, dates of tests type test, etc.), what resource do you use?”), 18.6% used the calendar offered by Moodle on the platform by default, 46.5% consulted the process calendar uploaded by the teacher on the UBUVirtual platform, 14% used the IPA, 11.6% consulted their colleagues and 9.3% had noted the information down since the beginning of the course.Regarding the second closed-ended question (“Would you like to receive notifications through an IPA, either on your mobile phone or on another platform?”), 81.4% of students opted to continue receiving notifications on the subject and university activities through the IPA.Answers were categorized for the study of the open-ended questions. Frequency and percentage analyses by category were then carried out on this categorization. All statistical analysis was processed with the ATLAS.ti v.8 tool. With respect to the first open-ended question (“What other information would you be interested in receiving from the UBUVoiceAssistant computer application?”), the answers showed that 10% of the users did not use the IPA, because of the need to open an Amazon account; 20% considered that the application was good, especially for people with special educational needs; 10% never used it; 10% indicated that they would like the application to include notifications when teachers upload resources on the platform; and 50% indicated that they would like the IPA to include information on all subjects during the academic year. Regarding the second open-ended question (“What information would you like the Moodle platform to give you?”), 60% indicated that they would like Moodle to give notices about activities, tests and exam dates. In addition, 40% indicated that they would like Moodle to give them information on resources or activities that the teacher would include in the platform. Regarding the third open-ended question (“If you are not using the UBUVoiceAssistant computer application, please tell us why and make suggestions for improvement”), 90% indicated that they used the IPA, although they would like information on all subjects to be included throughout the academic year. Meanwhile, 10% indicated that they used no IPA, as it is linked to an Amazon account, although they do find this type of application useful.The results indicate that the total accessing of the platform was lower in Group 1, where no IPA had been used, although the effect size was low. Likewise, more accessing of practical activities and teacher feedback was detected in Group 2, and more accessing of quiz-type activities in Group 1, with a high effect size in both cases (43% to 52%, respectively). With respect to learning outcomes, no better results were found in the group in which IPAs had been used. Likewise, student satisfaction with the development of the teaching was high in both groups, with no differences between either one. However, significant differences were detected for student perceptions of their knowledge, both before and after starting to teach. Differences were also found for student perceptions of the development of group work, which was higher in Group 2. Furthermore, in the qualitative study of the responses, greater satisfaction was found in the group in which IPAs had been applied. Along these lines, although both groups were satisfied with the development of the thematic units, the highest percentage was found in the group in which IPAs had been applied. In addition, the group in which no IPA had been implemented perceived the work during the state of alarm of the COVID-19 health crisis as more difficult than the group in which IPA had been used, a group that perceived the work during this period as very satisfactory. Along these lines, the group that had not implemented IPA indicated that more videoconferencing would have been necessary, and only 16.6% perceived that teaching had been “very good”, compared to 66.67% of the group that had implemented IPAs. In addition, this group explained that the methodology in use had facilitated the smooth development of teaching during this period. Nevertheless, both groups perceived the teaching resources used during the health crisis as adequate, although the percentage satisfaction was always higher in the group in which IPAs were implemented. These results support those found in other research on: the use of advanced learning technologies in the LMS as a good resource for learning regulation [13]; the use of advanced technologies in LMS learning with personalized attention [2,12,14]; the use of PBL in LMS environments for increasing collaborative work [6,7]; and the use of LMS with additional technologies, which, together with a pedagogical model similar to the one applied in this study [15], increased the motivation and the effectiveness of learning among students [3]. In addition, specifically in the group with access to an IPA, greater satisfaction was found with the teaching–learning process [22], with teacher guidance in the teaching–learning process [23,24], and greater general satisfaction [44].Regarding the assessment of students who had used IPA, it can be seen that the percentage of systematic use was around 14%, 66% of students opted for more visual resources within the Moodle platform (such as the default use of Moodle in the LMS and the calendar of processes and procedures that the teacher has included on the platform), and 20% of students used none at all. In addition, over 80% of students said they would like to receive information on assessment and test delivery processes and procedures through the Moodle platform with an IPA, as well as other information related to cultural events and events related to their area of knowledge at the university. In addition, some fears were expressed that these devices could invade privacy were linked to a reluctance to use IPAs. In summary, students appreciated the possibilities of using IPAs in university settings [45,46,47]; however, they understand that it is a new technology in this field and consider that there are aspects to be improved, both in terms of functionality and interface presentation [48,49,50,51,52,53].The results of this study should be treated with caution, because we have worked with convenience sampling that assembled a group of students from the specific knowledge area of Health Sciences. In addition, the results point to the existence of strange variables that may influence the results, such as the learning history of the participants. Future studies will therefore be aimed at increasing both the size of the sample and the knowledge branch of each student, as well as evaluating the student records of collaborative learning. Nevertheless, despite the areas of research improvement, it should be noted that there is very little research that refers to the use of IPAs as a support for university teaching, since their preparation and use in LMS requires a complex technological and fieldwork framework.The development of teaching in the university context is increasingly justified by the blended learning design and works towards the inclusion of different additional technologies and PBL resources. Within this framework, the pedagogical design of blended learning spaces in LMS is key to the consolidation of the teaching–learning process. Every day, technology offers new resources that can be incorporated into the LMS, including the use of an IPA. Its use is just beginning and requires important technical adjustments, although it is a very promising resource. University teaching has to implement further digitalization and move towards what has been called the smart university. This idea is gaining in strength, and situations such as the COVID-19 health crisis have only accentuated this trend. It is a present need as much a future one, that must be researched from both a technological and a pedagogical perspective, as well as from instructional standpoints. Moreover, both fields have to go hand in hand, since the functionality of technological resources has to be validated in both fields in an interactive manner, reiterating the need for further studies of blended learning. In addition, it is important to consider that the usage of a voice assistant could help students on their learning process, especially during the COVID-19 crisis, with the selected students specifically enrolled on the Health Science degree at the university. We believe that it is important to research about the advanced technological tools available during the current pandemic situation and how those tools can help all Health Science degree students during their learning path, remembering that in our case, the students sample for this study was taken from students within the area of Health Science. As a potential path for future work, we could consider researching how technological aids influence the mood of students studying for degrees who will be directly confronted with situations such as COVID-19.In summary, this study has contributed innovative results for university learning environments on the use of new technologies: particularly the LMS that incorporates an IPA. Nevertheless, this study has its limitations. As has been indicated, the study has worked with a specific sample size. Future studies will be directed towards expanding the sample, in terms of its size and the heterogeneity of the participating students. Likewise, some qualitative elements have been included in this investigation, although additional elements must be included in future studies, with which triangulation techniques may be applied, to expand the validity of the results. The inclusion of qualitative elements is a great challenge for the advancement of student assessment within virtual university environments. Nevertheless, advancement in this field can only happen with greater investment in both resources and investigation, to confront this challenge with greater assurance.UBUVoiceAssistant Computer application is in the process of being registered [60].The following are available online at https://www.mdpi.com/1660-4601/17/15/5618/s1, Table S1: Course Development Rating Scale, Table S2: IPA Functionality Rating Scale “UBUVoiceAssistant”.Conceptualization, M.C.S.-M., R.M.-S. and J.O.-O.; methodology, M.C.S.-M.; software, J.O.-O.; validation, R.M.-S., and M.C.S.-M.; formal analysis, M.C.S.-M.; investigation, M.C.S.-M., R.M.-S. and J.O.-O.; resources, M.C.S.-M., R.M.-S. and J.O.-O.; data curation, M.C.S.-M.; writing—original draft preparation, M.C.S.-M.; writing—review and editing, M.C.S.-M., R.M.-S. and J.O.-O.; visualization, M.C.S.-M., R.M.S. and J.O.-O.; supervision, M.C.S.-M., R.M.-S. and J.O.-O.; project administration, M.C.S.-M., R.M.-S. and J.O.-O.; funding acquisition, M.C.S.-M. and R.M.-S. All authors have read and agreed to the published version of the manuscript.This research was funded by the Consejería de Educación de la Junta de Castilla y León (Spain) (Department of Education of the Junta de Castilla y León), Grant number BU032G19, and grants from the University of Burgos for the dissemination and the improvement of teaching innovation experiences of the Vice-Rectorate of Teaching and Research Staff, the Vice-Rectorate for Research and Knowledge Transfer, 2020, and the Departamento de Ciencias de la Salud the University of Burgos (Spain). And the support for the dissemination of research results for teachers and doctorands of the Vice-Rectorate for Research and Knowledge Transfer, 2020.To all students who have participated in the study.The authors declare no conflict of interest.Two independent samples test in parametric test.Analysis of the asymmetry and kurtosis values of the distribution.Note. M = mean age; SD = standard deviation; A = asymmetry; K = kurtosis; ASE = asymmetry standard error; SEK = kurtosis standard error.Single factor fixed-effects ANOVA (use of IPAs vs. non-use)* p < 0.05.Categorization of the answers to the open questions of the scale. Analysis of percentages and frequencies found with ATLAS.ti v. 8Note. F = frequency; % = percentage.Intelligent personal assistant (IPA). Skill Alexa “UBUVoiceAssistant”.Diagram of the operation of the “UBUVoiceAssistant” application from Moodle.Process and procedure schedule.Description of the sample and the variables: gender and age.Note. Mage = Mean age; SDage = Standard Deviation age.A single factor fixed-effects ANOVA (IPA use vs. non-use).* p < 0.05. Note: N = number of participants; M = Mean; SD = Standard Deviation; η2 = eta squared (effect size); G1 = Use de IPA; G2 = No use of IPA.Single factor fixed-effects ANOVA (IPA use vs. non-use).* p < 0.05. Note: N = number of participants; M = mean; SD = standard deviation; η2 = eta squared (effect size); G1 = IPA use; G2 = IPA non-use. One participant in Group 1 and 3 participants in Group 2 never completed the course.

Med-MDPI/ijerph_5/ijerph-17-15-05619.txt

@@ -0,0 +1 @@
+Understanding how people’s worldviews and individual personality differences affect their thinking about anthropogenic climate change is critical to communication efforts regarding this issue. This study surveyed University of Georgia students to investigate the role that need for cognitive closure plays in level of climate change worry. The relationship between these two was found to involve suppression—a subset of mediation—by the social dimension of political conservatism. Political conservatism was also found to play a mediating role in the relationship between need for cognitive closure and support for governmental and personal climate solutions. However, social conservatism played this mediator role in women, and functioned as a suppressor for men. These findings help inform audience segmentation and creation of climate-related messages based on audience worldview and personality.Despite the vast amount of scientific evidence showing that human activities are causing the Earth’s climate to rapidly change; and that these changes will result in dire consequences in the future, there is substantial doubt and inaction regarding this issue in the United States [1,2] Only about 57% of American adults know that climate change is caused mostly by human activities, and only 49% know that most scientists agree that this is the case [2]. Enough doubt has been created about these facts, some of it purposefully, that the United States, one of the top emitters of fossil fuels worldwide, is making little to no progress to solve this issue [3]. It is of utmost importance for the future of humanity that Americans change their behaviors regarding climate change, from grassroots individual actions to sweeping policy and energy generation changes. A key part of making this behavior change happen is communicating the risks and ramifications of climate change in such a way that people are motivated to make a change. Thus, it is critical to study how climate change is communicated, what messages work for different audiences, and what makes different audiences more or less receptive to climate-related information. One way that audiences can be differentiated is by gender. It has been shown that women generally exhibit greater climate change knowledge and concern than men [4]. However, they also tend to underestimate their level of knowledge more than men do [4]. Another divider when it comes to climate change belief is political affiliation. There has been significant political polarization regarding the climate change issue since approximately 2000 [5]. Much of this partisan divide is based upon fear of increased regulation in conservatives [5]. This polarization makes messaging difficult, as people immediately side with their political allegiance when they hear the words “climate change”. Therefore, it is important to look at other factors that influence climate change beliefs, and maybe even be underlying in political affiliation. One such avenue is individual personality variables, which can play a role in receptivity to and effects of climate change communication.Individual differences in people’s worldviews can be a major barrier to climate change related communication. Messages that work for someone with one worldview will likely not work for someone with another. One aspect of a person’s worldview is their level of need for cognitive closure (NCC). NCC is defined as an individual’s need for a definite answer to a question and intolerance of ambiguity [6]. The impact of NCC on people’s behaviors and information processing has been studied through a number of different avenues relating to the climate change issue. These include conspiracy belief [7], precautions against long-term or future risks [8], open-mindedness and competing messages [9], and pro-environmental attitudes and behaviors [10].In a 2017 study, Marchlewska et al. investigated the connection between NCC and belief in conspiracy-related explanations for events. Results of this study showed that there was a positive link between NCC and belief in conspiracies, and this effect was strongest when the conspiracy theory was salient. Results also showed that people high in NCC latched on to conspiracy-based explanations when the cause of an event was uncertain, but not when the cause of the event was certain [7]. This study sheds light on what motivates people high in NCC to adopt a conspiracy-based belief in the face of uncertainty, which may help climate communicators to understand why some people do not believe the uncertainty-laden projected effects of climate change. One example of a conspiracy belief about climate change is “Climategate”, which occurred in November 2009. The Climate Research Unit at the University of East Anglia was hacked, and over 1000 emails between researchers were published. The released emails used scientific jargon, slang terms, and acronyms that raised alarm within the general public, and some sentences were taken out of their proper context by news outlets [11]. An investigation found no fault in the integrity of the research being done by the scientists whose emails had been hacked and published [12]. However, conspiracy beliefs about climate change persist despite this. University of Georgia professor and Weather Channel podcast host Marshall Shepherd has had plenty of experience with such beliefs. “I get emails weekly from people with their theories and thoughts on climate change or why I’m absolutely wrong on something,” he said. These beliefs can come from individuals themselves, or can stem from coordinated misinformation campaigns: “I think you have this sub-group of people in the American population…that very much propagate these conspiracy theories,” Shepherd says, “but then you do have this deeper, coordinated effort to propagate conspiracy….you see certain organizations and think tanks talking about different theories of climate change and that we need to teach alternative perspectives about climate change in schools—that’s coordinated effort, and so we need to understand that” [13]. In their 2010 book Merchants of Doubt, Naomi Oreskes and Erik Conway confirm the presence of intentional climate-misinformation campaigns [3]. Marchlewska et al.’s 2017 [7] study could help us to understand the role that NCC plays in misinformed, conspiracy-based climate change beliefs.Individual differences in personality can affect whether people tend to take precautions regarding their future. One such precaution is screening for chronic medical problems such as cancer, diabetes, or heart disease. Eiser and Cole [8] investigated how individual differences in young women ages 20–25 affected the behavior of getting a regular cervical cancer screening. Results showed that respondents lower in NCC were less likely to have had a test for cervical cancer or to be planning to get one. The authors say this might mean that for people high in NCC, the resolution of uncertainty in and of itself may be a motive for getting tested. That is, they do not solely seek a negative/all clear result, they want to know if they have something wrong with them. Based on these results, climate change may be able to be framed similarly to a long-term illness like cancer that can be prevented or dealt with efficiently through precautionary action. However, climate change messages exist in a highly competitive environment. While there are messages about how climate change must be dealt with, people are also being bombarded with messages saying that climate change is a hoax. This can cause a lot of conflict for people forming opinions about climate change and climate action strategies.One aspect of NCC is a person’s open- or closed-mindedness. Nisbet et al. [9] investigated in 2013 how open/closed-mindedness affects attitudes towards climate change in noncompetitive vs. competitive message environments. Low NCC is associated with open-mindedness, and high NCC is associated with closed-mindedness. This study found that when a video message focusing on the positive outcomes of climate policy/action was viewed in conjunction with a video message focusing on negative economic costs of climate policy/action, people who were less open-minded perceived fewer benefits to climate policy and were less supportive of it. On the contrary, those who were more open-minded and viewed the two messages in competition exhibited increased benefit perception and were more supportive of climate policy/action [9]. Results of this study help to reveal how people high in NCC process the often-conflicting messages about climate change that are portrayed by the media, which helps communicators in the effort to garner support for actions against climate change.Support of climate policy and action, as well as individual behaviors that protect the environment, are critical for the fight against climate change. Panno et al. [10] investigated how NCC affects pro-environmental behavior. This study was done with two surveys. In the first, respondents’ NCC levels and pro-environmental preferences and behaviors were measured, and political orientation was self-reported. This sub-study sought to understand the connection between NCC and pro-environmental behavior, and found that the two were related through political belief. In the second study, participants answered questions to measure their levels of NCC, pro-environmental attitude and behavior, and economic and social political beliefs. This second study found that this relationship was mediated by the social dimension of political ideology—conservatism specifically [10]. Communicators can use these results to target social aspects of conservative political belief in message building, as opposed to trying to cater to the economic aspect as well.Although NCC has been applied to study pro-environmental behaviors and attitudes, the present study is the first to use NCC to study concern for climate change specifically. NCC was chosen for this study because its use regarding environmental behaviors provides a framework that this study can be modeled after. NCC has also been shown to relate to political beliefs, and climate change is a very politically charged topic in today’s society. Finally, NCC was chosen because climate change can be considered a long-term risk, similar to the risk of cervical cancer—the perception of which Eiser and Cole studied with relation to NCC as described above [8]. Environmental behaviors, political views, and climate/long-term risk perception are all related to the hypothesis described below.One of the keys to combatting anthropogenic climate change is changing the behavior of the world’s population. Behavior changes can be brought about in a number of different ways. One method of changing behavior is to tailor a message encouraging behavior change towards personality aspects shared by a relatively large group of people. One such personality trait is level of need for cognitive closure (NCC), an indicator of how well a person deals with uncertainty in their life and how quickly they look for a definite answer to an open question. As climate change entails a certain level of uncertainty, and NCC has been studied already with respect to environmental beliefs, NCC presents an interesting avenue of study in the realm of how personality relates to climate change perspectives.The goal of this study is to examine the relationship between NCC and views regarding climate change; and the relationship between NCC and climate solution support, as well as the role that political beliefs may play in these relationships. There are many ways to segment and investigate audiences when it comes to climate change communication. This study focuses on people who may believe climate fallacies due to their aversion to the uncertainties associated with projected effects of climate change. These uncertainties have been used by climate denial campaigns to discredit scientists; and uncertainty can lead to a lack of self-efficacy in audiences. Therefore, this study focuses on individuals with a high level of need for cognitive closure, and the effects of high NCC on their climate change worry and solution support.This study sought to answer the question of whether level of need for cognitive closure (NCC) plays a role in climate change worry (CCW), and whether this relationship is mediated by political views. Additionally, we wanted to explore potential relationships between each of these variables and level of support for various climate solutions.Research Question 1. How do need for cognitive closure and conservatism affect climate change worry?Research Question 2: How do need for cognitive closure and conservatism affect support for climate solutions?In order to investigate these questions, a survey was distributed online via the Qualtrics platform to undergraduate and graduate students at the University of Georgia. As an incentive for participation, students were offered the chance to enter for a drawing of one of five $40 Amazon gift cards. The survey contained four scales, measuring participants’ levels of need for cognitive closure, levels of worry about climate change, levels of support for various climate change solutions, and both the social and economic aspects of their political beliefs. Demographic questions, including gender identification, race, religion, and age were asked at the end of the survey. The University of Georgia (UGA) is a public university in Athens, Georgia, with approximately 30,000 enrolled undergraduate students. Students are mostly white, with Asians, African-Americans, and Hispanics making up most of the rest of the student body. Table A5 and Table A6 in the appendix provide information on the ethnic and religious diversity of our sample. There are approximately 5000 more female-identifying undergraduates than male-identifying. The University of Georgia is involved in the Georgia Initiative for Climate and Society, has a Climate Action Planning Task Force, and runs an Office of Sustainability. Although not all UGA students will encounter climate-change-related information and action during their studies, the university displays a commitment to climate action (ethical approval number: PROJECT00000650, date 18 June 2019).The 15-item NCC Scale with Weights, created by Roets and Van Hiel [14], was used to measure respondents’ levels of need for cognitive closure. This scale was chosen because it assesses all dimensions of a person’s NCC level while reducing the number of questions from the 41 in their modification of the original scale to a more manageable—15 [6,15]. This allows for multiple scales to be used in this survey without requiring respondents to answer a cumbersome number of questions. Questions in the 15-item NCC Scale with Weights are rated on a 6-point Likert scale from 1 = Strongly Disagree, to 6 = Strongly Agree. Their responses are summed, and higher sums indicate higher NCC. The test-retest stability of this scale is 0.79, and the Cronbach’s alpha is 0.89, which is comparable to the original 41-item scale’s 0.9. Additionally, there is no difference in scale means between the 41-item and 15-item scales [15].The Climate Change Worry Scale, developed by Stewart [16] was used to measure respondents’ level of concern regarding climate change and its effects. This is a 10-item scale that assesses how often respondents have worries regarding climate change and its effects. Participants indicate how frequently each item, in the form of a statement, applies to them on a five-point Likert scale ranging from 1 = Never to 5 = Always. The Cronbach’s alpha for this scale is 0.95.Climate change solution support was measured using questions from the Yale Program on Climate Change Communication’s surveys on which their Climate Opinion Maps are based [2]. Eight questions were chosen from the original list of sixteen. This subset was chosen to focus on specific items related to the scale of various climate mitigation actions (i.e., federal, local, and personal), and to limit the number of questions in the survey so that participant fatigue could be minimized. The Cronbach’s alpha of this scale is 0.79.Respondents self-reported their social and economic political conservatism, respectively, on a scale ranging from 1 = Very Liberal, to 7 = Very Conservative. Self-reporting was chosen to minimize the number of questions in the survey, and to make the distinction between social and economic conservatism easy to find within the data. These questions had been previously used by Cacciatorie, Scheufele, and Corley in their 2011 study [17].The first step in analyzing survey results was to look at first-order correlations between each of the variables studied. These relationships were considered significant at α = 0.05. Following the analysis procedure of Panno et al. [10], these indicators of simple relationships were then used to inform the creation of mediation models. Detailed information can be found in the Supplementary Materials.A number of different mediation models were created for the first part of this study. Mediation models were used following the example of Panno et al. [10]. While Panno et al. went to a regular regression model before mediation, we were able to move straight to mediation models based on their findings [10]. The first, and most broad model involved CCW and NCC mediated by overall political conservatism for both men and women. This model was subsequently broken down, due to differences between the genders previously noted in the literature [10]. The second set of mediation models depicted the relationship between NCC and CCW as mediated by overall conservatism for men and women, respectively. Finally, a third set of models was created based on Panno et al.’s finding that the social dimension of conservatism played a larger role in the NCC/pro environmental attitude relationship. This third set contained models of the NCC/CCW relationship as mediated by social conservatism for both men and women separately, and economic conservatism for both men and women separately. In total, seven mediation models were a part of this analysis. Coefficients in these models were considered significant at α = 0.05. These models were created in R (version 2.3, R Fundation for Statistical Computing: Vienna, Austria,2006), an open-source statistical software, using the psych, expss, diagram, and lavaan packages.The second part of this study investigated relationships between level of need for cognitive closure, dimensions of conservatism, and climate change worry with level of support for personal versus governmental climate change solutions. First order correlations between each of these variables were calculated, and similarly to the first part of this study, were used to inform the creation of mediation models. These models were created for the relationship between NCC and personal/government solution support, mediated by overall, social, and economic conservatism, separated between men and women. A total of twelve mediation models were built to analyze each of these relationships. We chose these models because level of support for climate solutions is more behavioral than climate change worry—this allowed us to bring a behavioral aspect to the study, in a similar manner to Panno et al. [10]. R (version 2.3) was again used to create these models, using the psych, expss, lavaan, car, and diagram packages.In both of these studies, a Sobel test was used to determine which mediation effects were significant. The Sobel test is a special form of a t-test that is used to determine whether there is a statistically significant difference between the direct effect of an independent variable on the dependent variable, and the effect when the mediator is taken into account. This test was chosen because of its wide usage and acceptance in mediation research. The Sobel Test was run using a website widget, which can be found at http://quantpsy.org/sobel/sobel.htm.We received approximately 1100 responses to the survey. Of these participants, 731 were women, 363 were men, and 17 were other/non-binary/preferred not to answer. Respondents were 70% white, 6.5% Latinx, 6.6% black/African-American, and 11.3% Asian/Pacific Islander (5.9% other/prefer not to answer). Most respondents were either in the center of the political spectrum or slightly left of center (more liberal); however, the political spectrum overall was fairly well-represented (51% of the sample on the left, 49% on the right, see Figure 1). Respondents tended towards slightly higher levels of NCC, shown by the slight left skew of the histogram in Figure 2. Level of climate change worry showed a very large spread, with a drop-off at higher scores (Figure 3).The matrix of correlations between need for cognitive closure, conservatism, and climate change worry (Figure 4) shows that climate change worry was negatively correlated with all three conservatism items (overall/CCW = −0.54, economic/CCW = −0.48, social/CCW = −0.52). This indicates that climate change worry decreases with increasing levels of all three dimensions of conservatism. The social conservatism correlation is slightly larger than that of economic conservatism. Need for cognitive closure, however, was only significantly correlated with social and overall conservatism, with the slightly larger coefficient between the two also being social conservatism (0.09 vs. 0.07). The correlation between need for cognitive closure and climate change worry (−0.01) was insignificant, as denoted by the “X” in that particular box of Figure 4. Because the direct correlation between NCC and CCW was insignificant, but correlations between NCC/conservatism and conservatism/CCW were, we created mediation models to further investigate how these three variables interact with each other.A mediation model showing the relationship between need for cognitive closure and climate change worry as mediated by overall conservatism for both men and women (Figure 5) tells a similar story to the correlation matrix (Figure 4). In these models, the a coefficient represents the effect of NCC on conservatism, b represents the effect of conservatism on CCW, c represents the direct effect of NCC on CCW, and c’ represents the indirect effect of NCC on CCW, which takes the mediating variable into account. There was a strong and highly significant relationship (−0.55, p < 0.001) between overall conservatism and climate change worry. The relationship between need for cognitive closure and overall conservatism was smaller in magnitude but still significant (0.09, p = 9.786 × 10−5). Because of previously described differences between genders [18] this model was broken down into two parts: one using responses from men and the other using responses from women. A small number of responses from individuals identifying as non-binary or “other” (n = 17) were received, therefore, these responses were not used in these analyses. Table A1, found in the appendix, shows the a, b, c, and c’ values for each of the NCC/CCW models. The figures that display significant models in the triangle form shown in Figure 5 are contained in the appendix.The relationships between need for cognitive closure and overall conservatism are similar for men (a = 0.12, p < 0.01) and women (a = 0.13, p < 0.001). Comparable relationships between overall conservatism and climate change worry were also found for men (b = −0.59, p < 0.001) and women (b = −0.51, p < 0.001). However, men and women differ greatly when it comes to the direct NCC/CCW and mediated NCC/CCW relationships. Women exhibit partial mediation, with c’ < c, (c = −0.1, p = 0.00575; c’ = −0.03). Men, on the other hand, exhibit suppression. For men, c’ > c, with c’ being significant (c = 0.05, p = 0.351; c’ = 0.12, p < 0.001). Another interesting difference is that the c coefficients for men are positive, while for women, they are negative. This partial mediation vs. suppression difference remained when we broke these gender-based mediation models down further by the dimensions of conservatism: social and economic. The Sobel test was used to determine the significance of the mediation effects. The only effects that were significant were social and overall conservatism’s suppressing effect in men. Table A2 and Figure A2 and Figure A3 in the appendix model these relationships visually.Figure 6 shows correlations between climate change worry, need for cognitive closure, conservatism (overall, social, and economic), and support for both government-based and personal climate solutions. One thing to note from these correlations is that the coefficient for climate change worry and government-based solutions is close to twice that of climate change worry and personal actions (0.56, 0.33). Correlations between need for cognitive closure and these variables were small (i.e., −0.09, 0.02), while correlations between conservatism and solution support were larger (i.e., −0.26, −0.55). To further investigate these relationships, we created models representing the relationships between need for cognitive closure and solution support mediated by the dimensions of conservatism. Table A3 in the appendix shows the coefficients of each of these models.Both men and women show an interesting difference in b coefficients between solutions. The b coefficients for government solutions are larger than for personal actions. For example, in the social conservatism mediator/government solution model, the women’s and men’s coefficients are both −0.54. In the social conservatism mediator/personal solution model, the men’s b coefficient is −0.25, and the women’s is −0.27. The a coefficients are also fairly similar across genders—NCC and social conservatism is 0.13 for women and 0.17 for men, for example. However, these models exhibit the same partial mediation vs. suppression difference between genders as the previous NCC/CCW models. All types of conservatism function as suppressors in the men’s models. For women, all types of conservatism are partial mediators. Only the effects on government solution support are significant, as shown by Table A4 in the appendix. Table A4 shows that only government-based solutions exhibit a significant conservatism-mediated/suppressed relationship with need for cognitive closure. These significant models are included with the figures in Appendix A (Figure A1, Figure A2, Figure A3, Figure A4, Figure A5 and Figure A6; Table A1,Table A2, Table A3,Table A4, Table A5 and Table A6).The results of this study show that for college students, overall, men consistently exhibit suppression by conservatism in the relationship between need for cognitive closure and both climate change worry and solution support. This means that an already-existing relationship between NCC and CCW/solution support is strengthened by the presence of conservatism. Women, on the other hand, consistently exhibit partial mediation by conservatism for the NCC/government solution support relationships, but none of the others. Partial mediation means that the presence of conservatism is necessary for there to be any relationship between NCC and government solution support. This gender-based difference has some implications for climate messaging.For the NCC/CCW relationships, the only significant results were suppression in men by social and overall conservatism. This means that, for men, the presence of conservatism—social, especially—increases the positive effect of need for cognitive closure on climate change worry. Men who are socially conservative and high in NCC are therefore more likely to exhibit high climate change worry than their counterparts who are less socially conservative. Therefore, there is likely a certain group of socially conservative men for whom messages combining uncertainty resolution and the values of social conservativism would be particularly effective. However, message testing research should investigate this further.Of the two types of solutions presented in this survey study—personal actions and government actions—only the relationships between NCC and government solutions mediated by all dimensions of conservatism were significant. Personal solution support was not significantly related to NCC through conservatism. Within the NCC/government solution relationships, however, the aforementioned gender difference was observed. For men, who exhibited suppression, the presence of conservatism intensified the positive relationship between NCC and government solution support. That is, support for government solutions increases with higher NCC in men, and this increase is stronger when conservatism is present. For women, who exhibited partial mediation, the presence of conservatism plays a vital role in the negative relationship between NCC and government solution support. That is, support for government solutions decreases with higher NCC in women, and conservatism is a necessary link between these two—without it, the relationship is not significant.Taken together, these results show that high need for cognitive closure does play a role in climate change worry and solution support, mediated by various dimensions of conservatism. However, this role varies between men and women. While the particular power of social conservatism over economic is consistent with the findings of Panno et al. [10], this study found a significant difference between the genders that was not present in the Panno et al. [10] study. This difference may be present due to the highly politicized and divisive nature of climate change in American society. While environmental issues in general, such as those in the Panno et al. study, are also a topic of debate, climate change is a particularly hot button issue. Thus, it is possible that the use of climate change in the present study resulted in further division than observed in Panno et al. [10]. The present study’s findings can be used to inform audience segmentation and message creation. For example, messages could focus on losses that are results of climate change. For college students, these might include loss of or change to coastlines, spring break destinations, or towns where they aspire to live someday. Further, this message could be expanded to include the uncertainty element and utilize need for cognitive closure, especially for those who have the predictive power of need for cognitive closure increased by social conservatism. A message like this might touch on recent costly disasters, and acknowledge that there is limited uncertainty as to how these disasters will change with a changing climate—but we can take actions to reduce that uncertainty by reducing CO2 emissions. The suppression effect seen in men with social conservatism and government solution support means that for some people, combining government action examples with uncertainty resolution may increase support for those government solutions. In that case, the message could emphasize that the level of uncertainty regarding changing disasters can be reduced by taking actions such as a carbon tax or replacing fossil fuels with renewable energy sources—decreasing the likelihood that we could see worsened disasters. However, the mediation effect and negative relationship seen with all dimensions of conservatism in women with regards to government solutions means that this strategy may be more likely to backfire than to work with certain groups of people. Potential audiences for these messages should be studied and targeted carefully. As mentioned previously with the research questions, these messages are meant to resolve uncertainty and increase self-efficacy in those with high need for cognitive closure. Individuals who are low in need for cognitive closure will not be affected by messages about uncertainty, and thus, they are not the focus of the example messages or this study.It should be noted that these findings are not necessarily applicable to a general population, for two reasons. First, this study used college students as subjects; and while the respondents were somewhat diverse in race, (70% white, 6.5% Latinx, 6.6% black/African-American, and 11.3% Asian/Pacific Islander, see Appendix A for other statistics), this cannot be considered a representative sample. Second, the difference between genders raises some interesting questions. While general worldview does not necessarily strictly follow the traditional gender binary, it is clear from the results that suppression applies to men while partial mediation applies to women. What is unclear is the reason behind this difference.There are a few avenues for further research based on this study. Of course, one is to investigate the suppression vs. partial mediation difference between the genders. Another is to apply this framework to a more representative sample, to investigate whether these findings are valid beyond a college student sample. This framework would likely be effective in countries aside from the US that have a simple two-party political belief system. Adjustments may need to be made for use in countries such as Italy, where political parties are numerous. Another is to rearrange the mediation models in the solution-based part of the study to learn more about the potential effects of climate change worry on solution support, as a mediator or as an independent variable. Other individual personality differences could be added to these models to determine what other variables may result in the difference in conservative’s function between men and women. In addition, the role of conservatism itself in these relationships could be studied in more detail, to investigate these differences between men and women. Future messaging studies could test each of the aforementioned potential message formats on various populations based on personality and political leaning. Another interesting avenue for further research is the development of a “Need for Climate Change Closure” scale. Such a scale could adapt the pre-existing questions in either the 41-item or 15-item scale so that they ask about climate change specifically. Similarly, while the brevity of the self-reported conservatism [17] scale helps to keep the number of survey questions down and reduces participant fatigue, there are limitations to self-reporting political affiliation. The use of the terms “liberal” and “conservatism” encourage respondents to respond according to the group with which they identify. For example, a respondent may answer “very conservative” because they see themselves as a member of the conservative group of society, where their actual views may fall more centrally on this spectrum. “Political identity” may be a better term for the outcome of the questions in this scale, and this could be an interesting metric for future studies to consider. Additionally, this study did not investigate behaviors such as implementation of individual solutions into one’s lifestyle, or voting behaviors. Future work should build on the present study by looking in to how government solution support relates to voting behaviors. A previous study by Sinatra et al. investigating the outcomes of persuasive messages on college students’ intents to act on climate change showed positive changes in attitudes about climate change and willingness to act, and may serve as a helpful resource in this future avenue of study [19]. The college student age group represents the world’s leaders and voters of tomorrow, making them an important population to consider and study with regards to their support and implementation of vital climate change mitigating actions.The following are available online at https://www.mdpi.com/1660-4601/17/15/5619/s1.Conceptualization, M.O.; methodology, M.O., A.S., and A.G.; formal analysis, M.O.; data curation, M.O. and A.S.; writing—original draft preparation, M.O.; writing—review and editing, A.S. and A.G.; supervision, A.G. and A.S.; funding acquisition, M.O. All authors have read and agreed to the published version of the manuscript.This research received no external funding. Margaret Orr completed this research as part of an MS degree in Geography at the University of Georgia, where she was funded by a Teaching Assistantship.The authors wish to thank the Geography Department at the University of Georgia for providing TA funding for this research to take place. We also thank Master’s committee members Glen Nowak and Gabe Kooperman for their help in this project. Finally, we thank the reviewers for their helpful feedback.The authors declare no conflict of interest.Model of the relationship between NCC and CCW as mediated by conservatism for men. The numbers in this model represent the coefficients of each relationship in a regression model. The c’ coefficient is in parentheses. (*** = p < 0.01, ** = p < 0.05, * = p < 0.1).Model of the relationship between NCC and CCW as mediated by social conservatism for men. The numbers in this model represent the coefficients of each relationship in a regression model. The asterisks denote the level of statistical significance of the observed statistics. The convention is *** p < 0.001 and ** p < 0.05.a, b, c, and c’ values for models representing the NCC/CCW relationship with overall, social, and economic conservatism as mediators. The asterisks denote the level of statistical significance of the observed statistics.The convention is *** p < 0.001, ** p < 0.05, and * p < 0.10.Indirect effect (a*b) values and Sobel Test t and p values.The convention is ** p < 0.05, and * p < 0.10.Mediation model coefficients for the relationships between NCC and government/personal solution support, with dimensions of conservatism as mediator/suppressor. The asterisks denote the level of statistical significance of the observed statistics.The convention is *** p < 0.001, ** p < 0.05, and * p < 0.10.Sobel t and p values for solution support. The asterisks denote the level of statistical significance of the observed statistics.The convention is *** p < 0.001, ** p < 0.05, and * p < 0.10.Relationship between NCC and government solution support as mediated by social conservatism for men. The numbers in this model represent the coefficients of each relationship in a regression model. The asterisks denote the level of statistical significance of the observed statistics. The convention is *** p < 0.001 and ** p < 0.05.Relationship between NCC and government solution support as mediated by social conservatism for women. The numbers in this model represent the coefficients of each relationship in a regression model. The asterisks denote the level of statistical significance of the observed statistics. The convention is *** p < 0.001 and * p < 0.01.Relationship between NCC and government solution support as mediated by overall conservatism for women. The numbers in this model represent the coefficients of each relationship in a regression model. The asterisks denote the level of statistical significance of the observed statistics. The convention is *** p < 0.001 and * p < 0.01.Relationship between NCC and government solution support as mediated by economic conservatism for women. The numbers in this model represent the coefficients of each relationship in a regression model. The c’ coefficient is in parentheses. The asterisks denote the level of statistical significance of the observed statistics. The convention is *** p < 0.001, ** p < 0.05 and * p < 0.10.Ethnicity Counts.Religion Counts.Note: many of the “other” responses listed specific Protestant denominations, i.e., Presbyterian, Baptist, Methodist, etc.Histogram of political stances. Higher numbers indicate more conservative stances.Histogram of scores on the need for cognitive closure (NCC) scale. Higher numbers indicate higher level of NCC.Histogram of scores on the climate change worry (CCW) scale. Higher numbers indicate higher levels of climate change worry.Correlation matrix for climate change worry (CCW), need for cognitive closure (NCC), social conservatism, economic conservatism, overall conservatism. These numbers represent the R2 value for each of the relationships.Model of the relationship between NCC and CCW as mediated by conservatism for both men and women. The c’ coefficient is in parentheses. The asterisks denote the level of statistical significance of the observed statistics. The convention is *** p < 0.001 and * p < 0.10.Correlation coefficients (R2) for CCW, NCC, different dimensions of conservatism, and level of support for government-based and personal climate solutions. Each of the correlations displayed are significant (p < 0.05).

Med-MDPI/ijerph_5/ijerph-17-15-05620.txt

@@ -0,0 +1 @@
+The term Attention Deficit Hyperactivity Disorder (ADHD) has a long history of problems behind it. The origin of all these problems lies in the lack of agreement in the assessment procedures and evaluation instruments. The diagnosis is clinical and is determined by the observation and information provided by parents and teachers. So, this is highly subjective and leads to disparate results. Therefore, on the one hand the inaccuracy of the diagnosis of ADHD, which has been based on subjective criteria, together with the fact that hyperactivity is one of the main symptoms of this disorder, implies that several studies (with limitations) have been carried out to record objective measures of movement in subjects in at least the last ten years. In order to solve some of this derived problems and limitations of previous studies, a computer program has been developed to objectively record the amount of movement of subjects. The main objective of this study is threefold: first to register the amount of movement of both experimental group and control group, then to compare them with the movement registered by observers and finally to determine the validity of the software developed as a tool to support the diagnosis of ADHD. Results show that there are significant differences in the amount of objective movement between a clinical group of subjects with ADHD and a control group, obtaining a higher average of movement the experimental group. In addition, results also demonstrate that the developed software is a valid tool for the evaluation of movement that solves the limitations of previous studies. The proposed tool is developed from different aspects to give it a multidisciplinary character.The term Attention Deficit Hyperactivity Disorder (ADHD) is used by the American Psychiatric Association in the Diagnostic and Statistical Manual of Mental Disorders [1] to describe a persistent pattern of inattention and/or hyperactivity-impulsivity that is inconsistent with developmental level, and that impacts negatively and directly on social and academic/occupational activities. We are thus situated in the last interpretation of an old problem that, although highly topical, has a long history of queries behind it.The problems related to ADHD not only come from outside the scientific community, but have also spread within itself: multiple doubts in reference to etiology, prevalence, evaluation instruments and procedures, etc. The key to all these problems lies in the lack of agreement when considering diagnosis. Thus, in the absence of any biochemical, structural or genetic condition that unequivocally determines the existence of ADHD, the diagnosis is clinical, that is, based on the professional expertise of the doctor, and it is determined by the observation and information provided by parents and teachers [2]. This is highly subjective and leads to disparate results, largely due to the lack of agreement in the assessment procedures and evaluation instruments [3]. In addition, as it has been stated by the Spanish Association of Neuropsychiatry, “the main problem is that the clinical criteria used to diagnose this disorder are too vague. The diagnostic manual used, DSM-V, includes very broad and subjective criteria. How can be determine if a child is more or less prone to a higher degree of movement?” [4].Consequently, the inaccuracy of the diagnosis of ADHD, based on subjective criteria [5], together with the fact that hyperactivity is one of the main symptoms of this disorder [6,7,8], means that, for more than a decade, several studies have been carried out to record objective measures of movement in the subjects. However, these studies have a number of limitations. First of all, the use of accelerometers (actigraphy and inertial measurement units), that is, that some device has to be placed in the body of the subjects, limiting their ecological validity and, secondly, the use of infrared devices that only analyse some parts of the body [9,10,11,12,13].The present paper tries to fill the aforementioned shortcomings by using Microsoft Kinect V.2. This device is able to completely track and record 25 joints of six human bodies through an infrared system, without placing any type of sensor in the body of the subjects [14,15,16,17,18,19]. The commercialized version of this device allows developers to make their own programs by using gestures and body movements in a wide variety of applications (Ding and Chang [20]) besides, some emerging researches on the use of this device (that enable us to count limb movements in subjects with ADHD) conclude that it is a good device to measure movements [21]. Furthermore, since this is a device of small dimensions, it can be introduced into the natural environment of the subjects.For all the above, and trying to contribute to broaden the knowledge about the behaviour of subjects with ADHD, a computer program has been developed to objectively record the amount of movement of the subjects. Subsequently, study techniques workshops have been designed as an incentive used for attendance, both for subjects with a firm diagnosis of ADHD and for control subjects.The main objective of this study is to introduce a multidisciplinary tool developed for objective analysis of movement in subjects with ADHD. A computer program was developed to record the amount of movement of subjects. The validity of this software as a tool to support the diagnosis of ADHD has been carried out by comparing the amount of movement of the group diagnosed with ADHD and the control group, and it has also been compared with the movement registered by observers.This research involved 65 children, aged between 7 and 12 years old. The children were classified into two groups: experimental group and control group. The experimental group consisted of 32 children with a firm diagnosis of ADHD: 24 boys and 8 girls. In total, 24 of the 32 children had a diagnosis of ADHD with a combined presentation and 8 with a predominantly inattentive presentation, according to the DSM-V criteria. The age range of the participants of the experimental group was 7 to 12 years old (M = 9.75 and SD = 1.50).The control group consisted of 33 children from a standardized sample: 17 boys and 16 girls. The age range of the participants in the control group was 8 to 12 years old (M = 9.69 and SD = 1.50).The project was presented to the Ethical Committee for Clinical Research (CEIC) of the General University Hospital of Alicante (HGUA), obtaining its approval. From that moment onwards, informed consent to participate in the study was given to the parents of children diagnosed with ADHD by the Children’s Mental Health Unit (USMI).Once the procedure for the recruitment of the sample of the experimental group was started, a primary education school was contacted for the recruitment of the sample for the control group. The same number of boys and girls were randomly recruited as the experimental group.The first step for the organization of the study techniques workshops was to look for an only space, since the attendees (the sample) came from different educational centres. We searched for a classroom at the University of Alicante (UA) with writing pad chairs, because it would greatly facilitate the recording with the Kinect device.The different sessions of the workshops were organized by grouping children on age basis and ensuring that there were children from both the control group and the experimental group in the different workshops. Only between 4 and 6 people attended to each of the workshops, due to the limitations of the Kinect device (it can only detect and track 6 bodies).Families of children sent by the USMI that were on some type of medication, were invited to attend a new workshop to “review and expand” study techniques. Second to this workshop, they had to attend without having taken any medication. The families showed interest in collaborating, since all of them agreed, so that we could have all the subjects of the experimental group that has been prescribed some medication within two measurements of movements: with and without medication.Two Kinect devices were used in each workshop in order to ensure the capture of movement of the subjects and prevent any type of technological incident Figure 1.The classroom was arranged in such a way that the speakers never stood in front of the Kinect devices (Figure 2). Thus, the speakers would cross behind the chairs when it was necessary to hand out any type of material to the assistants or to answer any question.An observer attended each workshop together with the subjects and the speakers. Her role was to take notes on the subjects’ behaviour so that, in addition to having the objective movements of the subjects, the appreciations of an independent observer would be registered. Two of the behaviours registered by the observer, along the lines of the diagnostic criteria of the DSM-V and the World Health Organization [22], were the times that the subjects left seat and squirmed in seat. These behaviours were registered in a template that had previously been arranged.The computer application ADHD Movements was developed for the analysis of the movement of several subjects using Kinect. The version of Kinect for Windows allows developers to make their own programs using gestures and body movements in a wide variety of applications, for this reason it is considered an innovative interaction interface between people and computers, in addition to its simplicity in the acquisition of data [20]. This application was developed with Microsoft Visual Studio and using the Kinect for Windows Software Development Kit (SDK) 2.0 that enables the creation of applications that support gesture recognition. For this, one of the main characteristics of this SDK is the detection of the human body that it is represented as a skeleton consisting of segments (such as forearm, hand or foot), which are connected to each other by joints (such as elbow, wrist or ankle). ADHD Movements detects the number of visible skeletons and, to allow the observer to know if the detection is correct, it shows each skeleton and each joint on the screen Figure 3.The application numbers each skeleton (from 1 to 6) in order of detection and saves its location in the environment. In this way, if for any reason, a skeleton stops being detected (for example, a subject stands up and leaves the detection zone), when it returns to its location, the application assigns it the same number.Data obtained by the Kinect sensor includes information of depth measures and, therefore, of the coordinates (x, y, z) of the objects present in the scene. In the case of a human body, it obtains the coordinates of each of the 25 joints that it detects for each skeleton. Each joint is identified by its name and position and it is measured with three coordinates (x, y, z) where (x, y) define the position and (z) represents the distance to the sensor, according to the coordinate space shown in Figure 4.Using this information, the developed application estimates the distance covered by each joint of the skeleton in a three-dimensional space during each minute of the session for each of the subjects. Thus, for each joint, the distance between the different 3D points of the trajectory of the joint in the movement was calculated for each minute of the session.As an example, Figure 5 shows the trajectory of the head joint for one minute of a subject with a firm diagnosis of ADHD (centre), the same subject without medication (right) and a subject in the control group (left).The application obtains, for each subject, a table that shows for each joint the amount distance that it moves in each minute of the session. Sources of this computer program can be downloaded from the software repository of the research group located at https://web.ua.es/en/iamswarm/repository/sw-repository.html.Two Kinect devices were used to record the subjects’ movement data. Each of these devices recorded the data of the complete skeletons of all participants in the session of 30 min. In order to register the information of the session for its later analysis, the application provided by Microsoft, Kinect Studio v2.0, was used. This application allows you to save all the information captured by Kinect, such as depth data, skeletons, sound, etc. for further analysis.The computer application ADHD Movements was used for the analysis of the movement of all the subjects. The application  estimates the distance covered by each joint of the skeleton during each minute of the session for each of the subjects.As it has been previously mentioned, the Kinect device can detect up to 25 joints of 6 different bodies. Figure 6 shows the different joints detected by Kinect and their names.The Kinect device detects 4 joints: wrist, hand, hand tip and thumb from each hand. Kinect uses these joints for the recognition of gestures in a specific situation: that a subject is standing in front of the Kinect, showing the hand to the device.In the case of the study that has been carried out, with the subjects sitting in a natural position, that is, without clearly showing the hand to the device, it has been detected that the joint that is best recognized and registered of the four abovementioned, is the wrist. For this reason, in the results of this study, the hand, hand tip and thumb finger joints have been rejected.Likewise, for the lower limbs, Kinect detects the ankle and foot joints. However, as shown by the research carried out by [23], this device detects skeletons in a standing position much better than in a sitting one. For this reason, after detecting that the recognition and registration of the ankle is much better than that of the foot, the foot has been rejected.Thus, of the 25 joints registered by the Kinect device, only 17 have been analysed for the results of this study.Once all the editions of the study techniques workshops were made, a second observer visualized the data recorded by the Kinect and the times that the subjects leave seat and squirm in seat were registered in a template previously prepared.These data were compared with those previously collected by the observer who attended the different sessions of the workshops. For the study of the concordance of the data of the two observers, the Kappa index was used [24].In order to analyse the movement registered by observers between the experimental group and the control group, as well as to analyse the difference in movement registered by observers between the experimental group with and without taking their prescribed medication, the Student’s t-test was applied and d index proposed by Cohen [24] was included to measure the effect size of the differences found, since the t-test can erroneously detect statistically significant differences. The interpretation of this index is as follows: small effect size (0.20 ≤ d ≤ 0.50), moderate (0.51 ≤ d ≤ 0.80) and large effect size (d > 0.80).In order to analyse the differences between averages of movement of the different joints recorded by Kinect within the experimental group with and without their prescribed medication, receiver operating characteristic curves (ROC curves) have been used in addition to the t-test.The way of interpreting the area under the ROC curve is that a test with an area greater than 0.9 has higher accuracy, while between 0.7 and 0.9 it indicates a moderate accuracy, between 0.5 and 0.7, lower accuracy and 0.5 a chance result (Conzelmann et al. [21]).Statistical packages SPSS 24 and MedCalc 12 were used for the statistical analysis.Results show that, except in the left ankle joint, the subjects with a firm diagnosis of ADHD obtained a higher average in the movement of the joints analysed, than those in the control group. The differences found between the two groups are significant for 14 of the 17 joints (Figure 7). The magnitude of the differences found was small in the joints: spine mid, neck, head, left shoulder, left elbow, left wrist, right shoulder, right elbow, right knee and spine shoulder, oscillating between 0.21 and 0.46 (Table 1). In the joints: spine base, right wrist, left hip, left knee and right hip, the differences found, although statistically significant, did not obtain the minimum levels required by [20] to be considered relevant.The results revealed that there were statistically significant differences for the 17 joints analysed, being in all of them the movement average higher in the group without medication (Figure 8). The magnitude for the differences was small for the joints: spine mid, neck, left shoulder, right shoulder and spine shoulder, oscillating between 0.22 and 0.42 (Table 2). The magnitude for the differences was moderate for the joints: left elbow, left hip, left knee, left ankle, right knee and right ankle, oscillating between 0.56 and 0.79.The magnitude of the differences was high for the joints: spine base, left wrist, right elbow, right wrist, and right hip, oscillating between 0.83 and 1.12.In the head joint, although the difference found between both groups was statistically significant, it did not obtain the minimum levels required by Cohen to be considered relevant [25].In the joints spine base (Figure 9), left elbow, left wrist, right elbow, right wrist, left hip, right hip and right ankle, as shown in respective graphs, the differences in movement were constant in time and maintained over time. In addition, in most of the joints, there were large peaks in which, obviously, the amount of movement of the subjects without medication intake was much higher.All of the above implies an area under the ROC curve with values of 0.79 for the base spine, 0.73 for the left elbow, 0.81 for the left wrist, 0.77 for the right elbow, 0.83 for the right wrist, 0.77 for the left hip, 0.79 for the right hip and 0.72 for the right ankle. For the rest of the joints analysed, except for head, neck and spine shoulder joints, the differences in movement were remarkable in practically all the minutes, but they did not remain constant over time. For the joints head, neck and spine shoulder, although there were peaks in which the amount of movement of the subjects without medication was much higher, at different times the group of subjects with medication recorded more movements than the group without taking medication.Results showed that girls obtained a higher average in movement in comparison with boys for 14 of the 17 joints analysed (Figure 10). Thus, the only joints in which boys obtained a higher average are spine base, right elbow and right wrist.The differences found between the two groups were significant for 11 of the 17 joints (Table 3).This section offers an analysis of the difference in movement registered by observers for the movements left seat and squirmed in their seat. To evaluate the agreement between observers, Cohen’s Kappa index was calculated [24], obtaining a result of κ = 0.93. The results showed that for the two movements registered in the experimental group and in the control group (Table 4), the highest average corresponded to the experimental group, although only in the squirm in seat movement the average difference was significant. In this case, the magnitude of the difference found was moderate: d = 0.69.On the other hand, results revealed that for the two movements registered in the experimental group with and without the medication intake, the highest average corresponded to the group without prescribed medication (Table 5), although only in the movement squirm in seat the difference of averages was significant, with a high magnitude: d = 0.80.On the other hand, results revealed that for the two movements registered in the experimental group with and without the medication intake, the highest average corresponded to the group without prescribed medication (Table 5), although only in the movement squirm in seat the difference of averages was significant, with a high magnitude: d = 0.80.The identification of movement, posture, or a gesture made by a human body in real time is a difficult challenge because it has been found that the human body can perform a great deal of movement and it can have different size etc. [26]. Notwithstanding the above, after having carried out a series of tests with different technologies, the analysis of various technological devices, and the confirmation that Kinect has been previously used successfully in a wide range of research fields [27,28,29,30,31,32,33,34,35,36,37], it has been concluded that the Kinect V.2 is a suitable device to recognize and capture the movement of subjects in a teaching/learning situation, and thus achieve the objectives of this research.Therefore, after having analysed existing technology, the computer application ADHD Movements was developed to detect the number of visible skeletons and, from each of them, analyse and draw on the screen each of the joints. From each joint that makes up all the skeletons, the application estimates the distance it covers in the 3D space during each minute. So, for each joint, the distance between the different 3D points of the trajectory of the joint in the movement has been calculated for each minute of the session.Once the computer application was developed and verified, the movement of the subjects was captured.Results show that there were significant differences in the amount of objective movement between a clinical group of subjects with ADHD and a control group, obtaining a higher average of movement in the experimental group in all the analysed joints (except in the left ankle joint). In addition, the differences found between the two groups are statistically significant in 14 of the 17 joints analysed.In the left ankle joint, the control group obtained a higher movement average than the experimental group, although there was no statistically significant difference. This may be due to the dominance or preference for the use of the right or left leg of the subjects that have been part of the research groups.From the revision of the literature it is concluded that this is the first study that counts the amount of movement in different joints with Kinect in order to find differences between a group of subjects with a firm diagnosis of ADHD and a control group. However, given the importance of hyperactivity in ADHD [7,38], objective measures of movement have been studied for more than a decade, although motion capture has almost always been done with accelerometers (actigraphy and inertial measurement units), with infrared systems placing some device in the body of the subjects or analysing only a part of the body [6,14,15,16,17,18,19].In all the studies reviewed, children with ADHD showed more body movement than those without ADHD [6,14,39,40] so the results of the study coincide. As for the objective movement difference estimated in the group of subjects with ADHD with and without medication intake, the results show that there are statistically significant differences for the 17 joints analysed, being in all of them the average of movement higher in the group without taking medication. The results we have obtained are similar to those of previous studies that indicate that medicated ADHD children show significantly less motor activity than non-medicated ADHD children [41,42,43].Along the lines of a recent study in which the effect of methylphenidate was studied in children with ADHD [44], and in which head movements were not found to be significantly different between the groups (medicated ADHD, non-medicated ADHD and control children), in the study, in the head joint, although the difference found between both groups is statistically significant, it does not obtain the minimum levels required by Cohen to be considered relevant [25].The non-specificity of some diagnostic criteria defined by the DSM-V and the WHO for ADHD, such as squirm in their seat, means that although applications have been developed specifically to estimate the quantity of movement [45], the estimation of this depends on what the observer associates with the gesture. For this reason, both the movement to leave the seat, as well as the one previously mentioned, in the study, were registered by different observers.It is necessary to mention that this work is not exempt from limitations that must be considered when interpreting the results and their implications, as well as in the preparation of future studies in order to solve them and increase the understanding of the findings.Firstly, the number of participants in the experimental group (32) may restrict the ability to detect significant differences between the groups. This limitation must be taken into account nevertheless minimized if we consider that other studies have presented a lower sample size [46,47]. In addition, the experimental group consisted of 24 boys and 8 girls, that is, 75% of the sample of participants with a firm diagnosis of ADHD were male. This data is in accordance with current epidemiological studies that indicate that the prevalence rate is greater in male [48]. However, it limits the extrapolation of the results analysed according to the sex of the participants. For this reason, in order to check whether the results are scalable, it would be interesting for future research to use large samples of subjects.Besides, related to the sample size is the fact that experimental group consisted of 24 subjects with a diagnosis of ADHD with a combined presentation and eight with a predominantly inattentive presentation, according to the DSM-V criteria. In this study we have only analysed hyperactivity as a core feature of ADHD [7,8], together with the conditions of compliance with the diagnostic criteria in order to classify a predominantly inattentive presentation in ADHD that excludes those related to hyperactivity. Therefore, it would be interesting that in the future the objective movement tests carried out in this study could be carried out also in ADHD subjects with a predominant hyperactive/impulsive presentation.Secondly, in this study the Kinect V.2 sensor was used to record the movement of subjects due to its simplicity and effectiveness [20]. In addition, different studies carried out with the Kinect device in several fields show that, although this is not as accurate as some traditional measurement technologies in research laboratories, it provides a good quality relationship for motion tracking systems with respect to the length of the body segments, joint angles and the displacement of the joints in the different body gestures [23]. All the above is not an obstacle to emphasize that the device has a series of limitations that must be considered. Thus, the Kinect device shows excellent results in wide movements such as sitting or standing up, but it shows poor precision in fine or small movements such as closing a hand [49]. This fact has led us to dismiss the results of the hand, hand tip, thumb and foot joints. On the other hand, the Kinect device detects and records the human body better in standing position and in the study the data has been captured in real education/learning situations, where the subjects are sitting. For this reason, it is possible that in certain positions of the subjects (for example, crossing the legs) occlusions have occurred. In these cases, the skeleton detection software of the device interpolates the positions of the undetected joints, being able to produce slight deviations between the real position of a joint and the interpolated position.The results indicate that for the two registered movements, the highest average corresponds to the experimental group with respect to the control group, although only in the movement squirm in seat the difference in averages is statistically significant.In the same way, when comparing the movements registered by the observers between the experimental group with and without the medication, the highest average corresponds to the group without medication. Only in the movement squirm in seat the difference of averages is significant, with a high magnitude (d = 0.80).The absence of significant differences in the movement leave their seat when comparing the experimental group with the control group, may be due to both the “novelty” of the teaching/learning situation (it was the first time they were in that classroom and with those speakers) or to the effect of the medication. However, the results of the comparison of the group with and without medication cannot be due to this reason, since the assistants already knew the speakers and the classroom of the study techniques workshop.In any case, in the absence of previous studies in which a register of this type of movements associated with the diagnostic criteria of DSM-V and WHO for ADHD in a teaching/learning situation, we cannot compare results.The above results lead to a series of conclusions:The software developed (ADHD Movements) for the Microsoft Kinect V.2 device is valid to capture the movement of 17 joints of up to 6 subjects in a teaching/learning situation.Students with ADHD present more movement and squirm more in their seat, than students without ADHD.Students with a firm diagnosis of ADHD without the prescribed medication present more movement and squirm more in their seat than ADHD students with the prescribed medication.ADHD students with and without taking their prescribed medication present a similar amount of movement in the head joint.Girls with ADHD present more movement than boys with ADHD.The software developed (ADHD Movements) for the Microsoft Kinect V.2 device is valid to capture the movement of 17 joints of up to 6 subjects in a teaching/learning situation.Students with ADHD present more movement and squirm more in their seat, than students without ADHD.Students with a firm diagnosis of ADHD without the prescribed medication present more movement and squirm more in their seat than ADHD students with the prescribed medication.ADHD students with and without taking their prescribed medication present a similar amount of movement in the head joint.Girls with ADHD present more movement than boys with ADHD.The work presented here was developed in collaboration amongst all authors. All authors have contributed to, seen and approved the manuscript. All authors have read and agreed to the published version of the manuscript.This work has been supported by the Ministerio de Ciencia, Innovación y Universidades (Spain), project RTI2018-096219-B-I00. Project cofinanced with FEDER funds.The authors of this paper thank the University of Alicante for the grant that allowed part of this research.The authors declare no conflict of interest.Arrangement of Kinect devices in the classroom. Space previously analyzed with BIM (Building Information Modelling) tools.Classroom layout in the study techniques workshops.Screenshot of the application ADHD movements.Coordinate system with respect to the skeleton detected.Trajectory of head joint for a control subject (left), an ADHD subject with medication (centre) and the same ADHD subject without medication (right).Joint map detected by Kinect.Average differences in movement for every joint in the experimental group and the control group. ** = p < 0.001; * = p < 0.05.Average differences in movement for every joint within the experimental group with and without medication intake. ** = p < 0.001; * = p < 0.05.Spine base: movements average/minute and ROC curve.Average differences in movement for every joint in the experimental group, according to sex. ** = p < 0.001; * = p < 0.05.Average differences in movement for every joint in the experimental group and the control rgroup.Average differences in movement for every joint within the experimental group with and without medication intake.Average differences in movement for every joint in the experimental group, according to sex.Average differences in movement registered by observers in experimental group and control group.Average differences in movement for every joint within the experimental group with and without intake of prescribed medication.

Med-MDPI/ijerph_5/ijerph-17-15-05621.txt

@@ -0,0 +1 @@
+This study aims to identify and describe the patterns of shared perspectives of students and supervisory staff associated with their interaction in drug use prevention. We applied the Q methodology to cluster participants into groups according to the similarities of their Q sorts. A total of 31 pairs of students and their supervisory staff participated in the study to rank the designed Q statements. The Q factor analysis for supervisory staff revealed a five-factor solution that accounted for 58% of the total variance. Another five-factor solution for the students explained 49% of the total variance. One similarity between the groups was the need to enhance the involvement of significant others to help the students quit drugs. A major identified difference between the groups was that whereas the students highlighted the importance of health consequences of drug use in helping them stop use, the supervisory staff did not. The elucidation of similarities and differences between supervisory staff and students could offer more insightful strategies of preventing the drug use.Many drugs contain psychoactive substances and toxic chemicals that are detrimental to human beings, particularly adolescents. The use of these drugs by adolescents is associated with poor neurocognitive performance [1] and brain function [2]. Higher levels of drug use during adolescence predict drug abuse and dependence in adulthood [3]. Recently, campus drug use prevention encountered a critical challenge. The use of novel psychoactive substances (NPS) or so-called “legal highs” has grown rapidly [4]. NPS may be labelled “not for human consumption” to avoid regulation. Many NPS are created by minor alterations in the chemical structure of traditional drugs [5]. Such alterations make the NPS users unaware of what they have consumed. As a result, these effects may lead to serious adverse health consequences. Thus, the importance of finding countermeasures for drug prevention among adolescents is urgent.The adolescents who exhibit more knowledge about and positive attitudes toward drug use prevention are more likely to avoid drug use [6,7]. Therefore, providing tailored skills and responsive strategies to help adolescents handle possible situations of drug use can effectively decrease the risks of drug use [8,9]. Since adolescents spend much of their time at school, drug use intervention programs are primarily offered in schools to reach students with drug use problems. The results of a meta-analysis examining school-based drug use prevention programs indicated promising effects of these programs on adolescents with substance abuse problems [10].However, school-based programs targeting students should include strategies to keep students from dropout, which is an essential determinant for the effectiveness of a school-based program. Student dropout from schools often co-occurs with substance use [11]. Empirical studies have recognized that positive youth-adult relationships are beneficial for youth development [12]. Many studies support the association between young people’s positive relationships with adults and improved psychological and behavioral outcomes [13]. Caring adult relationships may be helpful in assisting adolescents to resist the temptation of drug use and in building school bonding, thus preventing students from school dropout. A meta-analytic review of adult mentoring effects on youth delinquency risk found that mentoring interaction had modest positive effects on drug use prevention [14]. A study of examining the effects of youth-adult relationships revealed that program staff with a specific purpose would be even more influential than school teachers [15].While recognizing the students with drug use problems, building supportive relationships between students and school staff would be a possible solution for preventing students from using drugs again. Perspectives that emerged from students and their supervisory staff would enhance the understanding of how youth-adult relationships can be improved. Regarding youth-adult relationships in drug use prevention for students with drug use problems, both students’ and adults’ perceptions of the interaction process are crucial to understand how positive influences can be effectively delivered. The Q method was successfully used to investigate diverse subjective perceptions in multiple health issues [16,17,18,19,20,21,22]. Instead of evaluating each item separately on a Likert-type scale, individuals were expected to prioritize all items in the Q sort procedure, which requires the agreement degree of all items to be simultaneously evaluated and weighed based on their viewpoints [16]. The Q sort procedure employed the holistic approach and offered an alternative to understand youth-adult relationships in drug use prevention. Hence, the study aimed to identify and describe the groups of shared perspectives associated with supervisory staff and students’ experiences in preventing drug use by applying the Q method.In Taiwan, once a student’s drug use is verified by a urine test, a school staff member is assigned to the student and is scheduled to meet with the student at least once per one or two weeks. During the supervisory period, the staff member has a responsibility to assist the student in remaining drug free for at least 12 weeks. In the current study, we defined the staff member as supervisory staff.The study adopted the purposive sampling approach by deliberately choosing a particular group of participants due to the qualities these participants possess. It is a nonprobability sampling technique where the participants are gathered in a process that does not give all the participants equal chances of being included [23]. The inclusion criteria for supervisory staff were those who (I) were assisting students with drug use problems during the study period, (II) had completed a 5-day counseling training workshop, (III) were official school personnel during the study period, and (IV) were willing to complete the study and had signed an informed consent form. The inclusion criteria for students were those who (I) were identified as having drug use problems by urine test, (II) were enrolled in a supervisory program to prevent drug use at the time of the study, (III) had no existing cognitive impairment, and (IV) were willing to participate in the study and had signed an informed consent form.The study recruited supervisory staff from a 5-day counseling training workshop hosted by the Ministry of Education. The workshop aimed to enhance supervisory staff knowledge of the skills required to assist students with drug use problems at vocational and senior high schools. All the attendees (n = 100) were invited and informed that their client students would also be invited to participate in the study. Those attendees who were willing to participate (n = 38) were provided an information letter detailing the study’s purpose and data collection procedure. The supervisory staff whose client students were not available, since the students had dropped out, were on sick leave, or had declined to participate in the study, were excluded. Subsequently, 31 supervisory staff–student pairs agreed to participate in the study. The participant rate was 81.58% (31/38).Through supervisory staff, we could have an opportunity to approach these potential student participants. However, we recruited students directly because of the importance of students’ availability to participate, and willingness to communicate supervisory experiences. Students were contacted by one member of the research team, who explained the study in detail and obtained informed consent from the students. Students were free to reject the invitation and decline to participate in the study. The supervisory staff and students were from six vocational and senior high schools: four in New Taipei City, one in Taichung City, and one in Changhua County. All the participants provided written consent.A structured questionnaire was used to collect participants’ background information. For supervisory staff, data such as their gender, age, duration of services, and duration of employment, were collected. Similarly, data such as gender, age, and status of living with parent were collected from students. Any experience of drug use among students were identified through a urine test. The drugs considered by the study were ketamine, amphetamines, and ecstasy.In general, Q methods comprise two techniques: Q sorting procedures and by-person factor analyses [24]. These are also known as Q factor analyses [25]. Designated Q statements were developed to address the area of interest and enable Q sorting usage. Through Q sorting, participants’ perspectives associated with a specific area of interest were investigated by simultaneously ranking all Q statements on a Q sort grid [17]. Rather than evaluating each statement separately on a Likert-type scale, the participants were expected to prioritize the statements in order according to the Q sorting procedure. They weighed the degree of agreement of each statement in relation to other statements. The results of the Q sorts revealed new insights that might not have been elicited using the traditional Likert-type scaling survey [16].Factor analysis is a technique of data reduction. Application of factor analysis typically leads to the emergence of a number of factors that are used to facilitate a simplified explanation of the associations captured in the correlation matrix yielded from a data set [26]. Traditional factor analysis used the technique to cluster a group of variables (or scale items) into a factor as an alternative manifestation of these variables (or scale items). Therefore, the purpose of data reduction was obtained, and a simplified explanation of data was revealed by factor analysis. On the other hand, Q factor analyses generated clusters of persons, rather than clusters of Q statements, by using the PQ Method (V2.35) [27]. Each resulting final factor represented a group of individuals with similar perspectives. Further, participants were clustered into groups (factors) according to the similarities of their Q sorts [25]. In other words, a group of persons who share a similar perspective emerged. A specific composite Q sort, which hypothetically represented the group perspective, was derived for each of the final factors [28].Finally, the Q methodology uses characteristic statements that are ranked at the extremes of a composite Q sort, which enables the interpretation of each identified group perspective (factor) [17]. Due to the strengths of grouping participants according to their perspective tendency [27], the compared perspectives of the supervisory interaction can be adequately addressed and illustrated.Designated Q statements were provided to address the raised issues related to drug use prevention during supervisory interactions. The statement sentences were developed according to a literature review [6,29,30,31,32,33,34,35,36,37,38,39] and experiences of youth counsellors. The research team discussed the draft statements with two senior youth counsellors; both worked with students of drug use problem for more than 10 years. After several group discussions, the statements were categorized as follows: strengthen information with personal relevance [6,36,37,38], health consequences of drug use [30,32,33], overcome barriers to change [6,28,33,34,35], development of preventive strategies [35,36,37,38], and supportive relationships [29,35,39]. The statements were later reviewed by three health professionals with expertise on drug use to ensure their appropriateness. The areas of academic expertise of the health professionals included health education, nursing, and public health. They were particularly experienced in conducting school-based drug use prevention programs and had published research articles on substance use prevention. Finally, a set of 39 Q statements was compiled (Table 1).The study was conducted after receiving ethical approval. The research team visited the relevant schools to explain and demonstrate to participants how Q sorting was completed. Subsequently, supervisory staff and students met separately in quiet rooms. Only a single participant was allowed in each time slot. All participants completed a paper-based questionnaire to obtain background information before Q sorting.Figure 1 illustrates the user interface for the online Q sort procedure. Participants were provided with a laptop as well as account names and passwords to perform the sorting tasks. They were asked to fill their Q sort in a Q sort grid, generating a distribution of degrees from +4 (strongly agree) to −4 (strongly disagree) statements. The Q sorting helped to clarify the participants’ subjective opinions on these statements.The Q sort task required the participants to consider the statements according to their experiences with drug use prevention. Participants were asked, “In order to prevent drug use, are the following statements important to you during the supervisory interactions?” We advised the participants to initially divide the Q statement into three groups (positive, negative, or neural). We provided a printed list of Q statements so that each statement could be marked as positive or negative. If a statement could not be classified as positive or negative, then it was classified as neutral. When complete, participants refined the positive-ranked statement group with degrees from +4 to +1 and assigned each statement to cells in the grid based on the number of degrees determined. Participants used the mouse to move each statement from the left panel to right panel in the Q sort grid until all positive rank statements were placed. Participants then repeated the process on the left side of the Q sort grid for the negative statements. If the positive and negative order cells were not filled, participants selected statements from the neutral group and placed them in empty cells. The remaining statements were placed in the middle column which was ranked “0”. Participants could adjust the position of statements repeatedly until they felt comfortable.Participants (supervisory staff and students) received a coupon worth NTD$200 as an appreciation for their provision of Q sorts.PQ Method 2.35 was used to analyze the data collected from Q sorts [39]. The PQ Method software is a statistical program dedicated to the statistical analysis of Q studies (http://schmolck.org/qmethod/). Electronic data from the participants’ online Q sorts were imported to PQ Method software for analysis. Q factor analysis was applied to constitute supervisory staff and student groups based on the similarities in Q sorts. Further, a scree plot was employed to determine the number of retained factors. Finally, Q factor analyses were separately performed for supervisory staff and students.In this study, sample size estimation was performed according to two rules [25]. First, there was one participant for every three Q statements. Second, it was suggested that at least three participants load highly on each perspective. According to the first rule, there were 39 Q statements, and at least 13 participants were expected for each group of supervisory staff and students. According to the second rule, the study revealed five perspectives, due to which at least 15 participants were included in each group of supervisory staff and students. However, it was impossible to know in advance the number of perspectives that would be the best solution for the Q factor analysis. Therefore, 31 participants in each group was considered sufficient because the suggested number was doubled (15 × 2 = 30) according to the aforementioned calculation.As shown in Table 2, supervisory staff included 23 men and eight women. Their mean age was 38.61 years (SD = 5.60). The duration of services as a supervisory staff and employment was 3.07 and 8.75 years, respectively. Similarly, the majority (85.71%) of the students were male students. Most (77.42%) were older than 18 years, with a mean age of 18.26 years (SD = 1.46). More than half (58.06%) of the students lived with both parents.Q factor analyses were separately performed for supervisory staff and students. Following analysis, five-factor solutions with eigenvalues above 1 were extracted for both supervisory staff and students. The result of the five-factor solution for supervisory staff accounted for 58% of the total variance. The supervisory staff were clustered into five groups (five-factor solution) according to the similarities of their Q sorts. Another five-factor solution represented five groups of students with similar perspectives. This explained 49% of the total variance.Q factor analyses generated clusters of individuals with similar perspectives. Hence, we used the term “groups” to indicate the factors resulting from Q factor analyses. Composite Q sorts of the resulting factors were used to interpret identified group perspectives. The characterizing statements, ranked at the most positive ends of each composite Q sort (+3 and +4), were used to illustrate perspective patterns of the participants who significantly loaded onto the specific group. Within each group, there were five characterizing statements, including three +3 and two +4 statements. A total of 25 characterizing statements were revealed for each group of supervisory staff and students (Table 3 andTable 4). We applied a radar chart to obtain an overview of the similarities and differences between supervisory staff and student groups (Figure 2). The marks in the figure represent the total numbers of characterizing statements for each category and indicate the degree to which the categories were endorsed by the supervisory staff and students. The farther the mark from the center, the more evident the perspectives associated with the respective category.According to Figure 2, the most evident perspective was associated with “Supportive Relationship,” where eight and seven characterizing statements were located for supervisory staff and students, respectively. The least evident perspective was associated with “Strengthen Information with Personal Relevance,” where two and three characterizing statements were located for the supervisory staff and students, respectively. The different perspectives between supervisory staff and students were associated with “Health Consequences of Drug Use,” where only one characterizing statement was located for supervisory staff, but six characterizing statements for students.The Q factor analysis revealed five groups of supervisory staff perspectives. These perspectives are discussed in the following subsections.Supervisory staff group 1’s perspectives included “Enabling students to become aware of the influences of friends with drug use problems” (+4), “Not hanging out with friends who have drug use problems” (+3), “Avoiding situations that may increase risks of using drugs” (+3), “Strategies to make friends who are positive influences” (+4), and “Enhancing the involvement of significant others to help students quit drugs” (+3). Five participants significantly loaded on this group.Supervisory staff group 2 agreed with the importance of “Working with students to effectively manage the underlying reasons for their drug use” (+3), “Strategies to make friends with positive influences” (+3), and “Enhancing the involvement of significant others to help students quit drugs” (+4). Further, they agreed with the importance of “Provision of examples related to drug use” (+3) and “Enabling students to become aware of the influences of friends with drug use problems” (+4). Nine participants significantly loaded on this group.Supervisory staff group 3 agreed with the importance of “Strategies to strengthen willpower to resist temptation to use drugs” (+3), “Management for preventing relapse, especially strategies for coping with low mood” (+3), and “Avoiding situations that may increase chances of using drugs” (+3). Further, they agreed with the “Provision of examples related to drug use” (+4) and “Working with students to effectively manage the underlying reasons for their drug use” (+4). Six participants significantly loaded on this group.Unlike group 3, supervisory staff group 4 expressed perspectives to overcome barriers of change. The characterizing statements were “Enabling students to become aware of the influences of friends with drug use problems” (+4), “Resisting a boy/girlfriend’s influences to use drugs” (+3), and “Not hanging out with friends who have drug use problems” (+3). They agreed with the importance of “The effect of drug use on behaviors” (+3) and “Enhancing the involvement of significant others to help students quit drugs” (+4), as well. Three participants significantly loaded on this group.The characterizing statements of supervisory staff group 5 included “Management for preventing relapse, particularly strategies for coping with low mood” (+3), “Management of physical dependence” (+4), and “Management of mental dependence” (+4). They agreed with “Enabling students to become aware of the influences of friends with drug use problems” (+3) and “Working with students to effectively manage the underlying reasons for their drug use” (+4), as well. Three participants significantly loaded on this group.Among students, five groups of perspectives emerged from the Q factor analysis. These perspectives are discussed in detail in the following subsections.Student group 1 emphasized statements such as “Provision of examples related to drug use” (+3), “A brave person resists the temptation to use drugs” (3), and “The effect of drug use on daily activities” (+4). Student group 1 agreed with “Educational needs of assisting students to find purpose in life” (+4) and “Effectively utilizing school resources for drug use prevention” (+3), as well. Seven participants significantly loaded on this group.The students associated with Group 2 agreed with the importance of “Not hanging out with friends who have drug use problems” (+3), “Avoiding situations that may increase chances of using drugs” (+3), “Educational needs of leisure skill building and developing ability to engage in appropriate activities” (+3), “Strategies to make friends who are positive influences” (+4), and “Enhancing the involvement of significant others to help students quit drugs” (+4). Nine participants significantly loaded on this group.Student group 3 emphasized the health consequences of drug use. The characterizing statements included “The effect of drug use on sleep” (+3), “… on mental health” (+3), “… on daily activities” (+3), and “… on behaviors” (+4). They agreed with the statement “Strategies to convince my friends to participate in drug rehabilitation program with me” (+4), as well. Five participants significantly loaded on this group.Student group 4 emphasized statements such as “Resistance to the immediate benefits of drug use, such as excitement, pleasure, euphoria” (+4); “Resistance to the peer pressure to use drugs” (+3); and “Not hanging out with friends who have drug use problems” (+3). They agreed on “Encouraging self-determination to quit drugs forever” (+4) and “The effect of drug use on mental health” (+3), as well. Four participants significantly loaded on this group.Student group 5 agreed with the importance of “Resistance to the dependency on drug use” (+3), “Educational needs of assisting students to find purpose in life and enabling students to show potential” (+3), “Development of school bonding by enhancing regular school attendance” (+3), “Strategies to interact with the person who knows my history of drug use” (+4) and “Enhancing the involvement of significant others to help students quit drugs” (+4). Three participants significantly loaded on this group.The results of the analyses showed agreement on the importance of supportive relationships, for example, enhancing the involvement of significant others to help an individual quit drugs, for both supervisory staff and student groups. Earlier studies have revealed that the perceived lack of support from family and friends is a pivotal factor in adolescents’ risk of relapse [40]. One difference between the supervisory staff and student groups was that whereas students highlighted the importance of health consequences of drug use, supervisory staff did not. Introducing the health consequences of drug use frequently has been applied as a motivation strategy for behavior change [41]. An enhanced perception of the negative outcomes of drug use can elicit discrepancies between individuals’ behaviors and goals and motivate them to make behavioral changes [10,42]. Unlike students, supervisory staff may feel that health hazards were knowledge-based information and had less impact on behavior change when compared to fostering positive attitudes or refusal skills. This difference between supervisory staff and students warrants further investigation.Another interesting difference was that three Q statements were endorsed by more than half of the supervisory staff groups, whereas none were endorsed by more than half of the student groups. A possible reason is that supervisory staff training/supervising experience generated some degree of consensus across various supervisory staff groups. The staff emphasized the importance of interpersonal influence during interaction, as well. They strongly agreed that an alliance between supervisory staff and students could create a climate of trust and establish a necessary supportive relationship [43]. Supervisory staff firmly believe the key elements of provision of companionship, support, and guidance [14], regardless of the groups that they loaded onto.Our findings revealed variabilities in supervisory practices. An earlier study revealed that adult perceptions influenced their practices of mentoring [44]. The perspectives of some staff, who loaded onto groups 1 and 4, were in line with earlier research in that peer influences were one of the most significant contributors to drug use among adolescents [6,45,46]. These staff emphasized external influences during supervision, whereas others prioritized internal influence. The staff loading onto groups 3 and 5 recognized the importance of assisting students to develop preventive strategies during supervision, such as coping with low mood [35,36,37] and management of physiological and psychological dependence [30,32,33].Students who joined a supervisory program were expected to stay drug free; however, not all could fulfill this expectation [47]. The students associated with groups 1 and 3 emphasized the health consequences of drug use. In addition, self-encouragement to avoid the temptation to use drugs was emphasized. Examples could offer students opportunities to explore the pros and cons of behavioral consequences [8,48].Students loading onto groups 2, 4, and 5 emphasized involving significant others, people who know the students’ history of drug use, and encouraging regular school attendance. Earlier studies have suggested that avoiding classes is significantly associated with drug use among vocational and senior high school students [45]. Our findings strengthen the importance of not only peer pressure avoidance but also encouragement of school attendance during supervision to prevent drug use. Students emphasized the development of an ability to engage in appropriate activities and a sense of purpose in life, as well. An individual’s purpose in life was proposed as a spiritual mechanism that contributed to his or her recovery from substance abuse and dependence [49].The current study has some limitations. In this study, strategies for assuring accuracy of Q sorting were that participants were asked to approve their final Q sort and were allowed to adjust the order of each statement if they felt it necessary. However, participants expressed that qualitative information associated with their Q sort would enhance appropriate interpretation of the study findings. A lack of qualitative data may decrease the degree of revealing in-depth participant thoughts. Further study is encouraged to include collection of qualitative information during the Q sort process. In addition, providing a comparison of gender differences would advance the understanding of the influence of youth-adult interaction. However, there were only three female students in the current study. Gender comparison is encouraged in studies with sufficient female sample size. Further, our study considered only in-campus, rather in-community, participants. All the participants were students of vocational or senior high schools. The adolescents who drop out of school may have different perspectives regarding drug use prevention than those attending schools. Hence, we suggest that future studies conduct a Q methodological survey among a sample of adolescents that includes those within the community.The novel approach of using the Q sort methodology in the study helped uncover significant similarities and differences between the perceptions of supervisory staff and students. Regarding youth-adult interaction for drug use prevention, students highlighted the importance of the health consequences of drug use, whereas supervisory staff did not. Both students and supervisory staff emphasized that enhancing the involvement of significant others helps students quit drugs. In addition to involving significant others, supervisory staff emphasized the importance of recognizing peer influences during supervision. Further, the students pointed out that interactions with people who knew their history of drug use were important during supervision. Finally, supervisory staff emphasized the development of preventive strategies, whereas students emphasized enhancing their regular school attendance.An examination of the similarities and differences between supervisory staff and students’ perceptions of supervisory interaction clarified how to support students in preventing drug use. While preventing students from drug use, enhancing the involvement of significant others is suggested as an approach to help students quit drugs. To gain insight into the effectiveness of youth-adult interactions, the study’s findings suggested a need to extend the dyadic relationships to include the students’ social contexts [44].Supervisory staff practices vary according to whether they emphasize internal or external influences. Not all students who joined the supervisory program were ready to change themselves. In addition to fostering positive attitudes or refusal skills, a careful assessment of students’ needs is important for appropriately advising and supervising students for drug use prevention. Students who were not ready to change themselves required information on the health consequences of drug use [42], for example, statements advocating the value of change may motivate students to change their behaviors [50].Some students emphasized the importance of promoting regular school attendance and interacting with people who know their history of drug use. However, supervisory staff did not recognize the importance of these two statements. Accordingly, counseling training programs or workshops might find the study’s findings helpful and emphasize the significance of conducting priority assessments of students. In addition to assessing students’ needs, understanding their priorities contributed to the establishment of effective adult-youth relationships in drug use prevention.Conceptualization and Design: C.-M.H. and J.-L.G. Formal analysis and Investigation: C.-M.H.; J.-L.G. and J.-Y.L. Project administration and Resources and Validation: C.-M.H.; J.-L.G.; J.-Y.L.; H.-P.H. and C.-Y.L. Visualization: C.-M.H. and J.-Y.L. Writing—original draft: C.-M.H.; J.-L.G.; J.-Y.L.; H.-P.H. and C.-Y.L. Writing—review and editing, C.-M.H. and J.-L.G. All authors have read and agreed to the published version of the manuscript.This work was supported by a grant from the Ministry of Science and Technology, Taiwan (MOST 104-2511-S-003-023-MY2).We express our gratitude to the participants and six schools that participated in this study. This article was subsidized by the National Taiwan Normal University (NTNU), Taiwan, ROC.None declared.Illustration of online Q sort procedure. The content of this interface was translated into English; the original interface was in Chinese.A glance at the similarities and differences between the perspectives of supervisory staff and supervisory staff by using a radar chart. The marks represent the number of characterizing statements for each category and indicated the degree of categories which were endorsed by the supervisory staff and students.List of the Q statements.Background information of the supervisory staff and students.Characterizing statements—supervisory staff.Note: a Total numbers of statements in each category. The numbers (4) and (3) in parentheses represent that the statements most accurately reflected the experience of participants who loaded significantly onto the given group.Characterizing statements—students.Note: a Total numbers of statements in each category. The numbers (4) and (3) in parentheses represent that the statements most accurately reflected the experience of participants who loaded significantly onto the given group.

Med-MDPI/ijerph_5/ijerph-17-15-05622.txt

@@ -0,0 +1 @@
+Walkability has been associated with urban development and political plans, contributing to more connected cities with improvements in communication, shopping, and pedestrian base. Among these services, fitness centers are becoming important elements for communities due to their impact on the health and welfare of citizens. The present study aims to examine how an area’s Walk Score® affects fitness center services, specifically membership costs, opening hours, and aquatic services. Data from 193 fitness centers were retrieved, representing all the areas of the municipality of Madrid, Spain, including fitness centers in the 21 city districts. A nonlinear relationship between an area’s Walk Score® and fitness centers’ monthly fees is observed. Only in premium fitness centers, a weak curvilinear model is observed, following a quadratic equation, showing that fitness centers with higher prices are in less walkable areas. Additionally, the association between Walk Score® and a fitness center’s opening hours reveals that fitness centers with wider hours of operation tend to be in moderately to highly walkable locations. Lastly, the existence of a swimming pool is related to a lower Walk Score®. Thus, fitness centers in less walkable areas try to offer additional services as differentiation from competitors, whereas centers in walkable locations use this advantage as a strength.The concept of walkability is defined as “the extent to which the built environment is friendly to people who walk to work, for leisure or recreation, to travel, for exercise, or to access services” [1]. Overall, it can be broadly understood as the extent to which an area, usually an urban area, is walking-friendly [2]. Proven benefits of living in walkable environments include a healthier lifestyle [3,4] and less polluted and congested streets [5,6] as well as economic benefits [7]. This evidence has led to an increase in research on the design of walkable cities [1]. Following previous work, there are some urban elements that are central for walkability. These are the connectivity of the path network, linkage with other modes of transport (bus, subway, or train), safety from both traffic and social crime, quality of path (width, paving, or signing) and path context (street design, visual interest, or landscape), and varied land use patterns (reaching most local services on foot within 10–20 min, including uses such as shops, cafes, banks, laundries, grocery stores, parks, or fitness centers) [5]. Thus, walkability is tightly associated with urban development and political plans, contributing to a more connected city with improvements in communication, services, shopping, and pedestrian base [8].City services have been widely analyzed in recent years in relation to the concept of smart cities [9], user information and communication technologies [10], and the habits of individuals based on their relationship with the environment [11,12]. Among the different approaches, it is not easy to find a variable to measure the quality of life of citizens tightly related to the services that cities offer—with the exception of satisfaction surveys and consumption habits. Different perspectives on walkability also lead to diverse focuses [13]. Consequently, a wide range of quantitative and qualitative tools have been used to assess the walkability of urban environments. Some examples are the Pedestrian Environment Review System (PERS), Pedestrian Level of Service (POS), or Geographic Information Systems (GIS). In view of the range of studies, diverse issues in measuring walkability have also emerged [14,15], including trip purpose, temporal issues (different times of day), walking barriers, and the perceived quality of walkable spaces. These issues highlight the difficulties in assessing walkability. In this regard, following Vale et al. [15], there are four main methodological categories for assessing walking accessibility: distance-based, gravity-based, topological or infrastructure-based, and walkability or walk score-type. This research focuses on this last approach.Since its creation in 2007, Walk Score® has been one of the most widely used methods worldwide for walkability assessment. Walk Score® is a United States-based company that provides walkability services and apartment search tools through a website and mobile applications. Walk Score® uses data provided by the Google™ AJAX Search application program interface (API) through a geography-based algorithm [16]. The Walk Score® algorithm calculates a score of walkability based on the distance to 13 categories of amenities (grocery stores, coffee shops, restaurants, bars, movie theaters, schools, parks, libraries, book stores, fitness centers, drug stores, hardware stores, clothing/music stores). Each category is weighted equally, and points are summed and normalized to yield a score of 0–100 [17].The score has been validated by the scientific literature as a reliable way to measure access to walkable amenities mainly in the United States [18,19], and its use is being extended to other regions, including Europe [14,20,21] and Asia [22,23]. However, some researchers claim that Walk Score® and other applications do not replace conventional street network measures but are complementary [24]. A recent systematic review indicates that the analysis of walkability using Walk Score® is inconsistent due to research results tending to only partly support the validity of Walk Score® [2]. Despite the criticism, the research community considers that there is no reason to believe that Walk Score® is substantially different than in the United States-based studies [20,21], and it is valid in high population density urban areas [19].In addition, Walk Score® has been used in the scientific literature to evaluate physical activity, health status, and sedentary behavior [20,25,26], tourism accommodation and services [14,23], eating habits [27], and walking and gaming mobile apps use [28]. Nevertheless, there is no specific research about Walk Score® and fitness centers, considering that gyms and sports centers are part of the Walk Score® algorithm. This knowledge gap in the literature regarding fitness centers should be addressed, since these sports services are fundamental for active lifestyle promotion and have important impacts on health [29,30]. Furthermore, it is difficult to find previous research that links Walk Score® with important service variables such as price and opening hours. As an example, other authors have found how Walk Score® could affect pricing in other tourist services [23]. For that reason, this paper tries to answer the research question: “How does location affect fitness center services?”.In this vein, several authors have raised the importance of the evaluation of fitness services, meaning a detailed analysis of the provided service that contributes to making managerial actions more precise. Additionally, authors have examined specific business models based on price [31] or the analysis of the importance and performance of different services and management decisions [32]. For this purpose, after the literature review, a possible relationship between the Walk Score® and different variables such as prices, opening hours, and specific services at fitness centers is considered, depending on the city district and its walkability. Traditionally, price has been identified as one of the main reasons to enroll in a fitness center [32]. Moreover, opening hours are crucial, even more so in a big urban area such as a capital city. Additionally, extra services such as aquatic services contribute to attracting more people to the centers. Thus, this study aims to examine how the location, measured by Walk Score®, affects fitness center services through different variables such as membership costs, opening hours, and aquatic services in fitness centers in the municipality of Madrid.The sample is composed of 193 fitness centers (179 private, 14 public) located in the city of Madrid, Spain, covering the 21 districts of the municipality (Figure 1). Madrid, located in the center of the Iberian Peninsula, is the capital city of Spain and has a population of almost 3.3 million inhabitants. Madrid has a land area over 600 square kilometers.The selected fitness centers were retrieved from Madrid Council’s Open Data Portal, which allows access to a complete database of the business census of the city. The June 2019 data package was used [33]. Several stages were addressed before reaching the final sample of 193 centers. Firstly, the complete business census was filtered by activity, obtaining the number of 730 sports centers, thus eliminating non-sports businesses. Secondly, those oriented to fitness activities were selected, as well as those currently inactive were deleted, obtaining 252 centers. Thirdly, for each of them, the monthly fee, business model (low-cost ≤ 30€; mid-market = 30€–60€; premium ≥ 60€), and the existence or not of a swimming pool were checked. Any center with a lack of any of the aforementioned information was deleted from the database, resulting in a final number of 193 fitness centers. Lastly, the Walk Score® for each center location was manually derived from the website www.walkscore.com, using the exact address of the company [34].For the analysis of the data, Walk Score® was used as a categorical variable, instead of continuous, as recommended by previous work [23,35]. Therefore, four quartiles were established (≤90; 91–95; 96–98; ≥99). The data analysis was performed with IBM SPSS 23.0 Statistics software (IBM Inc., Chicago, IL, USA), including descriptive statistics, t-test, Mann–Whitney U test, and quadratic regression, as suggested in previous research [14,23]. The critical level of significance was set at p < 0.05.Data from 193 fitness centers were retrieved (Figure 2). A representative distribution in all the areas of the municipality was ensured, including centers in the 21 districts of the city of Madrid, Spain (Figure 3).Table 1 shows the means of Walk Score®, monthly fee, and opening hours. These results are presented according to the business model (low-cost, mid-market, and premium). An average Walk Score® of 92.02 (SD = 9.02) out of 100 was determined. The average fee per month was 41.46 (SD = 28.70), and the average daily opening hours were 15.36 (SD = 3.16). Additionally, 20.20% of the centers have a swimming pool.A nonlinear relationship between Walk Score® and the monthly fee was obtained. According to the business model, no correlation was observed in either the low-cost or mid-market centers. Nevertheless, after measuring a normal distribution of data, the t-test only established statistically significant differences in premium centers, with lower fees in the range of 96–98 in contrast to a ≤90 Walk Score® (Table 2). These data are presented graphically in Figure 4. No statistically significant differences were observed between other Walk Score® ranges.A regression analysis of Walk Score® was conducted, following a curve estimation procedure. Different curves were estimated, namely, linear, quadratic, cubic, growth, and exponential. The quadratic model received the highest R-square (Table 3). Nevertheless, the analysis reveals a weak relationship between Walk Score® and the monthly fee (R2 = 0.05, F (2, 190) = 5.27, p = 0.006). Therefore, medium levels of Walk Score® (96–98) weakly contribute to lower fees.The Mann–Whitney U test was performed to compare the average opening hours between the four Walk Score® groups (Table 4). The results show statistically significant differences between the ranges of ≤90 and 96–98, as well as between the groups of 91–95 and 96–98. Therefore, medium levels of Walk Score®, specifically between 96 and 98, are associated with wider opening hours.The Mann–Whitney U test was also conducted to compare the average Walk Score® between fitness centers with a swimming pool and centers lacking a swimming pool. The results show statistically significant differences when comparing both clusters. Therefore, a lower Walk Score® is associated with swimming pools in the fitness centers (Table 5).Walkability is an important concept in urban planning, with great implications for the population, since walkable places are usually related to economic performance, including real estate development and values as a result of their attractiveness to permanent and temporary populations [7,36]. This paper deals with an innovative approach for measuring the relationship between walkability and specific variables of fitness centers (price, opening hours, and aquatic services) as fundamental services for physical activity promotion, which help to improve citizens’ health.Firstly, regarding the association between Walk Score® and the monthly fee of fitness centers, a weak nonlinear relationship was observed, meaning that there is not a strong association between the two variables. These results show coherence with previous work on tourist attractions and walkability [14]. However, our results contrast with research on tourist accommodations. Although hotels and AirBnB locations also showed a nonlinear relationship, the range of 93–96 Walk Score® displays higher prices [23]. This weak link between walkability and price has implications for sports centers and other leisure-oriented businesses.For the aforementioned reasons, there is no direct correlation between fitness centers with a higher Walk Score® and higher fees. Only in premium centers, a certain degree of association is proven. When analyzing premium centers, a quadratic equation curve fits in the model, as was presented in previous research [14]. Thus, fitness centers with higher fees are often located in less walkable locations. This circumstance may be related to the specific features of the fitness industry and the diverse business models, together with the preferences and needs of the users of sports centers, since as was observed in previous studies, price and location are among the criteria of greatest weight for users in Spain [37], the United States, and Canada [38]. Especially for fitness centers with a low-cost model, service convenience (where walkability would be integrated) is tightly related to user perceived quality, user satisfaction, and client loyalty [39]. In fact, regarding fitness centers and their locations, previous studies have shown that clients are willing to commit to even a 30 min commute time to the facility if the perceived quality is better [39], while a 15 min commute time positively influences the client’s adherence to the center, leading to the client’s involvement in a longer membership and longer member continuity [37,40,41]. This is especially important for increasing levels of physical activity and, subsequently, impacts citizens’ health.A good explanation for why users are willing to walk longer or to travel longer distances for premium services could be related to perceived service quality and added value. Previous studies have proven that the enjoyment of positive customer experience is associated with higher engagement levels, recommendation rates, and membership renewal intentions, with positive effects on all these variables [42,43,44,45]. For that reason, customers of premium fitness centers could be willing to walk longer distances or travel to less walk-friendly areas. This phenomenon has also been detected in the tourism sector [2,46] since tourists are willing to walk longer distances to locations with a lower Walk Score® in order to enjoy the most popular tourist attractions.Secondly, regarding the relationship between opening hours and Walk Score®, our findings show that a moderate to high Walk Score® (96–98) correlates with longer opening hours. There is no direct correlation between fitness centers with a higher Walk Score® and wider opening hours. Good management practices encourage sport facilities managers to reinforce their company strengths [32]. Therefore, for fitness centers whose location is considered a strength, maximizing and enlarging opening hours contributes to maximizing this strength. Identical to the comparison between Walk Score® and price, low-cost centers need to pursue high service convenience, where opening hours may be influential, achieving ultimately a good perceived quality, user satisfaction, and client loyalty [38].Lastly, regarding the relationship between the existence of a swimming pool at the gyms and Walk Score®, our findings show that a lower Walk Score® correlates with the existence of aquatic services. In this regard, the ability to enjoy guided and free aquatic physical activities is an attractive aspect for gym users, especially affecting client satisfaction in the case of group class swimmers and future intentions in the case of free swimming users [47]. It has been shown that a swimming pool is not considered essential, but influential and attractive for customers at the moment of gym enrolment, even if swimming pool use rates are low [37]. However, the swimming pool is not always a profitable space, especially for fitness centers whose main focus is not related to aquatic activities [40]; therefore, gyms located in better locations (i.e., with a more expensive rental or surface fee) prefer not to dedicate a big surface area to these aquatic services.This paper contributes to understanding the association between the geographical distribution of fitness centers and the variables of price, opening hours, and aquatic services. Firstly, no direct correlation is shown between Walk Score® and monthly fee. A weak quadratic model is followed only by premium centers, with higher prices in less walkable areas, meaning that members of these centers are willing to travel to less walk-friendly areas. Secondly, the association between Walk Score® and opening hours is not totally confirmed. Fitness centers in moderately to highly walkable locations tend to widen their average hours of operation, but there is not a direct correlation. Thirdly, the existence of a swimming pool is associated with a lower Walk Score®. In this regard, fitness centers in less walkable areas try to offer additional services such as aquatic services to differentiate themselves from better located competitors. However, centers in walkable locations do not need to invest in swimming pools and prefer to use this area for other purposes. All these data are relevant for increasing the adherence to fitness services. A better tailored experience would help to promote physical activity services participation and engagement.This research has clear practical implications, mainly for managers of sports services. Business location is particularly important for increasing levels of adherence to fitness centers, helping to increase levels of physical activity. Understanding the geographical distribution of fitness centers would help them to tailor the offer to potential customers. Walk Score® can be a useful open resource for managers of sports services at the time of deciding the best location and fitness center characteristics.However, this work deals with the limitation of the local features of the municipality of Madrid. The urban distribution of Madrid shares many characteristics with other European capitals. However, our conclusions could be limited by specific local conditions and particular issues of Madrid. Nevertheless, there is no evidence in recent previous literature [19,20,21] to believe that Walk Score® substantially differs outside the United States, mainly in high population density areas. Despite the applied approach, we recognize that there are other approaches, especially regarding spacial autocorrelation. Future studies should address different cities for further examination of fitness center parameters’ geographical distribution. Additionally, more extensive areas could be assessed, including both urban and rural spaces.Conceptualization, J.L.-Q.; Data curation, J.L.-Q. and P.B.; Formal analysis, J.L.-Q. and J.B.; Methodology J.L.-Q. and Á.F.-L.; Writing—original draft preparation, J.L.-Q., J.B., and Á.F.-L.; Writing—review and editing, P.B. and Á.F.-L. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The authors declare no conflict of interest.Districts in the city of Madrid: (1) Centro; (2) Arganzuela; (3) Retiro; (4) Salamanca; (5) Chamartín; (6) Tetuán; (7) Chamberí; (8) Fuencarral-El Pardo; (9) Moncloa-Aravaca; (10) Latina; (11) Carabanchel; (12) Usera; (13) Puente de Vallecas; (14) Moratalaz; (15) Ciudad Lineal; (16) Hortaleza; (17) Villaverde; (18) Villa de Vallecas; (19) Vicálvaro; (20) San Blas-Canillejas; (21) Barajas.Distribution of the fitness centers in Madrid, sized by Walk Score®.Fitness centers’ distribution by district.Monthly fee according to Walk Score®.Descriptive statistics.Monthly fee t-test in premium centers according to Walk Score®.Regression curve estimation.Opening hours comparison according to Walk Score® groups.Walk Score® comparison in fitness centers with and without a swimming pool.

Med-MDPI/ijerph_5/ijerph-17-15-05623.txt

@@ -0,0 +1,9 @@
+A consecutive crash series is composed by a primary crash and one or more subsequent secondary crashes that occur immediately within a certain distance. The crash mechanism of a consecutive crash series is distinctive, as it is different from common primary and secondary crashes mainly caused by queuing effects and chain-reaction crashes that involve multiple collisions in one crash. It commonly affects a large area of road space and possibly causes congestions and significant delays in evacuation and clearance. This study identified the influential factors determining the severity of primary and secondary crashes in a consecutive crash series. Basic, random-effects, random-parameters, and two-level binary logistic regression models were established based on crash data collected on the freeway network of Guizhou Province, China in 2018, of which 349 were identified as consecutive crashes. According to the model performance metrics, the two-level logistic model outperformed the other three models. On the crash level, double-vehicle primary crash had a negative association with the severity of secondary consecutive crashes, and the involvement of trucks in the secondary consecutive crash had a positive contribution to its crash severity. On a road segment level, speed limit, traffic volume, tunnel, and extreme weather conditions such as rainy and cloudy days had positive effects on consecutive crash severity, while the number of lanes was negatively associated with consecutive crash severity. Policy suggestions are made to alleviate the severity of consecutive crashes by reminding the drivers with real-time potential hazards of severe consecutive crashes and providing educative programs to specific groups of drivers.With the increasing number of traffic crashes occurring in the past decade, both researchers and practitioners in the road safety field have been focusing on topics such as crash risks, severity levels, and safety behaviors to propose prevention measures for alleviating social and economic losses [1,2,3,4,5,6,7]. Some scholars are focused on investigating the risk factors and severity levels of crashes in specific scenarios, as the mechanisms of crashes may vary considerably in different situations, e.g., with different traffic conditions, involving different numbers of vehicles, or caused by different at-fault driver groups. Secondary crashes, defined as the crashes occurring spatially and temporally within the impact area of a preceding crash, have always been considered a special crash scenario. They commonly occur as a consequence of the primary crash and may possibly cause greater burdens to the traffic system, as multiple crashes may severely block the roads and causes congestion, hindering victim evacuation and road clearance activities.The occurrence of secondary crashes has been the main focus of previous research on this topic. Some studies have explored the occurrence of such type of crashes and its contributory factors, most of which incorporated a logistic regression to model the probability of occurrence [8,9,10], and some have adopted variants of logistic regression for addressing specific modeling issues [11,12]. Specifically, Xu et al. [11] have proven the existence of unobserved heterogeneity by adopting a random-effects logit model to discover the contributions of various factors on secondary crash propensity. Common understanding has formed that the occurrence of secondary crashes is significantly dependent on the duration of the primary crash, queue length, traffic condition, and road geometric design [13]. Various identification methods for secondary crashes applied in previous studies, including static methods with fixed spatial and temporal thresholds (usually high in numbers), and dynamic methods such as shock wave and speed contour methods, are both based on the assumption that congestion is incurred by the primary crash given a certain period of time [9,13].Conceptually, secondary crashes are induced by queuing effects caused by primary crashes, and thus the effective area of the primary crash can be relatively large in time and space. In previous research on secondary crashes with static identification methods, the distance between the primary and secondary crashes are usually longer than 2 miles and the time gaps are larger than 2 h [14,15]. With this assumption, the traffic hazards produced by the primary crash are cumulative in time and space and spread through the queuing downstream vehicles.There is a special type of immediate secondary crash that occurs in a considerably short period of time within a vicinity of the primary crash hasn’t gained enough attention in previous research. When the secondary crash happens shortly after the primary crash, traffic queues haven’t formed and the clearance of primary crash doesn’t apply. In this case, congestion is no longer a factor that induces the subsequent secondary crash, and the traffic hazards produced by the primary crash transmit mainly through vehicular interactions at a near-free-flow speed, similar to the mechanism of a chain-reaction (multiple-vehicle) crash. In this paper, the aforementioned primary crash and the subsequent immediate secondary crash is named as a consecutive crash (CC) series. One primary crash is possible to induce one or more secondary CCs. By definition, a CC series is different from a chain-reaction crash in that the latter is considered as one single crash with multiple vehicles involved as a result, and can normally be decomposed into several collisions, but a CC series is a group of multiple crashes, each of which can involve single or multiple vehicles.As many similarities between CCs and chain-reaction crashes exist, the crash mechanisms of the two can be analogized. Chain-reaction crashes have been proven to be caused by a sudden change of the lead vehicle, e.g., a sudden slowdown or lane switch [16,17]. Traffic condition and drivers’ reaction speed have been found to affect the occurrences of a chain-reaction crash based on simulation methods [18]. In addition to simple chain-reaction crashes, traffic hazards may also spread backwards in a non-linear or irregular manner based on traffic wave theory. Imagine that the drivers of the following vehicles have different alertness levels and reaction swiftness, the same hazard produced in the front may cause different driving behaviors and vehicular interactions in its behind. Hence, collisions may not always occur to adjacent vehicles in the same lane. Instead, secondary crashes may happen several vehicles away from the primary crash shortly within its influential area, and then subsequently induce a series of CCs in a short period of time. This effect is also probable to be magnified on high speed roads and with busy traffic, such as on freeways, resulting in higher CC risks and severity levels [16,18].Although there are considerable similarities between CCs and chain-reaction crashes, their impacts on the traffic system and society are quite different. CCs can cause relatively more burdens to the affected road and traffic because they involve multiple crashes in a limited road area, which possibly causes congestion and are difficult to evacuate. Clearance and traffic management of multiple crashes within a limited area of a road network are also relatively more cumbersome, and hence injuries may not be promptly transferred to hospital and higher injury severity or even more fatalities may result. Despite these challenges, little research has been focused on understanding the crucial aspects of this type of crash. Some experience from previous studies on the occurrence of primary and secondary crashes can be referred to, but they hardly addressed crash severity of secondary crashes and its influential factors, not to mention the severity issues of CCs.This study aims to identify the factors associated with a CC series (including a primary CC and at least one secondary CC) on freeways while addressing unobserved spatial heterogeneity in the dataset. The crash data used in this study include 8779 crashes on the freeway network of Guizhou Province, China in 2018, from which 349 crashes involved in CC series were identified based on a static identification method. Factors including road geometric design, traffic conditions, environmental and crash attributes of the primary and secondary CCs are incorporated into the model with elaborated interaction terms. A basic, a random-effects (RE), a random-parameters (RP) and a two-level binary logistic regression are estimated and compared to quantify the relationships between CC severity and various influential factors. The best model form for modeling CC severity while dealing with unobserved heterogeneity and data hierarchy is selected based on the AIC values and other commonly used metrics. Significant influential factors are identified based on the model form with the best performance and discussed accordingly. Policies are also suggested to alleviate crash severity levels of CC on freeways and mitigate the potential losses.The crash database used in this study was provided by Guizhou Traffic Information and Emergency Control Center, affiliated to the Department of Transportation of the Guizhou Province. The data were originally recorded by onsite traffic police officers for all crashes on national, provincial and local freeways in the province. Data on the speed limit and monthly traffic volume in 2018 was provided by the Department of Transportation of Guizhou Province. In this study, a static identification method was applied to identify CCs from the crash database based on two conditions: (1) a secondary crash occurs in the downstream of the primary crash, and (2) the secondary is within 1 km in space and 1 min in time from the primary crash [13]. In total, there were 135 primary crashes and 214 immediate secondary crashes extracted from the database, resulting in an average of 1.59 secondary CCs per primary CC. To facilitate the understanding of influential factors of CCs and the differences between the contributory factors to the severity of primary and secondary CCs, all of these 349 crashes were included in the modeling.The Transport Department of Guizhou has divided the road networks into 79 segments for management purposes. The length of these segments ranges from 40–80 km. Apart from their different locations and geographical and environmental conditions, different segments may be also under surveillance and management of different police departments from different cities/towns. The 349 crashes extracted for this study covered 12 out of the 79 road segments.The dependent variable was extracted from the crash records as the severity level of each relevant crash. Given that the fatal crashes only accounted for a very small proportion among both the total crashes and the 349 CCs incorporated in this study (less than 1%), the dependent variable was set to contain two categories: severe (with at least one injury caused) and non-severe (with property damages only). The dichotomous nature of the dependent variable gives rise to the choices on model forms of this study.Environmental factors and road geometric design attributes for each crash were recorded directly in the crash database, including time, location, weather, etc. The traffic volume information of the road segment where the crash happens was extracted based on the location information of each crash and transformed into 106 units as the numbers were usually very large. Given the location of each crash, the speed limit and number of lanes of the road segment was also collected. If a crash occurred in a tunnel or on a bridge, the information was also recorded by the variable “location”, as the special driving environments may affect the transmission of the hazards produced by the primary crashes.Crash-related characteristics such as the number of vehicles involved, crash type, and various types of vehicle involvement were also extracted from the database. Rear-end, rollover, and side-swipe crashes are worthy of special attention as three categories of the variable, crash type, with hitting objects as the base level. A standardized vehicle type categorization scheme of China was adopted in the database, classifying all vehicles into two main categories: light vehicles such as passenger cars and light commercial vehicles, and heavy vehicles which include trucks, buses, and trailer trucks. Heavy vehicles deserve more attention in this study as these vehicles are large in size and likely to induce more and propagate further hazardous driving conditions to vehicles in behind [11,19]. The involvements of truck, trailer truck, and bus in each crash were formulated into tree respective dummy variables to represent whether these three types of massive vehicles were involved. Number of vehicles involved in a crash is presumably contributive to the severity level of a crash on freeways and was also adopted as a dummy variable with the multiple-vehicle crashes as the baseline level. Besides, primary and secondary crashes were differentiated by the variable, “secondary CC”, as the contributary factors to the severity levels of primary and secondary CCs are assumed to be different. Interaction effects of this variable with other environmental factors and crash attributes are readily adopted in the models, to test the contributions of these factors on the severity levels of CCs.As secondary CCs are by nature a result of the hazards produced by the corresponding primary crash, the crash severity of a secondary CC is assumed to be affected by the crash-related attributes of the primary crash. Hence, attributes of the primary crashes including crash type, number of vehicles involved, and involvements of various types of heavy vehicles (i.e., truck, trailer truck and bus) were delineated for all observations. The interaction terms between crash attributes of the primary crash and “secondary CC” were also introduced to measure the effects of crash-related attributes of the primary crash on the severity level(s) of its related secondary CC(s).The descriptive statistics of the dependent and independent variables are shown in Table 1. For continuous variables, the sample means and standard deviations were provided. Categorical variables were all coded into dummy variables, and the percentage of each category together with that of the base level were provided. In addition, the conceptualization of the hierarchical data structure has been displayed in Figure 1, illustrating the two-level settings and the rationale for incorporating interaction terms.Logistic regression has been proven to be effective and widely used in measuring the relationships between binary injury outcomes, as the link function can transfer a linear function into a continuous probability function ranging from 0 to 1 [1,2,3,20,21]. Let Yi denote the outcome of CC severity i. Yi=1 means that crash i is a severe crash, while Yi=0 means that crash i is a non-severe crash. A binary logistic function is used to link the probability of Yi=1 (denoted as πi) with the independent variables as follows [22]:(1)logit(πi)=log(πi1−πi)=β0+∑k=1KβkXik+εi
+        where Xik is the value of the kth independent variable for crash i, β0 is the intercept of the model, βk is the estimated coefficient for Xik, and εi is the random error term following a logistic distribution.Unobserved heterogeneity has been shown to widely exist in road crash modeling [20,21,23,24,25,26,27], as factors that were not accounted for by the independent variables may induce individual heterogenous effects on the outcome variable [26]. To address the unobserved heterogeneity possibly existing in our dataset, a RE modeling approach was adopted by incorporating a random intercept term, vi, which was normally distributed across individual observations with a mean of 0 and a standard deviation of σv. The RE logistic function is formulated as follows:(2)logit(πi)=log(πi1−πi)=β0+∑k=1KβkXik+vi+εiA maximum likelihood estimation method is applied to estimate the coefficients, β0 and βk [22]. In addition to the coefficients for the independent variables, the variance of the random intercept, σv is also estimated simultaneously. Statistically, scholars commonly use another parameter, ρ, to represent the proportion of variance explained by the random effect [28]:(3)ρ=σv2σv2+σε2
+        where σε2 is the variance of εi.While RE model is considered as a special case of a RP model where only the constant term is assumed to be random [26], a RP model has become increasingly prevalent in recent traffic safety studies due to its ability to address the heterogeneity in all independent variables. A RP modeling approach was therefore used by allowing the estimated coefficients for desired independent variables to vary across observations, as a comparison with the RE model. The RP logistic function is described as function (1), with the estimated coefficient βk allowed to be random as follows:(4)βk=β¯k+μik
+        where β¯k is the estimated mean of the kth coefficient, and μik is a random term following a normal distribution with mean of zero and variance of σk2.A simulated maximum likelihood estimation method with 200 Halton draws is applied to estimate the coefficients, β¯k and σk [22]. A Z-test was applied to each estimated coefficient, and only the coefficients with the a significant mean, β¯k, and a significant standard deviation (SD), σk, were treated as random coefficients.Previous studies have found that as most crash data were hierarchical in nature, spatial heterogeneity may exist on the higher level(s) of the hierarchy [29,30,31]. Vanlaar [32] found in his research on drunk driving that ignoring a hierarchical data structure could lead to underestimation of standard errors in predictors. Dupont et al. [33] also pointed out that traditional model structures ignoring hierarchical heterogeneity in the data structure might induce problematic inferences and conclusions. To address the within-road-segment correlation while capturing spatial heterogeneity across various road segments, a two-level modeling scheme was proposed here. In this case, level 1 was the crash level and level 2 was the road segment level. The basic logistic function should be rewritten into a two-level model form as follows:(5)logit(πij)=log(πij1−πij)=β0j+∑k=1Kβ1jkXijk+εij
+        where j is the index of road segment, β0j is the crash-level intercept and β1jk is the estimated coefficient of covariate Xijk. Note that β0j and β1jk both vary across road segments (i.e., on level 2) and εij varies across all observations (i.e., on level 1). To address the cross-road-segment variations, the level 2 model is specified as follows:(6)β0j=γ00+∑l=1Lγ0lZjl+μ0jβ1jk=γ1k+μ1jk
+        where γ00 and γ1k are the fixed intercepts on the road segment level; Zjl is the lth road-segment-level independent variable for segment j; γ0l is the estimated fixed coefficient for Zjl; μ0j and μ1jk are the random effects varying across road segments for the crash-level intercept and crash-level covariate k with means of zero and variances of σ02 and σk2, respectively [34]. A similar simulated maximum likelihood estimation method with 200 Halton draws was used to accomplish model estimations. A t-test was performed for all estimated parameters including the random and fixed parameters on both crash and road segment levels.As the estimated coefficients may not always directly represents the effects that a contributory factor has on the indicator, an elasticity analysis is necessary for quantifying the effect of each independent variable based on the observed and estimated information [28,35,36]. The elasticity for a continuous independent variable k on the probability of a severe crash is calculated from the partial derivative of each observations [28]:(7)Eik=∂πi∂XikXikπi
+        where the Eik is the elasticity outcome for continuous variable k of crash observation i. As the probability for a crash to be severe is not differentiable with dummy independent variables, a pseudo-elasticity is defined for indicators as follows [33,34]:(8)Eik(p)=∂πi(Xik=1)−∂πi(Xik=0)∂πi(Xik=0)
+        where Eik(p) is the pseudo elasticity of dummy variable k of crash observation i. The final elasticity of a variable is calculated as the sample mean of the elasticity outcomes for all observations.To facilitate model comparison, likelihood-based model performance indices such as log-likelihood and the Akaike Information Criterion (AIC) value were calculated. For the RE, RP, and two-level logistic regressions, a McFadden Pseudo R2 value was also calculated as it is widely used as a measure for model fitting evaluation for simulated-based model estimations [22]. Similar to R2 value in ordinary least square estimation, a McFadden Pseudo R2 value varies between 0 to 1, and a value closer to 1 indicates a better model fit.Moreover, the AIC is capable of reflecting the amount of information loss in each model [37] given the likelihood performance and number of predictors used in the model. The definition of the AIC value is as follows:(9)AIC=2n−2ln(L^)
+        where n is the number of estimated parameters, and L^ is the maximum value of the likelihood function for the model. The model with the lowest AIC value has the best performance in terms of using as few parameters as possible to maximize the likelihood function of the model.To identify the significant factors associated with the severity level of CC series and compare the differences in influential factors between primary and secondary CCs, four models based on the logit link function have been estimated: basic, RE, RP, and two-level logistic models. The same set of dependent and independent variables were incorporated into all four proposed models. The environmental (level 2) independent variables and the crash-related (level 1) independent variables were all included. To test the contributions of attributes of the primary crash on the crash severity of its corresponding CC(s), the interaction terms between the crash-related attributes of the primary crash and the indicator variable representing secondary CC (namely “consecutive secondary crash” in Table 1) were adopted in the models. Besides, the indicator variable for secondary CC was also interacted with environmental variables and incorporated in all four models, to test the direct effects of the environmental variables on CC severity. Spearman’s correlation tests were performed for all pairs of independent variables to identify possible strong collinearities from the variable set [28]. All Spearman’s correlation values were smaller than 0.5, indicating that no strong collinearity was observed from the variables.For the basic and RE binary logistic models, all independent variables in the aforementioned variable set were first adopted with their estimated parameters tested for statistical significance. In the final results in Table 2, the insignificant continuous variables and the categorical variables with all sub-categories that were insignificant at the 90% confidence level have been excluded. In the basic logistic model, ten coefficients were significant at the 95% confidence level and two coefficients (for truck involvement and tunnel × secondary CC) were marginally significant at the 90% confidence level. In the RE model, the same significant variables were found, but the random effect turned out to be insignificant, reflecting that the random intercept was not adequate to address the unobserved heterogeneity in the dataset.For the RP model, a stepwise method for selecting proper RPs was adopted. The variables were set to be random one by one following an order from the lower-level variable (crash level) to the higher-level (environmental) variable, and then to the interaction terms (the order for adopting random interaction terms was also from the lower to the higher level). Note that parameters of all categories for a categorical variable were considered to be random in the same step. Only the variables with both the mean and the standard deviation (S.D.) were significant at the 95% confidence level were kept random, and the rest were all set as fixed parameters. In the final model (see Table 2), only the significant continuous variables and the categorical variables with at least one significant category were kept (confidence level is set to be 90%). Finally, nine fixed coefficients were significant at the 95% confidence level or above and two were marginally significant at the 90% confidence level. One variable was found to have a significant mean and a significant SD at the same time.According to the model forms of the two-level binary logistic model, a random term for each crash-level variable was estimated. As the interaction terms between the secondary CC indicator and crash-level attributes also represented crash-level effects, random effects were allowed for these interaction terms as well. For random term selection, only the variables with either a significant fixed effect or a significant random effect was adopted in the final model.That is to say, the crash-level variables (including the aforementioned interactions) whose fixed and random terms were both insignificant at the 95% confidence level were excluded as no significant effect were found for this variable from both levels. For level-2 variables, statistically insignificant continuous variables and categorical variables whose all sub-categories were insignificant have been excluded in the final results. Six level-2 variables had a coefficient that was significant at the 95% level or above; two level-1 interaction terms were significant at the 95% level. Although the estimated fixed intercept was insignificant, the random effect of the intercept varying across road segments was significant at the 95% confidence level.Major metrics reflecting model performances of the four proposed models for severity of CC series are listed and compared in Table 3. Based on the log-likelihood value at convergence and the AIC, the RE model (log-likelihood = −162.70, AIC = 355.4) didn’t improve in model performance compared to the basic binary logistic model (log-likelihood = −162.70, AIC = 353.4). The RP model had a relatively higher log-likelihood at convergence (−159.88) and lower AIC (349.8) than the basic and the RE logistic regression, indicating that the full heterogeneity model can better address the unobserved heterogeneity in the data and achieve a better model performance. The two-level logistic regression outperformed the other three models and had a substantial enhancement in all of the three major metrics, especially in the AIC value (325.4), meaning that within-segment correlation exists in our dataset and the two-level model structure fits the crash dataset better than the other three proposed model forms. Hence, the elasticity analyses were performed for the optimal model choice, the two-level model, to facilitate the discussion process. As shown in Table 4, all of the elasticity values were significant at the 99% confidence level, and the results echo with the estimated coefficients in Table 2. The elasticity values for five main effects were positive, namely speed limit, traffic volume, rainy, cloudy and tunnel. Two other main effects had a negative elasticity value: the number of lanes and the intercept.As the two-level binary logistic regression model outperformed the other three models in modeling the crash severity for CC series, the two-level form is considered superior among the tested model forms with the most unbiased results. Hence, the following discussions are mainly based on the estimation results of the two-level binary logistic model.As shown in Table 2, the random terms for the crash-level variables, the interaction between truck involvement and secondary CC, and the interaction between double-vehicle primary crash and secondary CC were not significant, indicating that although these two factors have significant fixed contributions on secondary CC severity, the road-segment-level spatial heterogeneity doesn’t locate in these two variables. Instead, the SD of the random intercept of the road segment level was significant at the 95% confidence level, indicating that spatial heterogeneity exists on the road segment level and locates mainly in the intercept. Given that the two-level structure has been proven more robust and unbiased, the individually heterogenous effect of “cloudy” in the RP regression was rather a false alarm.The interaction between truck involvement and secondary CC crash was found to be positively associated with the propensity of a severe crash (coefficient = 2.80). This result indicates that if trucks were involved in a secondary CC, the severity outcome of the crash tended to be more severe compared to the one without trucks involved. Similar results have been found in previous research in which the involvement of truck was reported to significantly increase the severity of the crash under various circumstances, provided that trucks are massive in weight and size and potentially more disruptive in a traffic collision [7,38,39]. Zhou and Zhang [40] studied the potential hazardous driving behaviors of commercial truck drivers and identified that 40% of truck drivers tended to drive in a substantially dangerous way. Secondary CCs occur under the direct effect of the primary crash immediately. For secondary crashes in a CC series, taking a swift action to the traffic hazards generated from the primary crash is crucial to alleviating its crash severity, yet trucks are cumbersome and slow down the changes in movements to a large extend. The nature of massiveness, the hazardous driving behaviors, and the cumbersomeness in reaction of a truck are the possible reasons that more severe secondary CCs tend to be induced when trucks are involved.The estimated coefficient for the interaction term between two-vehicle primary crash and secondary CC was significantly negative at the 95% confidence level (coefficient = −1.59). Compared to a primary crash with three or more vehicles involved, a primary crash involving two vehicles had a lower propensity to cause a severe CC in its behind. Nagatani and Yonekura [16], Sugiyama and Nagatani [17], and Nagatani [18] investigated the mechanism of multiple-vehicle crashes with various hazardous inputs such as a sudden lane change and a sudden brake based on a car-following model, similar to the mechanisms of CCs except for the less strong assumptions in the regularity of vehicular movements and reaction behaviors. Under most circumstances, the following vehicles’ sensitivity (agility in reaction) was associated with the occurrences of chain-reaction crashes and the number of vehicles involved in the crash chain. In the case of CCs, similar car-following behaviors are also valid and reaction speed is even more crucial if the primary crash involves more than three vehicles and engages a relatively larger area of the road and generates a larger number of traffic hazards from multiple origins, possibly resulting in multiple unsafe behaviors in the traffic behind. Compared with two-vehicle primary crashes, the more complicated combination of behaviors can possibly overload the following vehicles’ reaction capabilities and thus cause a more severe secondary CC subsequently.Although multiple-vehicle primary crashes were proven to have a higher likelihood of severe secondary CCs than two-vehicle primary crashes, the number of vehicles involved in the primary crash was not positively associated with the severity of secondary CCs. The estimated coefficient for the interaction between single-vehicle primary crash and secondary CC was insignificant, meaning that no evidence could be found to differentiate the effect of a single-vehicle primary crash on the severity of the secondary CC from that of a multiple-vehicle primary crash as single-vehicle crashes usually occur under extreme circumstances, such as losing control of the vehicle or rollover crashes [41] and could have ambiguous effects on the severity of the secondary CCs.In the two-level binary logistic model, six environmental factors were significant at the 95% level or above, five of which had a positive effect on the severities of crashes in a CC series based on the estimation results, including speed limit, traffic volume, rainy, cloudy, tunnel, and bridge. According to the elasticity analyses, the sample means of the marginal effects of these variables were also significantly positive. The number of lanes was the only environmental variable that was negatively associated with the crash severity of CC series.The elasticity value of speed limit was 2.60, indicating that 1% increase in speed limit will lead to a 2.6% increase in the probability of a severe CC on average. This result reveals a similar effect of speed limit to positively contribute to the probability of severe crashes on freeways or highways [38,42,43], especially to those fatal crashes. As vehicles tend to have a higher speed driving on a freeway with a higher speed limit, drivers have a relatively shorter time for reaction to emergencies and take action to avoid a fierce collision. Hence, for both primary and secondary crashes in a CC series, a higher speed limit is more likely to result in severe crash outcomes.Traffic volume of the road segment where the CC occurred had a significantly positive effect on CC severity (coefficient = 1.41) and a significantly positive elasticity effect (1.64). This result suggests that the denser the traffic is, the more likely for a CC to be severe. Zeng, et al. [44] studied the severity crashes on freeways and found that traffic volume significantly affected crash severity and impacted the threshold between median and severe crashes. Wang, et al. [45] also concluded that a higher traffic volume may increase the distance gap between the primary and the secondary crash based on a shock wave method. Although this result may apply to CCs, the longer distance gap could still add inequality in the spread pattern of the traffic hazards produced by the primary CC, especially within a time period that is short enough, and hence result in more severe secondary CCs.The number of lanes was the only environmental factor that had a significantly negative effect on the severity of crashes in a CC series among all the factors estimated (coefficient = −4.29, p = 0.008). The elasticity value of this variable was −5.11 and significant at the 99% level, indicating that 1% increase in the number of lanes on one side of the freeway may decrease the probability of a CC to be severe by 5.77%. Compared with a CC occurring on a three-lane freeway, the same CC occurring on a two-lane freeway had approximately 1.9 times higher probability to be severe if other factors hold the same. In previous research, hazardous lane-keeping behaviors of the front vehicle has been widely proven to affect crash injury severity positively. Jamal, et al. [46] studied the vehicular crash severity in Saudi Arabia and identified that a sudden deviation from the lane of the at-fault vehicle caused a higher crash severity. Shao, et al. [47] applied a RP ordered logit model to prove that a sudden stop of the front vehicle in the same lane significantly raised the injury severity of rear-end crashes with trucks involved. Hence, the severity of the primary crash in a CC series can be affected by the hazardous lane-related behaviors, such as a sudden lane change or a drastic change in speed, of the front vehicle(s), and a wider road with more numbers of lanes provides an wider space to avoid a fierce collision under such circumstances. For the secondary CCs, hazards for a severe crash are received in two-ways: one is directly from the primary crash and the other is through (probably hazardous or abnormal) behaviors of the vehicles between them. In both cases, more lanes provide wider spaces to prepare and take action and can possibly reduce the severity level although a crash is still unavoidable.A CC in a freeway tunnel had a 19.4% higher propensity to be severe than that occurring on open freeways (coefficient = 2.47, elasticity = 19.4). In its nature, a freeway tunnel possesses traffic hazards given its constraint driving space and tedious driving environment [48,49]. Poor lighting and fatigued driving in a tunnel are also often considered as potential hazards causing crashes, especially severe ones [50,51]. Based on the definition, a secondary CC happens very shortly after the primary crash occurs, meaning that there is extremely limited time for the following drivers to react to the primary crash. Hence, the poor driving environments and the fatigued-prone mental and physical conditions of the drivers in tunnels may result in a belated reaction than normal and thus cause severe CCs.For the various weather conditions estimated in the two-level binary logistic regression model for CC severity, rainy (coefficient = 2.84) and cloudy (coefficient = 2.42) showed significant positive effects at the 95% confidence level or above. Compared with sunny days, a CC happening on a rainy day had a 52.5% higher possibility to be severe, and that happening on a cloudy day had a 66.6% higher possibility to be severe. These two types of extreme weathers display similar associations with crash severity in other contexts and can both incur poor visibility and hinder drivers’ perceptions and reactions to potential hazards or a primary crash, and thus causes more severe CCs [19,46,52,53]. Besides, pavements of freeways tend to be more slippery on rainy days [19,20,46], probably leading to a longer braking distance, and consequently result in a severe CC. It is worth noting that previous research found that drivers tended to drive more carefully under hazardous weathers such as on rainy days and thus probabilities for crash fatalities on rainy and cloudy days are lower especially during the day [54]. Based on the results in our study, although skillful drivers may get accustomed to extreme weathers promptly and drive carefully, the secondary CCs happen “within a blink” after the primary crash, and the objective elements (i.e., the existing hazards of extreme weathers) play a more crucial rule and cause more severe crash outcomes.This study investigated the influential factors associated with the severity of crashes in a CC series. The severity of CCs occurring on the freeway network of Guizhou Province, China in 2018 were adopted in the modeling. Four model forms, including basic, RE, RP, and two-level binary logistic regressions were proposed to link the probability of a severe CC to the potential influential factors (i.e., environmental factors and crash-related attributes). To identify the specific influential factors affecting secondary CC severity, interactions between the dummy indicator for secondary CC and various factors (i.e., environmental factors and primary crash attributes) were created and incorporated in the models. The two-level logistic model was proven to outperform the other three model forms for modeling CC severity with our dataset, according to modeling performance metrics such as log-likelihood, McFadden Pseudo R2 and the AIC value.According to the coefficient estimates of the two-level logistic model, significant influential factors affecting CC severity were discussed. On the crash-level, the involvement of trucks in a secondary CC had a higher probability for it to be severe, and a primary CC with two vehicles involved had a significantly lower probability to incur a severe secondary CC than that with three or more vehicles. The facts that a secondary CC occur very shortly after the primary CC and that prompt reactions are crucial to its crash severity help to explain the crash-related results. On the road-segment level, environmental factors such as speed limit, traffic volume, tunnel, and rainy and cloudy weathers had significantly positive associations with CC crash severity. The number of lanes, on the contrary, had a negative effect on the severity of CC as a larger road space prepares the following vehicles for a possible escape from a severe hitting.As discussed before, CCs contains multiple crashes can typically affects a large area of the road and paralyses the traffic system. Evacuations of the injured are normally difficult, especially if severe congestions are formed. Hence, precautions for lowering the crash severity of CCs are crucial. Based on the modeling results of this study, some policy suggestions can be made to reduce the severity levels of CCs. First, drivers should be reminded with hazardous situations where severe CC tend to occur. Dynamic warning signs should be placed in freeway tunnels and two-lane open roads with high speed limits (120 km/h in the case of China), especially in extreme weather such as on rainy and cloudy days. Similar warning systems are suggested to be inserted into mobile applications for navigation or the navigation systems in the vehicles to remind drivers of keeping a long-enough headway and alert of the potential hazards in the front. Besides, truck drivers are advised to experience trainings theoretically and practically with the potentiality for severe CCs to follow if they are involved in a primary crash. The truck drivers’ responsibility in maintaining a healthy driving environment on the road and preventing severe subsequent CCs from happening should be strengthened.Based on local policies on crash data disclosure in Guizhou Province, only the data from 2018 was adopted in this study. Future works could benefit from applying multi-year data and study the temporal effects for CC severity [23]. Besides, the temporal threshold for the identification of CCs in this study is subject to minimum time interval (1 min) between two crash records in the database. Although 1 min is considered short enough as the behaviors and hazards induced by the primary crash needs time to spread, future studies are suggested to further split the temporal threshold into shorter periods and define the secondary CC in each period distinctively (i.e., CC occurring within 30 s, CC occurring 30 s to 1min, …), and compare the patterns in the severity of these subgroups of secondary CCs.Conceptualization, F.M.; data curation, C.S. and Z.Z.; formal analysis, F.M.; funding acquisition, C.S. and L.Y.; investigation, F.M. and Z.Z.; methodology, F.M.; project administration, L.Y.; resources, C.S.; software, F.M.; supervision, L.Y.; validation, P.X., K.G. and Z.Z.; writing—original draft, F.M.; writing—review & editing, P.X. and K.G. All authors have read and agreed to the published version of the manuscript.This study was funded by the NATIONAL NATURAL SCIENCE FOUNDATION OF CHINA, Grant No. 7177113, NATIONAL KEY R&D PROGRAM OF CHINA, Grant Nos. 2018YFC0807000 and 2019YFC0810705, and the SHANGHAI SAIL PROGRAM, Grant No. 19YF1451800. The funders had no roles in study design, data collection and analysis, decision to publish, or preparation of the manuscript.We would like to acknowledge the Guizhou Transport Information and Emergency Control Center and Department of Transportation of Guizhou Province for providing the crash database and the traffic volume.The authors declare no conflict of interest.Conceptualization of the model structure for the severity of CC.Descriptive statistics of dependent and independent variables.† Reference group.Estimation results for the proposed four model forms.* means the estimated coefficient is significant at the 90% confidence level. ** means the estimated coefficient is significant at the 95% confidence level or above.Model comparison based on log-likelihood, McFadden Pseudo R2, and AIC.Elasticity analyses results for the two-level binary logistic model.** means the estimated coefficient is significant at the 95% confidence level or above.

Med-MDPI/ijerph_5/ijerph-17-15-05624.txt

@@ -0,0 +1 @@
+Restoring bone loss is one of the major challenges when facing hip revision surgery. To eliminate the risk of disease transmission and antigenicity of allografts and donor-morbidity of autografts, the use of synthetic bioceramics has become popular in the last decade. Our study investigated the effectiveness of impaction bone grafting (IBG) of contained acetabular defects (Paprosky 2 and 3a) using a porous ceramic-based hydroxyapatite bone substitute (Engipore, provided by Finceramica Faenza S.p.A., Faenza, Italy) mixed with a low percentage of autologous bone (obtained from reaming when available). We retrospectively assessed 36 patients who underwent acetabular revision using IBG using a porous ceramic-based hydroxyapatite bone substitute with cementless implants with a mean follow-up of 4.4 years. We evaluated, at regular intervals, patients clinically (using the Hip Harris Score and Oxford Score) and radiologically to evaluate the rate of incorporation of the graft, the presence of radiolucent lines or migrations of the cup. Clinical scores significantly improved (WOMAC improved from 49.7–67.30, and the HSS from 56–89). The rate of implants’ survival was 100% at our medium follow-up (4.4 years). We reported five cases of minor migration of the cup, and radiolucent lines were visible in seven patients at the last-follow up. The graft was well-incorporated in all patients. The results presented in this study suggest the HA bone substitute is an effective and safe bone graft when facing hip revision surgery; thus, longer follow-up studies are required.With the ageing of the population and a life expectancy increase in the Western world, there will be a greater demand for total hip replacement procedures and, subsequently, of hip revisions.In the United States, more than 300,000 total hip replacements (THAs) are performed yearly and, according to the National Inpatient Sample of the US, over 250,000 THA revisions were performed between 2009 and 2013 [1]. Given the premises, the revision surgery rate is expected to grow by 137% by the end of 2030 [2].The main indications for revision surgery are hip instability and aseptic loosening of the implant, accounting for 42% of all revision procedures [3].The greatest concerns for orthopedic surgeons dealing with a THA revision is how to address bone loss and poor bone regenerative potential to restore the anatomical hip center and improve joint biomechanics.Impaction bone grafting (IBG) has been described for the treatment of contained defects where a good primary stability could be obtained [4]. Although long-term results with this procedure are mainly related to the surgical technique, good outcomes have been reported in the literature.IBG of the acetabulum, first performed by Parker et al. in 1975 [5], was made popular by Slooff et al. in 1984 using a cemented acetabular component [6]. More recently, favorable results have been reported when morcellized allografts were used in combination with uncemented cups [7,8,9].There are multiple viable sources for bone grafting. Autografts are described as the gold standard source to treat bone loss in terms of the properties of osteoconduction, osteoinduction and osteogenesis, but they are rarely used in THA revisions due to high rates of morbidity at harvest sites and their limited availability [10]. Allografts retain osteoconductive properties and may exhibit osteoinductivity potential. For these reasons, although the option presents high costs and there are still concerns about the related risks of disease transmission and antigenicity [11], they could be considered the second-choice option to address bone defects in THA revision procedures.All the above concerns have led to growing research in developing synthetic bone substitutes for natural bone stock replacement.Bioactive ceramics are synthetic bone substitutes which have received great attention recently due to their potential in stimulating cell proliferation, differentiation and bone tissue regeneration [12]. Several synthetic ceramics have been tested and used, such as calcium phosphate, tricalcium phosphate, calcium sulphate and hydroxyapatite (HA) [13].In this scenario, hydroxyapatite (HA), a major component of natural bone, can combine with tissues by chemical bonds to form new bone tissue when implanted [14].The ceramic-based hydroxyapatite bone graft substitute Engipore (provided by Finceramica Faenza S.p.A., Faenza, Italy) consists of porous hydroxyapatite that allows the product to serve as a scaffold to guide bone regeneration, fostering cell attachment and proliferation and promoting osteointegration [15]. The aim of this retrospective study was to analyze the osteointegrative properties of Engipore bone grafts in revision surgery after primary THA.This is a retrospective case series of patients who underwent an acetabular revision associated with the use of the bone substitute Engipore between January 2014 and December 2016.In this period, 105 THA revision procedures were performed: we selected 36 patients whose indication for THA revision was aseptical loosening of the acetabulum after primary THA with a contained bone defect (Paprosky 2 and 3a) treated with cementless implants.There were 21 women (58.3%) and 15 men (41.7%). The mean age at surgery was 72 (age range: 40–81 years). Patients with an uncontained defect requiring the use of mesh or a cage to reconstruct the acetabulum were excluded from this analysis. It was the first revision in all cases.We performed a postero-lateral approach to the hip in all patients, obtaining a good exposure of the acetabulum. After implant removal, necrotic and soft tissue surrounding the acetabulum and the bone defect were carefully removed in order to evaluate the severity of bone loss according to Paprosky’s classification (see Table 1) [16].We prepared the acetabulum with progressive hemispherical reamers and then packed the bone loss with reverse reaming and trial prosthesis using a mixture of Engipore chips mixed with the patient’s blood and autologous bone recovered from reaming when available.The bone substitute Engipore is a biomimetic and biocompatible porous stoichiometric hydroxyapatite bone substitute comprised of calcium ions, phosphate ions and hydroxyl groups, which is very similar in microstructure and chemical composition to the mineral component of human bones. The trabecular structure, which resembles the mineralized phase of natural bone, is characterized by a 90% porosity rate, allowing physiological fluids absorption, the promotion of cell migration and adhesion for mineral matrix synthesis, thus offering an ideal environment for new bone formation and tissue restoration [17]. The composition, shape and handling properties of this bone substitute make it an ideal bone graft candidate for hip revision surgery. Moreover, this material can be safely mixed with autologous bone. The chips provided by Finceramica used in our study came in the size of 2–4 mm.Our case series is comprised of selected patients with contained defects, so it was possible to achieve a primary press-fit stability placing uncemented cups in all cases (Regenerex Revision Shell in 17 cases, Delta TT Revision Cup in 19 cases), 1–2 mm larger than the last trial used. Regardless of the primary stability of the implant, a median of 4 screws (2–6) was used to fix the shell (Figure 1). The post-operative rehabilitation protocol included: mobilization of the hip, avoiding luxation movements, and partial weight bearing was given for 6 weeks and then gradually progressed to full weight bearing and was performed by all the patients. Clinical and radiographical evaluations of all patients were performed pre-operatively and post-operatively at regular intervals (1, 3, 6 months and yearly after). The radiological follow-up included anterior-posterior X-rays of the pelvis and antero-posterior and lateral X-rays of the hip. The clinical evaluation was conducted using the Harris Hip Score (HHS) [18] and the West Ontario and McMaster Universities Osteoarthritis Index (WOMAC) [19]. We eventually reached by phone patients (or their relatives) who did not fully accomplish follow-up visits to assess whether they had underwent further surgery. Radiographic analysis was carried out by two senior surgeons (N.P. and M.S.), assessing graft incorporation, bone resorption or migration of the implant. We used the De Lee and Charnley classification [20] to assess lines of radiolucency around the acetabular component. A radiolucent line wider than 2 mm was considered significant. We considered the surgery to have failed if the cup had migrated 3 mm or more or if revision was necessary. The grade of heterotopic ossification was evaluated according to Brooker et al. [21]. The mean follow-up was 4.4 years (minimum 3.1—maximum 5.8 years). Given the small sample size, statistical testing of correlations was not determined to be useful.This is retrospective study which was approved by internal revision board and for these reason needs only a tacital approval of the ethical committee—ethical approval was not needed for this study.Thirty-six patients met the inclusion and exclusion criteria defined for our retrospective case analysis. Five patients died of disease not related to the surgical procedures. Among the remaining 31 patients, 25 patients returned a complete questionnaire; meanwhile, a full radiological follow-up was available in 28 cases.We recorded a significant improvement in clinical function: the WOMAC score changed from 49.7 pre-operatively to 67.3 post-operatively, and the HHS changed from 56.1 pre-operatively to 89.4 post-operatively. At the final follow-up, most of the patients (67%) stated they would undergo surgery again. Data about survivorship of the implant were available for all patients. There were no cases of acetabular revision in the selected patients, so the rate of survivorship of the implant at our medium follow-up was 100%. We reported five cases of acetabular migrations wider than 3 mm (range: 3–6 mm), but surgery was not necessary because four patients reported no functional impairment and one patient had functional impairment but refused further surgery. All migrations occurred within one year after surgery and then remained stable during the follow-up and were not associated with significant worsening of the clinical scores. According to De Lee and Charnley et al. [20], we found at the initial follow-up two patients with radiolucent lines bigger than 1 mm in zone 1, three patients in zone 2, one patient in zone 3, two patients both in zone 1 and 2 and one patient in all three zones (see Table 2). No significant association between the presence of radiolucent lines and clinical outcome was noticed.In the last X-ray evaluation, radiolucent lines remained still visible and no signs of progression were detectable, except for two cases, in which radiolucent lines were filled by, presumably, the formation of new bone (see Figure 2).Mild heterotopic ossification (type 1–2 according to the Brooker classification) was found in three patients. Early complications (during the in-patient stay) occurred in five of the 31 patients. These included a superficial wound infection in two patients, treated successfully with antibiotic therapy; one case of deep infection treated successfully with DAIR (debridement, antibiotics and implant retention); one case of deep-vein thrombosis and one case of postoperative early dislocation, which was successfully treated with close reduction. One patient sustained a periprosthetic fracture of the femur (type B1 according to the Vancouver Classification) after falling and was treated successfully with osteosynthesis with a plate.There is a great debate in the literature regarding clinical and radiological outcomes of hip revision surgery using synthetic bone substitutes. Kurien et al. [22], in a systematic review about the evidence of the use of bone graft substitutes, stated that there are few synthetic graft substitutes with level I evidence. A certain number of papers have reported good results on the use of bone substitutes in hip revision surgery, but they were all a heterogenous case series mix with different kinds of implants, wide bone loss severity and type of bone substitutes used. Our case series, even if small, has an asset of being a homogeneous court of patients characterized by an isolated acetabular contained defect treated with hydroxyapatite (mixed with autologous bone and blood obtained from reaming when available) and with cementless implants. IBG is recognized as an efficient method to treat contained acetabular defects [16,23], even if the procedure is described as successful for selected cases of the Paprosky 3 type [24]. A porous, cementless coated socket would lead to bony ingrowth and osteointegration, providing a stable and solid fixation [7]. Few authors have reported the results of the use of an isolated bone graft substitute, not augmented with allografts or autografts. Oonishi et al. [25] documented the use of hydroxyapatite granules to fill massive bone loss in 40 patients using cemented sockets, obtaining good clinical and radiological results in a 4–10 year follow-up. They reported three cup migrations associated with mild clinical impairment. Good osteointegration to native bone was observed in all 40 patients. Schwartz et al. [26] used a biphasic phosphor-calcium bone substitute to face severe acetabular bone loss using both jumbo cups and a screwed support ring. At a mean 10 year follow-up, they reported no cases of migration of the cup in living patients and good bone osteointegration. Coralline hydroxyapatite was used by Wasielewski et al. [27] in complex acetabular revision surgery, reporting one case of failure. No resorption of the graft was noticed and all cases showed good osteointegration. Our clinical results encourage the choice of synthetic bone substitutes, rather than a metallic augment, when facing THA revision. Patients were eventually satisfied in terms of pain relief and functional recovery, and most of them stated they would undergo surgery again. The radiographical results are quite in contrast to the clinical outcomes: we registered five cases of acetabular migration, of which one was considered a frank radiological failure but not associated with worsening of clinical scores. The incidence of radiolucent lines we reported around the acetabular shells was concerning, but similar to other studies in the literature [27,28]. Although these radiolucent lines showed up early after surgery, they did not become wider with the progress of time; thus, we agree with Schmalzried et al. [29], and we consider that these radiolucent lines could represent a predictive factor of future aseptic loosening, so patients need to be kept monitored. The rate of graft incorporation in our series was satisfactory, similar to others reported [25] or even higher compared to other series in the literature [30]. We reported only one patient with graft incorporation failure in all three zones, but the case was not associated with any clinical impairment. Synthetic bioceramics such as Engipore offer several advantages over autografts and allografts in terms of safety [10]. The synthetic fabrication of Engipore makes the product free of any risk of disease transmission or immunoreaction, and the availability of a wide range of shapes and formats makes it a valid option for different surgical applications. Moreover, its availability off-the-shelf eliminates the donor-morbidity of autograft harvests and reliance on bone banks for allograft supply.Based on this clinical experience, as a general comment, the identification of the most appropriate surgery, together with a proper application and packing of the chips in the surgical implantation phase, are key aspects for a successful outcome.The clinical and radiological results presented in our study suggest that the bioceramic bone substitute Engipore can be used as an ideal bone substitute in THA acetabular revision surgery, even if mixed with a low percentage of autologous bone. The product is safe, with no risk of disease transmission or an antigenic response. It led to a low rate of failure and a satisfactory grade of osteointegration. The results presented in the study must be interpreted with caution due to the low number of patients enrolled and the retrospective design of the study and follow-up period. Prospective, randomized controlled clinical studies would be beneficial to confirm its safety and efficacy at a longer follow-up.Conceptualization, P.D.P. and M.S. (Matteo Simonetti); methodology, N.P.; validation, E.B.; formal analysis, E.B. and N.P.; investigation, M.S. (Michelangelo Scaglione).; data curation, M.S. (Matteo Simonetti); writing—original draft preparation, M.S. (Matteo Simonetti); writing—review and editing, M.S. (Matteo Simonetti); supervision, M.S. (Michelangelo Scaglione) and P.D.P.; project administration, P.D.P. and M.S. (Michelangelo Scaglione). All authors have read and agreed to the published version of the manuscript.This research received no external funding.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.Clinic case: 76 y female with loosening of the acetabular cup (a) Pre-operative x-rays, (b) Intraoperative pictures show the acetabulum impacted with Engipore chips filling the bone defect (Paprosky 3a), (c) Post-operative x-rays at the last follow-up (3.2 years) showing good osteointegration of the implant.X-rays show good osteointegration of the implant and formation in a 72-year-old woman treated with cup revision for a Paprosky 3a. (a) One month follow-up, presence of radiolucent lines, (b) One year follow-up, shows filling of the radiolucent line with the formation of new bone.Paprosky’s classification of acetabular bone loss.Clinical and radiological results at the mean follow-up (4.4 years).HHS, Harris Hip Score; WS, WOMAC Score.

Med-MDPI/ijerph_5/ijerph-17-15-05625.txt

@@ -0,0 +1 @@
+Physical activity counseling in primary health care is regarded as a useful complementary preventive and therapeutic measure and is advocated by leading public health institutions. This integrative review summarizes the available data on physical activity counseling in primary care in Germany. A systematic literature search in various databases (peer reviewed and grey literature) was carried out for quantitative and qualitative studies on physical activity counseling and use of “Exercise on Prescription”. The 25 studies included show a very high methodological diversity and, in some cases, considerable risks of bias, with limited comparability across studies. Counseling was provided in all studies by physicians. They report frequent physical activity counseling, which is partly confirmed and partly refuted by patient data. The use of “Exercise on Prescription” is at a very low level. Information on the frequency of physical activity counseling in Germany varies depending on data source and is sometimes contradictory. Our review provides a synthesis of various perspectives on routine physical activity counseling in primary care in Germany. Future studies using standardized and validated instruments in representative samples are needed to further knowledge on counseling and to be able to establish trends in prevalence. Strengthening the topics of physical activity and health and physical activity counseling in medical curriculum is strongly recommended.The evidence on the wide-ranging health benefits of regular physical activity (PA) is overwhelming [1,2]. PA reduces mortality risk, the risk of chronic diseases with the highest disease burden, such as cardiovascular and metabolic diseases, cancers, and diseases of the musculoskeletal system, and is also an effective (complementary) therapeutic measure for these clinical conditions [2]. Nevertheless, PA levels remain low worldwide [3] and in Germany [4].The relevance attributed to routine PA promotion in primary care is based on two further aspects in addition to the health effects of PA. Through universal access to health care in most Western countries, physicians can reach practically all social-economic groups, and physicians are considered the most important source of health information. Because of this high public health potential, PA counseling in health care has been advocated by a number of public health institutions, including the World Health Organization [5]. In Germany, the Annual Meeting of German Physicians has also recently confirmed the importance of PA counseling as a part of physicians’ routine [6].In international practice, two general approaches in PA promotion in health care are established: PA counseling, where counseling is provided by physicians and/or other health care professionals and patients implement the recommendations on their own; and exercise referral (also called exercise on prescription, green prescription), where physicians refer patients to an existing group offer, usually in a community setting. An increasing number of countries worldwide have established exercise referral schemes and developed PA counseling programs [7,8].Exercise referral schemes [9] and PA counseling [10,11] have been shown to increase participants’ PA levels at least at short or middle term, and PA promotion interventions in primary care can yield clinically relevant effects [12].Little is known about the current level of routine PA promotion in primary care in Germany. The main aim of this study is to provide an overview in the form of an integrative review [13] of the prevalence of PA counseling in primary care and the use of the German Exercise on Prescription (EoP) program. Further, we aim to summarize data on the content and effects of, as well as barriers to routine PA counseling.The following study was prepared according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [14]. The systematic literature search, data extraction, and the assessment of the risk of bias in the individual studies were performed independently by two researchers (E.F., T.W.). Differences in opinion relating to inclusion and exclusion criteria were discussed until consensus was reached.The literature search was performed in the following databases: PubMed, Web of Science, Google Scholar (first 10 pages), Karlsruher Virtueller Katalog (database for diploma, master, state examination, bachelor, and master theses), diplom. de, dissonline, base-net Bielefeld Academic Search Engine, DART-Europe E-theses Portal, Open Access Theses and Dissertations, as well as in relevant the German language journals not listed in PubMed (Bewegungstherapie und Gesundheitssport, Deutsche Zeitschrift für Sportmedizin, Prävention und Gesundheitsförderung, Public Health Forum, Journal of Public Health, Der Kardiologe, MMW—Fortschritte der Medizin, Der Internist, Der Orthopäde, German Journal of Exercise and Sport Research—Sportwissenschaft, Zeitschrift für Allgemeinmedizin) for the period 2000–2019 in German and English using the search terms Bewegungsberatung, Rezept für Bewegung, physical activity counseling AND Germany, exercise prescription AND Germany (search terms linked with AND were considered together). In addition, the reference lists of the included sources were searched, and a forward reference search was performed.The following a priori inclusion criteria were defined: (1) studies on prevalence of routine PA counseling or use of Exercise on Prescription in primary care in Germany, (2) publication language English or German, (3) quantitative or qualitative studies, (4) peer reviewed and not peer reviewed (grey) literature. We excluded studies on short-term PA counseling interventions (i.e., non-routine PA counseling) and studies in which PA counseling did not take place in primary care, as well as studies on preventive counseling services in which the share of PA counseling could not be determined. Data extracted from the included studies are summarized in Table 1 and Table 2.The risk of bias was assessed using the 10-item instrument developed by Hoy and colleagues [40] for quantitative studies. The instrument addresses four domains of bias and provides a summary risk-of-bias assessment. The overall interrater agreement is 91% with a Kappa statistic of 0.82 [40]. Risk of bias in qualitative studies was assessed using the 10-item Critical Appraisal Skills Programme (CASP) checklist [41].The search yielded 626 records. After deduplication, we screened 587 titles and abstracts and reviewed 92 full texts subsequently. After applying the inclusion and exclusion criteria, 25 articles from 20 studies were included in the descriptive analysis [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39], cf. also Figure 1.Nineteen studies were quantitative, eight of which were conducted with patients, ten with physicians, and one study was based on patient records. Of the six qualitative studies, three were conducted with physicians and one with patients. In two studies, physician–patient discussions formed the data basis, cf. also Table 1 and Table 2. Four studies are grey literature [33,37,38,39].PA counseling [17,23,24] and the use of EoP per se [18,25,39] were primary research questions in three studies each. The remaining publications represent secondary research questions of other, usually more comprehensive studies, such as cardiovascular disease prevention in primary care [26] or the National Health Survey [15,16,20].Due to the great methodological diversity of the included studies, a meta-analysis was not feasible.The results of the methodological assessment are presented in Table 3 and Table 4. No study has given a formal definition of “physical activity” or “physical activity counseling”; various terms and periphrases were used instead. All quantitative studies that did not evaluate data in patient records used self-developed survey instruments (questionnaires), with one or more items for PA counseling or use of EoP. Physicians were typically invited to provide information on the prevalence of counseling using different level Likert scales. The overall sample of patients to whom the counseling prevalence refers varied and was not explicitly mentioned in every study. None of the physician surveys provided information on how inactive or insufficiently active patients were defined and identified. Patient surveys included questions on PA counseling and prescribing exercise in different past time-periods.Beyond data on prevalence, some studies provide information on the content and methods of counseling, such as recommendations for specific types of PA [23,37]; general information on the health benefits of PA [18,23]; recommendation on the frequency and intensity of PA (Kroll 2014); patients’ preferences [35]; and disease-related, individual exercise capacity [23]; use of written materials [24,29]; referral to group offers or to therapists [17,24,29,32]; written agreement on goals and follow-up [27,29], and motivational counseling [29].Two studies [26,29] and [24] have assessed physicians’ self-rated counseling competence and knowledge. The physicians report high to very high competences and at the same time express doubts that they can actually bring about behavior change in patients [23,24,26,29]. Similar views are also voiced in qualitative studies [34,39].Some studies have assessed barriers to routine PA counseling [23,24,25,39]. These included lack of remuneration, lack of time, patients’ disinterest and lack of compliance, lack of information, and lack of networking with partners outside the health care system [17,25,39].The effects of counseling or prescription of PA were assessed in three studies using non-validated self-reports with different follow-up periods [17,18,20]. No study has used objective measurement methods. Kroll documented the effects of counseling in her qualitative study [37].The first aim of this review was to present data on the prevalence of routine PA promotion in health care in Germany as comprehensively as possible. Our approach was that of an integrative review to “enhance a holistic understanding” of this topic [13]. The second aim was to offer and discuss findings on contents of and barriers to PA counseling. The great methodological diversity, which is inherent in the method of integrative reviews, and the substantial methodological limitations of the studies included make it difficult to draw a conclusive summary. Since to date no review on PA counseling in primary care in Germany has been published, we adopted an approach that allows for the synthesis of different perspectives on the topic. Thus, e.g., the juxtaposition of contrasting physician and patient reports adds a further dimension relative to presenting just “one side” [13].Physician-reported prevalence of counseling is high. The largest nationwide study, with over 4000 respondents, found that 71.8% of primary care physicians offered PA counseling to more than half of their patients [26]. Furthermore, more than 80% of neurologists surveyed in a nationwide study stated that they “frequently” counseled their patients on PA [23]. Moreover, 90% of the general practitioners surveyed in Berlin report offering PA counseling always or frequently if it is indicated [31]. General practitioners in and around the city of Würzburg also give recommendations on PA physical activity to 53.5% of older patients [24]. However, knowledge and use of EoP is limited: less than 8% of the physicians surveyed use it as part of their PA counseling [25] or do not use it in the intended sense [39].Some, but not all, of the patient-reported data seem to contradict those of the physicians. The representative data of the National Health Surveys show a considerably lower prevalence: 8.6% of patients between 18 and 64 years of age report having received PA counseling in the past 12 months [16]. According to the 1998 National Health Survey, the prevalence of counseling in the 18–79 age group was as low as 6.85% [20]. However, two smaller studies documented an almost fourfold (32.8%) [19] and sevenfold (48%) [17] prevalence of counseling, respectively, in older patients. In a sub-sample of the Leipzig Life Study, 21.5% of patients reported having received PA counseling from their primary care physician [22].The only study based on patient records found a counseling prevalence of 21.4% [33].Counseling prevalence seems to be higher in patients with diabetes [16,19] coronary heart disease [16,19], myocardial infarction, osteoarthritis, multi-medication [19], and hypertension [16] than in people without these conditions. These patient-reported data are consistent with those of physicians: physicians with a high proportion of high-risk patients seem to offer counseling more frequently [29]. These results are also in line with data from Sweden [42], the U.S. [43,44], and a systematic review [45].Current data from Germany provide little insight into how PA counseling is offered. It remains largely unknown whether counseling is based on a theory of behavior change, whether physicians use motivational techniques and, if so, which ones, how they define “inactivity”, for which patients they consider counseling to be indicated, how often follow-ups take place. These data would be of major interest when it comes to effectiveness, since though the specific intervention components associated with best result cannot be clearly defined, interventions that include multiple behavioral change strategies such as goal setting, written prescriptions, providing feedback, and follow up, seem to yield better outcomes [12].Primary care physicians’ attitudes and perceptions on PA counseling is very similar to those reported from other countries [45]. Physicians typically regard lifestyle counseling in general [26] and PA counseling in particular [25,39] as an important part of their routine as medical professionals, but face a number of barriers. Besides lack of time [23,24,39], patient-related factors such as disinterest, lack of motivation, and lack of compliance [23,24,25,39] are often reported to be important barriers to routine counseling.There seems to be a disconnect between physicians’ and patients’ perception of success in behavior change, which is very similar across countries. While physicians in Germany [23,24,25,34,39] and elsewhere [45] cite patients’ disinterest and reluctance to act upon advice as one of the major barriers to counseling, patients’ reports seem to at least to some extent contradict these relatively widespread assumptions. Indeed, several German studies show that patients value physicians’ advice. More than three-quarters of older patients stated that they had decided to keep up with an exercise course recommended by their family doctor, and 82% were generally more interested in a course if their family doctor recommended it [17]. More than half of the patients who received an EoP from their physician reported that they did more exercise and were more active in their everyday life [18]. In the National Health Survey, compliance rate upon counseling was 52% [20]. Appreciation of physicians’ support in increasing PA has been found in various countries and patient groups [46,47,48].Lack of remuneration for counseling is mentioned in every study that identified the barriers [23,24,25,26,29], but interestingly, it is not always considered the most important factor.The widespread call and advocacy for routine PA promotion in primary care notwithstanding there seems to be a paucity of current representative data on PA counseling prevalence. Representative patient-reported data indicate that in 2010 about one third of all U.S. patients who had seen a physician or other health professional in the previous 12 months had received advice on PA [43]. In a national sample, which was representative in some but not all relevant terms, 18.2% Australian adults reported having received PA counseling from their physician in the previous 12 months [49].In a nationwide Brazilian study, over 80% of physicians reported regularly providing PA counseling [50]. A nationally representative survey of primary care physicians in the United States found that 93.9% and 86% provide guidance on PA “often” or ”always” to patients with and without chronic diseases respectively [44]. In a national survey among Canadian primary care physicians, 85% of respondents reported asking their patients about PA, whereas only 15.8% provided written advice [51]. Similar rates have been reported from Ireland [52]; 88% of survey participants reported asking about PA, but the vast majority (82.6%) did not provide written prescription [52]. These findings collectively suggest considerably higher physician-reported prevalences than patient-reported ones.Based on electronic patient records, an EoP was issued to 3% of all patients in primary and secondary care in a Swedish County Council [42].Involving allied health care professionals, such as nurses, physiotherapists, or exercise scientists, into PA counseling in primary care is practice in some countries [53]. This interdisciplinary model has been shown to produce better result than physician-only approaches [53]. We could identify no study in Germany where professions other than physicians were involved. The less than optimal cooperation between professions and sectors was cited as a barrier in various studies [17,18,25,34,39]. We see improved interdisciplinary work as a key element to enhance the prevalence of PA counseling in primary care.Direct comparison between countries is challenging for various reasons. Assessment methods (self-report vs. patients’ records), data sources (patients vs. physicians), patient and physician characteristics differ in different countries. Interestingly, data showing that physicians tend to offer advice on PA more readily to already diseased populations than to currently healthy participants seems to be consistent across countries, data sources, and assessment methods [16,42,45,49]. Encouraging patients with chronic diseases and compromised health to be more physically active is very welcome. On the flipside, PA counseling seems to be underutilized as a preventive tool.To the best of our knowledge, this is the first study to give an overview of PA counseling in primary care in Germany. We have followed the strict criteria of the PRISMA recommendations. In order to provide the most comprehensive overview possible, we have included both quantitative and qualitative studies from peer reviewed and grey literature. At the same time, our review must be seen in the light of the limitations of the studies included.There are no widely accepted reporting schemes for survey studies, which leads to inconsistent reporting [54]. In the included studies, with a few exceptions, response rates were low, and most studies did not provide information on item non-response (complete vs. partial answers to the questions). We cannot exclude the possibility that the data presented here contain a positive bias. Self-selectivity may have played a role for both physicians and patients, and physicians may have indicated more frequent counseling activity (social desirability). Overall, the methodological limitations greatly reduce the generalizability of the results.Data on the prevalence of PA counseling in Germany vary according to data source and are sometimes contradictory. Direct comparison with other countries is challenging due to methodological issues. Perceived barriers to routine PA counseling in primary care seem to be very similar to those reported from other countries. To improve comparability among studies and to improve overall methodological quality, standardized instruments should be developed and validated. Surveys in representative samples using such instruments are needed to further knowledge on counseling and to be able to establish prevalence trends. Conducting studies on counseling methods and contents can add valuable information beyond prevalence. Strengthening the topics of physical activity and health and physical activity counseling in medical curriculum is strongly recommended.E.F., T.W., D.A.G., and W.B. conceptualized the review and developed the methodology. E.F. and T.W. performed the systematic literature search and data analysis. E.F. wrote the first draft. T.W., D.A.G., and W.B. performed review and editing. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The authors declare no conflicts of interest.Flow chart.Quantitative studies.1 PA—physical activity; 2 n.r.—not reported; 3 BGS 98—National Health Survey Bundesgesundheits survey 1998; 4 DEGS 1—First Wave of National Health Survey DEGS; 5 OR—odds ratio; 6 CI—confidence interval; 7 getABI Study—German epidemiological trial on ankle brachial index for elderly patients in family practice to detect peripheral arterial disease; 8 EUROPREVIEW Study—cross-sectional study conducted by the European Network for Prevention and Health Promotion in Family Medicine/General Practice; 9 ÄSP Study—Physician Survey on Cardiovascular Disease Prevention.Qualitative studies.Assessment of risk of bias in quantitative studies.Assessment of risk of bias in qualitative studies.

Med-MDPI/ijerph_5/ijerph-17-15-05626.txt

@@ -0,0 +1 @@
+Background: Tuberculosis (TB) is a major global public health problem and one of the leading causes of death among infectious diseases. Although TB can be cured with first-line antibiotics treatment of 6 months regimen, non-adherence to the treatment remains the main challenge for TB prevention and control. Interventions to promote adherence need to address multiple underlying factors linked to non-adherence, which requires a synthesis of studies to understand these factors in the local context. Our review accordingly examines these factors for TB treatment in Ethiopia. Methods: Articles were searched from PubMed and ScienceDirect databases, as well as manual searches through Google and Google Scholar search engines. Both quantitative and qualitative studies that showed factors associated with or reasons for non-adherence, default or loss to follow up from TB treatment were included. A total of 276 articles were screened, and 29 articles were ultimately included in the review. Findings: The extracted factors were synthesized thematically into seven dimensions of patient-centred, social, economic, health system, therapy, lifestyle, and geographic access factors. More than 20 distinct factors were identified under these headings. Some of these factors may also apply quite widely in other settings, with greater or lesser influence, but some are particularly applicable to the Ethiopian setting. Conclusion: Helping patients to achieve full adherence to TB medication is a complex problem as it is influenced by interplay between many factors. Healthcare managers, providers, and researchers need to consider and address multiple underlying factors when designing adherence interventions. This work provides a reference set of such factors for Ethiopian interventions.Tuberculosis (TB) is an infectious disease that is one of the major causes of death, being in the top ten causes of death worldwide and the leading cause of death from infectious disease, ranking above HIV/AIDS in 2018. Globally, one-fourth of the population is either infected with TB or at risk of developing the disease, with an estimated 10 million people infected with TB worldwide in 2018 [1]. Ethiopia has achieved a 50% reduction of TB through the Millennium Development Goals (MDGs) [2]. However, Ethiopia remains one of 14 high-burden TB countries for TB, TB-HIV, and Multi-drug-resistant tuberculosis (MDR-TB). An estimated 165,000 cases of TB incidence were reported for Ethiopia in 2018 [1].Although TB can be cured with first-line antibiotics treatment of 6 months regimen, non-adherence is the main challenge for TB control and prevention programs. The patient needs to take >90% of TB medication to facilitate TB cure, and a patient who takes at least 95% is said to be ‘high adherence’. Treatment default is defined by the World Health Organization as a patient who interrupts treatment for 2 or more months [3]. The default rate is thus a crude method to monitor adherence [4]. Non-adherence to TB treatment increases the risk of morbidity, mortality, and drug resistance at both the individual and community level [5].The World Health Organization has recommended Direct Observation of Treatment (DOT) by a trained supervisor, in which a healthcare worker watches the patient take the medication every day, to ensure adherence to treatment [6]. However, implementing DOT in Ethiopia is challenging for both the patient and healthcare provider. For example, one study conducted in Addis Ababa found patients reported that a daily visit to a health facility for the first two months was very difficult for a range of reasons, including severe illness at the initiation of treatment, distance too far for walking, and high transportation cost. Because of these challenges, DOT has not been implemented on a daily basis in Ethiopian standard care after the first two months of treatment [7].A systematic review found that the pooled prevalence of non-adherence to TB treatment in Ethiopia was 21.3%. Forgetfulness, fear of drug side-effect, waiting time for 1 h or more during the service, and feeling a long distance to health facility were identified as factors associated with this non-adherence [8]. Another systematic review in the same setting found that the pooled prevalence of non-adherence to TB treatment and loss to follow-up were 20% and 5%, respectively. Being TB-HIV-co-infected, transport costs, lack of knowledge, drug side-effect, educational status, forgetfulness, being in continuation phase, perceived physical and psychological barriers, and psychological distress were identified as associated factors [9].Interventions to promote adherence require addressing multiple components to overcome the barriers to adherence [10,11]. This requires a synthesis of studies to understand the causal factors for non-adherence to TB treatment in the local context. The above-mentioned systematic reviews and meta-analysis conducted in Ethiopia by Zegeye et al. (2019) [8] and Tolla et al. (2019) [9] had their main objective to estimate the pooled prevalence of non-adherence to TB treatment. The associated factors for non-adherence to TB treatment were not comprehensively identified by them, and their reviews included only quantitative studies.This literature review synthesises both qualitative and quantitative studies to thematically present multiple factors that have been identified as influencing non-adherence to TB treatment in Ethiopia. This includes all types of TB such as active TB, latent TB, and TB-HIV co-infected patients. Therefore, this review could help healthcare managers, providers, and researchers to design and implement adherence interventions based on established contextual factors rather than ad hoc generalisations.Research articles were searched for from PubMed and ScienceDirect databases, as well as manual search through Google and Google Scholar search engines. Search expressions were developed for TB medication adherence or loss to follow up or default from TB treatment that were published in the English language with no publication date restriction (see Table 1). The same expression of search strategy was used for all databases and search engines. Article searching was undertaken from 15 April to 5 May 2020.Both qualitative and quantitative research articles were included, and articles that did not report original research were excluded. Articles that did not assess factors or reasons associated with TB medication non-adherence or default or loss to follow up from TB treatment were excluded, following the protocol of Figure 1. Data were extracted using an Excel template comprising the author, year, region, sample size, study design, population, and major findings. The article selection and data extraction were performed by the first author of this paper, and consistency was checked by the two other authors. Factors associated with TB treatment non-adherence were extracted from the quantitative type of studies, and reasons for non-adherence or default or loss to follow up were extracted from the qualitative type of studies. The extracted data were synthesized into groups based on the seven thematic dimensions of TB medication adherence factors proposed by Ogundele et al. (2015) [12]. These seven thematic dimensions are patient-centred, social, economic, health system, therapy, lifestyle, and geographic access factors.A total of 276 research articles were screened, and 29 studies were ultimately included in this literature review. Of these, 6 articles were qualitative studies, 22 articles were quantitative studies (18 cross-sectional, 3 prospective cohort, and 1 case-control), and 1 article used a mixed-method design. These studies were conducted in Addis Ababa (9); Southern Nations, Nationalities, and Peoples’ Region (SNNPR) (7); Amhara (5), Oromia (5); and Tigray (3) regions of Ethiopia. Five studies were conducted among latent TB-HIV co-infected patients, and the remaining 24 studies were conducted among active TB infected patients. Approximately 7382 participants in total were involved in the studies reported in these 29 articles.The synthesised findings from both quantitative and qualitative studies are presented below, grouped in seven dimensions of adherence influencing factors. The individual studies that showed factors or reasons linked with TB medication non-adherence are presented in Appendix A Table A1.Forgetfulness [13,14,15,16,17,18,19,20,21] and inadequate knowledge about tuberculosis and its treatment regimen [14,18,22,23,24,25] were the two major patient-centred factors. Three studies conducted in Oromia [26,27] and SNNPR [21] regions showed that the patient’s educational status was associated with non-adherence to TB medications: the more the patient was educated, the less likely was non-adherence to TB medication. Psychological distress was another factor: two studies conducted in Addis Ababa reported that this indirectly positively influences non-adherence to TB medication [28,29]. Another qualitative study conducted in Addis Ababa also stated that poor mental health status of a patient would make them reluctant to regularly attend follow up and clinic appointments [18].Several studies reported that patients not getting social support from families and neighbours in remembering to take their medication, food, and financial assistance were the major social factors that influenced non-adherence to TB medication [18,19,24,26,30,31,32,33]. Additionally, one study in Addis Ababa conducted among latent TB-HIV co-infected patients reported that the patients’ friends’ decision to take the medication would make them less likely to be non-adherent to isoniazid preventive therapy (IPT) [34]. Another study among the same subjects and setting found that patients who were comfortable to take IPT in front of other people were less likely to be non-adherent [35]. Being busy with work [14] and away from home for work or other social-related activities were also found to influence non-adherence to TB medication [14,15,16]. Perceived and experienced stigma and discrimination also led the patient to non-adherence [18,36,37,38]. These particular factors were highly noted in studies conducted among TB-HIV co-infected patients [18,31,37,38]. As one study indicated, because of fear of stigma and discrimination, the patients were not disclosing their HIV status to their family, which in turn influenced their non-adherence to TB medication [18].Beliefs about the disease and treatment, such as perceived wellness or cure, perceived risk, and perceived barriers over the benefits, were influencing factors for non-adherence to TB medication [13,26,28,29]. One study conducted in Addis Ababa reported that a patient’s belief in curability and severity of TB in the presence of HIV infection would make them less likely to be non-adherent [31]. Another study in Addis Ababa found that the perceived risk of discontinuing TB medication was the reason for adherence, while perceived wellness was the reason for patients have intention to discontinue TB treatment [13]. One study conducted in SNNPR also reported that belief in traditional healing influenced non-adherence to TB medication [36].The patient’s economic constraints (which impact the financial burden) was the main economic factor that influences non-adherence to TB medication [17,29,30,31]. Economic constraints limit the patient’s ability to have adequate food which influences non-adherence [31,36]. The cost of medication other than TB medications is also a factor for non-adherence in one study conducted in SNNPR. Another study conducted in SNNPR reported that the patient being not employed was associated with non-adherence [39].Poor healthcare provider–patient relationship with communication gaps was a major factor that influenced non-adherence to TB medication [14,31,32,36,39]. For example, one study conducted in Addis Ababa among loss-to-follow-up patients reported that the healthcare providers were seen as disrespectful of their patients and less committed to their profession [32]. The quality of healthcare service and a patient’s satisfaction with healthcare service affect non-adherence to TB medication [16,30,39]. When patients perceived that they received less professional care and less time spent with the healthcare providers, and waited a long time to get healthcare service, they were more likely to be non-adherent [16,32,39,40]. Health information/education is also crucial for adherence: a few studies showed that the patients who did not receive health information/education from health facilities were more likely to be non-adherent [15,18,22]. Additionally, one study done in Addis Ababa found that cues to action were reported as a factor for non-adherence [28]. Lack of supervision and healthcare providers incapable of managing the patient’s illness were also reported as influencing factors for interruption and default from TB treatment [38].Many studies reported that drug side-effects were the major therapy-related reason for non-adherence to TB medication [15,16,23,24,26]. Pill burden was also reported as a factor for non-adherence to TB medication among active TB and TB-HIV co-infected patients [31,36]. The presence of more than one co-morbidity including TB-HIV co-infection was also reported as a factor for non-adherence to TB medication [14,16,20,26,37]. One study conducted in Addis Ababa also found that being on Antiretroviral Therapy (ART) was a factor for non-adherence to TB medication [29]. Symptom presence after initiation of anti-TB treatment and slow progression of the health status were also found as non-adherence factors [19,20,26,33]. Being in the continuation phase of the treatment (after the initial 2-month clinic-based treatment period) was a factor for non-adherence and default [14,20,24,34,41]. This might be due to the patient’s perceived wellness or cure because a daily DOT was not implemented after the first two months of treatment in Ethiopia.Alcohol consumption was reported as a factor that influenced non-adherence to TB medication in several studies [14,16,27,29]. Cigarette smoking and khat (herbal stimulant) chewing were also found as factors associated with non-adherence [16,27].Healthcare inaccessibility from residence location was a major geographical access factor for non-adherence and default from TB treatment [17,19,22,26,30,36,40]. Due to inaccessibility, patients were unable to keep regular clinic appointments and follow up treatment in two studies, conducted in Addis Ababa and Amhara regions [16,35]. Distance of the health facility was related to transportation cost, which was also a factor for non-adherence [21,22,25,33].Adherence to TB medication is a complex and dynamic matter as it is affected by multiple factors. This review has identified the range of multiple factors that have been found to affect non-adherence to TB medications in Ethiopia. The influence of these factors individually and in combination might vary from one social or geographic setting to the other. Healthcare managers need to consider the underlying factors for non-adherence to TB medication in the local setting (ideally using locally available evidence) when they design and implement an intervention.Our literature review has found some similar influencing factors of non-adherence to TB treatment as those described in the previous two systematic reviews conducted in Ethiopia [8,9]. These factors were forgetfulness, inadequate knowledge about TB and its treatment regimen, psychological distress (poor mental health condition), perceived barriers, long waiting time, drug side-effects, TB-HIV co-infection, being on the continuation phase of treatment, healthcare inaccessibility, and travelling costs. Adherence interventions such as providing health information about TB and its treatment regimen, dealing with side-effects, providing reminders, and other health system interventions could be used to resolve these factors.However, in this review, we have found several additional factors of non-adherence to TB treatment that were not identified by the previous two systematic reviews. These factors were lack of social support, being busy with work, being away from home, perceived and experienced stigma and discrimination, beliefs such as perceived wellness/cured, perceived risk, financial constraints to buy food and medication cost other than anti-TB, poor healthcare provider–patient relationship such as communication gaps, disrespecting patients, quality of healthcare service, patient satisfaction, lack of health information/education, pill burden, the persistence of symptoms after initiation of treatment, and use of substances such as alcohol, smoking, and khat chewing. These factors and the possible interventions to overcome the factors are discussed below.We have found that lack of social support can influence non-adherence to TB medication. Thus, providing social support such as financial assistance for transport costs, food assistance, and reminders for medication intake from families and neighbours may help the patient to adhere to TB medication. In the same way, not getting social acceptance to take medication from families, friends, and neighbours can lead to non-adherence. A systematic review in developing countries reported the same finding [42]. Similarly, the perceived stigma and discrimination were reported as factors for non-adherence to TB medication. These were higher especially among TB-HIV co-infected patients. The other new social factors were being busy with work and away from home for work and social-related activities. These factors influence non-adherence might be due to the patient forgetting to take their medication when they are busy and away from home. A reminder from family or through mobile SMS text could be considered as a solution for these factors. Beliefs related to TB and its treatment can also influence non-adherence. This might be due to the patient’s perceived wellness or cure after taking some medications and thus interrupting their treatment. On the other hand, when the patient perceives that the disease is severe and the risk of discontinuing TB medication leads to poor health outcomes, their non-adherence to TB medication would be less likely. A systematic review conducted in developing countries also found that feeling well after the initiation of treatment was a factor for non-adherence to TB medication [42]. Providing patients with health information to change such wrong beliefs about the disease and its treatment could be considered as an adherence intervention to address these factors.The financial constraints to get adequate food, transport cost, and medication cost other than anti-TB medication were found to influence non-adherence or loss to follow up from TB treatment. A systematic review conducted in developing countries also reported that the financial burden can cause TB medication non-adherence [42]. Financial assistance or support of food and transport and making all medications freely available to the patient may overcome this factor.We also found that healthcare-system-related factors were other major factors for non-adherence to TB medication. Poor healthcare-provider relationship with communication gaps, disrespect of the patient, and lack of professional commitment influenced non-adherence to TB medication. A systematic review conducted in developing countries also found that poor patient–healthcare worker communication was a factor for non-adherence to TB medication [42]. The quality of health service as perceived by the patient and the patient satisfaction also influenced non-adherence to TB medication. It was also found that patients who did not receive health information/education were more likely to be non-adherent to TB medication. Therefore, health system interventions such as training for healthcare provider–patient communication and their relationships and health system strengthening to shorten waiting times and to raise the quality of health services could address these factors.Pill burden was also a factor for non-adherence to TB medication both in TB and TB-HIV co-infected patients. TB-HIV co-infection has additional burden from the pills and drug side effects. Persistence of symptomatic conditions after initiation of TB treatment was another influence for non-adherence. This might be due to the patient’s belief in the curability of the disease becoming less when symptoms persist despite the treatment initiation. Health information about TB, its treatment, and side effects could reduce the influence of these factors. New drug investigations to reduce the drug side effect and to make a shorter treatment regimen could also be considered if it is possible.Substance use such as alcohol use, cigarette smoking, and khat chewing was a factor for non-adherence to TB medication as reported by a few studies. The use of these substances might make the patient reluctant to follow the regular clinic appointment and follow up treatment. Other systematic reviews conducted in developing countries also showed that alcohol and tobacco were factors for non-adherence to TB medication [42]. Health information on how the use of substances affected treatment adherence and treatment outcome may help to resolve this factor.This review included studies conducted in Ethiopia; therefore the identified factors may not be generalizable to other settings. Adherence to TB medication is a complex problem as it is influenced by multiple factors so a single factor may not be shown as a cause–effect relationship.This review describes more than 20 factors that influence adherence to TB treatment in Ethiopia, demonstrating that it is a complex problem that is affected by the interplay of multiple factors. We have found major additional factors for TB medication non-adherence or default or loss to follow up. These were social support from families and neighbours such as food support, reminders, and encouragement; being busy with work; being away from home; perceived and experienced stigma and discrimination; beliefs such as perceived wellness/cure; perceived risk; economic constraints for having adequate food and medication cost other than anti-TB medication; poor healthcare provider-patient relationships such as communication gaps, disrespecting patients, quality healthcare service, and patient satisfaction; health information/education, pill burden, the persistence of symptoms after treatment initiation; and use of substances. Healthcare managers, providers, and researchers need to address these underlying factors when they design and implement adherence interventions.Adherence to a medication regimen is defined by Osterberg and Blaschke (2005) as “the extent to which patients take medications as prescribed by their health care providers”(10).Z.S.N., L.P.-L. and A.J.M. contributed to the conceptual design, data extraction, synthesis and analysis, and preparation of the manuscript. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The authors received advice from Flinders University library in developing the protocol.The authors declare no conflict of interest.Summary of reviewed articles showing factors or reasons linked with TB medication adherence or default or loss to follow up from TB treatment.SNNPR–Southern Nations, Nationalities, and Peoples’ Region, TB–Tuberculosis.Flow chart for selection of reviewed articles.Searching strategy.

Med-MDPI/ijerph_5/ijerph-17-15-05627.txt

@@ -0,0 +1 @@
+The outbreak and worldwide spread of COVID-19 has resulted in a high prevalence of mental health problems in China and other countries. This was a cross-sectional study conducted using an online survey and face-to-face interviews to assess mental health problems and the associated factors among Chinese citizens with income losses exposed to COVID-19. The degrees of the depression, anxiety, insomnia, and distress symptoms of our participants were assessed using the Chinese versions of the Patient Health Questionnaire-9 (PHQ-9), the Generalized Anxiety Disorder-7 (GAD-7), the Insomnia Severity Index-7 (ISI-7), and the revised 7-item Impact of Event Scale (IES-7) scales, respectively, which found that the prevalence rates of depression, anxiety, insomnia, and distress caused by COVID-19 were 45.5%, 49.5%, 30.9%, and 68.1%, respectively. Multivariable logistic regression analysis was performed to identify factors associated with mental health outcomes among workers with income losses during COVID-19. Participants working in Hubei province with heavy income losses, especially pregnant women, were found to have a high risk of developing unfavorable mental health symptoms and may need psychological support or interventions.At the end of December 2019, the Chinese city of Wuhan reported a novel pneumonia caused by coronavirus disease 2019 (COVID-19), an infectious disease caused by an acute severe respiratory syndrome coronavirus, which is rapidly spreading both domestically and internationally [1,2]. On 30 January 2020, the World Health Organization (WHO) held an emergency meeting and declared the worldwide COVID-19 outbreak a public health emergency of international concern [3]. The emergence and rapid increase in the number of COVID-19 cases has posed and continues to pose complex challenges for global research, public health, and medical communities [4,5]. As of 1 June 2020, there were more than 6.15 million confirmed cases of COVID-19 across more than 215 countries and regions, including more than 372,130 deaths.With the rapid spread of COVID-19, the local government in Wuhan immediately adopted a city closure policy, encouraging citizens to work at home and teach online, and shut down non-essential services to mitigate the impact and risks of the disease. Then, the governments of other provinces with low numbers of infected people in China and many other countries around the world entered states of emergency for the health response and issued a series of policies, including ordering citizens (regardless of having symptoms of infection or not) to self-isolate at home, and maintaining social distance from other people. However, concerns have arisen about the potential psychological impact of these measures [6,7,8]. Studies proved that COVID-19 has caused a high prevalence of mental health problems in China [8,9,10,11,12] and other countries around the world [13,14,15,16]. Some researchers have attempted to understand the outbreak of this novel coronavirus from a global health perspective [17,18,19]. However, most studies focused on the psychological effects of people who were infected with COVID-19, medical workers, or people in specific regions [10,11,12,13,14,15,20]. Studies showed that the economic impact caused by severe acute respiratory syndrome (SARS) will produce psychological morbidities in individuals who are directly or indirectly exposed to life-threatening situations [21]. The occurrence of such psychological morbidities among workers can impact their daily functions and lead to immediate economic and physiological consequences, such as lost job productivity, depression, and anxiety [22,23]. To the best of our knowledge, no previous study focused on mental health problems among people with income losses caused by COVID-19. To address this gap, the aim of our study was to evaluate the mental health of Chinese workers with income losses exposed to COVID-19 by quantifying the degrees of depression, anxiety, insomnia, and distress, and analyze the potential risk factors related to these symptoms. In this study, besides age, sex and other demographic characteristics, participants from Hubei province and outside Hubei province were taken as the research objects for comparison of regional differences. The ultimate goal of this study was to assess the mental health burden of people with income losses during COVID-19 and to provide guidance for the promotion of mental well-being among this population.This was a cross-sectional study conducted using an online survey and face-to-face interviews to assess mental health problems and their associations with income losses among Chinese citizens who were exposed to coronavirus disease 2019 (COVID-19) from 25 April to 9 May 2020. Eligibility criteria included (i) currently living in China, (ii) aged 18 years or older, and (iii) with income losses caused by COVID-19. Participants were encouraged to participate in online surveys or complete offline questionnaires. A total of 421 of 600 contacted individuals completed the survey for a participation rate of 70.2%, and 23 people with no loss of income were excluded from the study. The final sample included 398 respondents, with a response rate of 66.3%. This study was approved by the Ethics Committee and Institutional Review Board of Wuhan University, Wuhan, China (Ref: 20200411), and conducted in accordance with the ethical guidelines of the Declaration of Helsinki of the World Medical Association. All data were deidentified before being provided to the investigators. Consent from each participant was obtained at the beginning of the survey.The questionnaire consisted of 37 factors to record demographic indicators and symptoms of depression, anxiety, insomnia, and distress caused by COVID-19 of the participants (See Appendix A).The following demographic data were included in this study: sex (male or female), age (18–25, 26–30, 31–40 and >40 years old categories), educational level (<undergraduate, college, masters, and higher), marital status (married and other including unmarried, widowed, and divorced), working location (Hubei province, and outside Hubei province), loss of income caused by COVID-19 (light, middle, or heavy, >0% to 25%, 25–50%, and >50% less than pre-epidemic income, respectively), and place of residence (urban or rural).Mental disorders, including depression, anxiety, insomnia, and distress, caused by COVID-19 were assessed in our study by Chinese versions of validated measurement tools [24,25,26,27]: the Patient Health Questionnaire-9 (PHQ-9; the total score ranged from 0 to 27) [24], the Generalized Anxiety Disorder-7 (GAD-7; the total score ranged from 0 to 21) [25], the Insomnia Severity Index-7 (ISI-7; the total score ranged from 0 to 28) [26], and the revised 7-item Impact of Event Scale (IES-7; the total score ranged from 0 to 28) [27]. The response options are: 3 = nearly every day, 2 = more than half the days, 1 = several days, and 0 = not at all for PHQ-9 and GAD-7; 4 = always, 3 = often, 2 = sometimes, 1 = rare, and 0 = never for ISI-7 and IES-7. The total scores of these survey scales are interpreted as follows: PHQ-9, extremely severe (22–28), severe (15–21), moderate (10–14), mild (5–9), and normal (0–4) depression; GAD-7, severe (15–21), moderate (10–14), mild (5–9), and normal (0–4) anxiety; ISI-7, severe (22–28), moderate (15–21), subthreshold (8–14), normal (0–7) insomnia; and IES-7 severe (22–28), moderate (15–21), subthreshold (8–14), and normal (0–7) distress. The cutoff score for detecting possible major symptoms of depression, anxiety, insomnia, and distress caused by COVID-19 are 10, 10, 15, and 15, respectively. A higher score indicates participants with greater self-reported severe symptoms [24,25,26,27]. The psychometric properties and internal reliabilities of the 4 scales have been previously confirmed in Chinese populations [24,25,26,27]. In [24], statistical tests were performed to determine the reliability and validity of PHQ-9. Results showed that the internal consistency value of PHQ-9 was 0.854 and the test–retest reliability value of PHQ-9 was 0.873, proving the PHQ-9 is a valid and reliable tool to evaluate depression in Chinese people. He [25] tested the reliability and validity of Chinese version of GAD-7. The results show that the Cronbach ‘α coefficient of GAD-7 is 0.898, and the test–retest reliability coefficient is 0.856, proving the Chinese version of GAD-7 has good reliability and validity in the application of evaluating anxiety. Doris S.F. Yu [26] tested the reliability and validity of Chinese version of ISI-7, finding that Cronbach’s alpha of the Chinese version of the ISI-7 was 0.81, with item-to-total correlations in the range of 0.34–0.67. In [27], Chan reported that the Cronbach ‘α coefficient of IES-R is 0.89, which proved the IES-R is a valid and reliable tool to evaluate distress among Chinese people. In our study, the Cronbach’s alpha coefficient of our questionnaire is 0.97. The Cronbach’s alpha coefficients of the Chinese versions of PHQ-9, GAD-7, ISI-7 and IES-7 were 0.920, 0.945, 0.879 and 0.909, respectively.First, we used descriptive statistics to describe the socio-demographic characteristics of these participants. Second, the prevalence rates of depression (PHQ-9 score ≥ 5), anxiety (GAD-7 score ≥ 5), insomnia (ISI-7 score ≥ 8), and distress (IES-7 score ≥ 8) were estimated. Finally, multivariable logistic regression models were used to explore factors associated with depression, anxiety, insomnia, and distress among workers with income losses exposed to COVID-19 in China, and the associations between risk factors and outcomes are presented as adjusted odds ratios (aORs) with a 95% confidence interval (CI), after adjustment for confounders, including sex, age, marital status, educational level, working position, place of residence, degrees of income losses. Data analysis was performed by SPSS statistical software (version 25.0, IBM Corp., Armonk, NY, USA,), with p-values < 0.05 indicating statistical significance. The significance level was set at α = 0.05, and all tests were two-tailed.As shown in Table 1, the proportion of men to women was close, at 50.5% and 49.5%, respectively, and the proportion of marital status (recoded into married and other including unmarried, widowed, and divorced) was similar to that of sex, at 49.5% and 50.5%, respectively. We classified their income losses caused by COVID-19 as one of the demographic variables. Response options were slightly affected (>0% to 25%), moderately affected (25–50%), and heavily affected (>50%).Table 1 shows that the proportions of light, middle, and heavy income loss (>0% to 25%, 25–50%, and >50% lower income than pre-epidemic income, respectively) caused by COVID-19 were 33.9%, 17.6%, and 48.5%, respectively. As Hubei was most severely affected province by COVID-19 in China, all 398 participants were grouped by their geographic location. The proportions in Hubei province, and places outside Hubei province were 44.2%, and 55.8%, respectively. Most of these participants were aged from 26 to 40 years, lived in urban areas, and had a college degree or above.Generally consistent with the existing COVID-19 research results [8,9,10], the prevalence rates of our participants who had symptoms of depression, anxiety, insomnia, and distress cause by COVID-19 were 45.5%, 49.5%, 30.9%, and 68.1%, respectively. As shown in Table 2, multivariable logistic regression analyses showed that, after controlling for covariates, the adjusted odds of depression, anxiety, insomnia and distress were lower among participants who under 30 years (e.g., depression among participants aged 26–30 years: OR = 0.228, 95% CI: 0.097–0.535, p < 0.001; depression among participants aged 18–25 years: OR = 0.187, 95% CI: 0.072–0.489, p < 0.001) compared with who aged over 40 years, and greater among those working in Hubei province (e.g., depression: OR = 2.647, 95% CI: 1.662–4.217, p < 0.001) than outside Hubei province. For the population whose income was heavily affected by COVID-19, they were prone to experiencing mental symptoms of depression, anxiety, and insomnia (e.g., depression among participants with light income losses: OR = 0.215, 95% CI: 0.124–0.371, p < 0.001). Those from urban area had lower adjusted odds of depression anxiety, insomnia and distress than those from rural area (e.g., depression: OR = 0.391, 95% CI: 0.226–0.675, p = 0.001). At the same time, being married (OR, 3.348; 95% CI, 1.896–5.911; p < 0.001) was associated with a greater risk of feeling depressed than being unmarried. In sex statistics, we set an additional question (If you are a woman, please indicate whether you are pregnant). In this study, as shown in Table 3, multivariable logistic regression analyses showed that, after controlling for covariates, we found that pregnant women with income losses during COVID-19 were associated with a greater risk of feeling depressed and anxiety (depression: OR = 2.956, 95% CI: 1.208–7.229, p = 0.018; anxiety: OR = 3.146, 95% CI: 1.217–6.133, p = 0.018) than unpregnant women (Table 3).According to Lai, J et al. [10], the cutoff scores for detecting possible major symptoms of depression, anxiety, insomnia, and distress caused by COVID-19 are 10, 10, 15, and 15, respectively. Thus, the prevalence rates of our participants who had severe mental symptoms of depression, anxiety, insomnia, and distress were 19.1%, 21.9%, 7.8%, and 25.9%, respectively. Similar to findings regarding prevalence of mental symptoms, as shown in Table 4, multivariable logistic regression analyses showed that, after controlling for covariates, the adjusted odds of severe symptoms of depression, anxiety, and distress were lower among participants who aged 26–30 years (e.g., severe depression: OR = 0.243, 95% CI: 0.091–0.645, p = 0.005) compared with who aged over 40 years, greater among those with heavy income losses than light and middle income losses (e.g., severe depression among participants with light income losses: OR = 0.246, 95% CI: 0.121–0.502, p < 0.001), and lower among those from urban area than those from rural area (e.g., severe depression: OR = 0.337, 95% CI: 0.185–0.615, p < 0.001). For those working in Hubei province, they were more prone to experiencing severe mental symptoms of anxiety and distress than those working outside Hubei province.We enrolled 398 respondents and found a high prevalence of mental health symptoms among workers with income losses caused by COVID-19 in China. This latest national sample indicated the prevalence rates of any disorder (excluding dementia), anxiety disorders, and depressive disorders were 16.6%, 7.6%, and 6.9% in China, respectively. Compared with national data, we found much higher prevalence rates of participants with symptoms of depression, anxiety, insomnia, and distress caused by COVID-19, at 45.5%, 49.5%, 30.9%, and 68.1%, respectively. Our findings are consistent with those of previous COVID-19 studies, including a study in mainland China that found that the prevalence of depression as measured during the COVID-19 pandemic was 48.3% [8] and a study in Hong Kong that found that the prevalence of depression caused by COVID-19 was 49.8% [9].Mental disorders, including depression, anxiety, insomnia, and distress, caused by COVID-19 were assessed in our study by Chinese versions of validated measurement tools [24,25,26,27]: PHQ-9, GAD-7, and ISI-7. In our study, the Cronbach’s alpha coefficient of our questionnaire is 0.97. The Cronbach’s alpha coefficients of the Chinese versions of PHQ-9, GAD-7, ISI-7 and IES-7 were 0.920, 0.945, 0.879 and 0.909, respectively, proving these scales have good reliabilities and validities in the application of evaluating mental disorders among Chinese worker with income losses. By reviewing the literature, we found that these Chinese scales are widely used in the study of psychological problems. Especially recently, these four scales have been used to study COVID-19. For example, researchers used them to assess the magnitude of mental health outcomes among healthcare workers treating patients exposed to COVID-19 in China [10], PHQ-9 and GAD-7 were used to evaluate depression and anxiety in Hong Kong during the COVID-19 pandemic [9], and GAD-7 was used to assess the prevalence of mental health problems and examine their association with social media exposure [8].In this study, besides age, sex and other demographic characteristics, participants from Hubei province and outside Hubei province were taken as the research objects for comparison of regional differences. The proportions of respondents from Hubei province and places outside Hubei province were 44.2% and 55.8%, respectively. The proportions of light, middle, and heavy losses of income (>0 to 25%, 25–50%, and >50% less income than pre-epidemic levels, respectively) caused by COVID-19 were 33.9%, 17.6%, and 48.5%, respectively. Most of these participants were aged from 26 to 40 years, lived in urban areas, and had a college degree or above. We found that workers with heavy income losses caused by COVID-19 reported more symptoms of depression, anxiety, and insomnia. Compared with participants outside Hubei province, those in Hubei province reported higher scores on all four scales. The prevalence rates of our participants who had severe mental symptoms of depression, anxiety, insomnia, and distress were 19.1%, 21.9%, 7.8%, and 25.9%, respectively. Our findings further indicated that pregnant women scored higher than non-pregnant women on PHQ-9 and GAD-7 measuring symptoms of depression and anxiety. These findings are consistent with the previous studies’ findings that exposure to a public health emergency can cause mental health problems.This study has several limitations. First, it was limited in scope. Almost half of the participants (44.2%) were from Hubei province, limiting the generalization of our findings to less affected regions. This survey was mainly conducted online, so some respondent bias, such as few elder citizens’ participation, may have affected the results. Second, the survey was conducted over two weeks and lacked longitudinal follow-up. It was hard to determine whether the mental health symptoms of workers with income losses could become more severe, so the long-term psychological implications of this population are worth further investigation. Last, although the response rate of this study was 70.1%, response bias may still exist if the non-respondents were either too stressed to respond or not at all stressed and therefore not interested in this survey.In conclusion, our findings showed that relatively high prevalence rates of symptoms of depression, anxiety, insomnia, and distress were caused by COVID-19. The prevalence of mental health problems among workers caused by COVID-19 in China is high, especially those working in Hubei province with heavy income losses. In addition, pregnant women with income losses were associated with a greater risk of feeling depressed and anxiety than other women, and may need psychological support or interventions. These results further indicate that the long-term psychological implications of this population are worth further investigation.The following are available online at https://www.mdpi.com/1660-4601/17/15/5627/s1, Table S1: Prevalence of anxiety and associated factors, Table S2: Prevalence of insomnia and associated factors, Table S3: Prevalence of distress and associated factors, Table S4: Prevalence of severe anxiety and associated factors, Table S5: Prevalence of severe insomnia and associated factors, Table S6: Prevalence of severe distress and associated factors.Conceptualization, X.L.; data curation, X.L. and P.L.; software, X.L. and L.H.; validation, X.L., P.L., T.H. and L.H.; investigation, X.L., T.H. and P.L.; writing—review and editing, X.L., and L.L.; visualization, X.L.; supervision, L.L.; project administration L.L. 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. 61772375, 18ZDA325]; Hubei Provincial Natural Science Foundation of China [No. 2019CFA025]; National Key R&D Program of China (2019YFC010167); and Independent Research Project of School of Information Management Wuhan University (No: 413100032).The authors declare no conflicts of interest.The questionnaire consisted of 37 questions to record demographic indicators and symptoms of depression, anxiety, insomnia, and distress of all participants.Demographic dataThe following demographic data were included in this study: sex (male or female), age (18–25, 26–30, 31–40, or >40 years categories), educational level (<undergraduate, college, masters, or higher), marital status (recoded into married or other including unmarried, widowed, and divorced), working location (Hubei province or outside Hubei province), loss of income caused by COVID-19 (light, middle, and heavy, being >0 to 25%, 25–50%, and >50% less income than the pre-epidemic level, respectively), and place of residence (urban or rural).Mental Health ScalesThe English versions of the PHQ-9, GAD-7, ISI-7, and IES–R-7 scales were used in this study to measure the degree of symptoms of depression, anxiety, insomnia, and distress of our participants.Patient Health Questionnaire-9 (PHQ-9).General Anxiety Disorder-7 (GAD-7).Insomnia Severity Index-7 (ISI-7).Revised Event Impact Scale (IES-7).Demographic and occupational characteristics of participants.Prevalence of depression and associated factors.Table 2 lists the detailed results of PHQ-9 from multivariable logistic regression analysis; the results for the other scales are presented in Supplementary Materials (Tables S1–S3). Abbreviations: NA = Not Available; aOR: adjusted odds ratio; CI: confidence interval. PHQ-9: the Patient Health Questionnaire-9.Prevalence rates of mental symptoms and associated factors in female cohort.Abbreviations: NA = Not Available; aOR: adjusted odds ratio; CI: confidence interval. PHQ-9: the Patient Health Questionnaire-9; GAD-7: the Generalized Anxiety Disorder-7; ISI-7: the Insomnia Severity Index-7; IES-7: the revised 7-item Impact of Event Scale.Prevalence of severe depression and associated factors.Table 4 lists the detailed results of PHQ-9 from multivariable logistic regression analysis; the results for the other scales are presented in Supplementary Materials (Tables S4–S6). Abbreviations: NA = Not Available; aOR: adjusted odds ratio; CI: confidence interval. PHQ-9: the Patient Health Questionnaire-9.

Med-MDPI/ijerph_5/ijerph-17-15-05628.txt

@@ -0,0 +1 @@
+Background: Light pollution is increasingly an area of concern for health and quality of life research. Somewhat surprisingly, there are relatively few descriptions of perceptions of light pollution in the literature. The current study examined such perceptions in a Irish sample. Methods: A survey was circulated as part of a citizen science initiative of a national newspaper; the survey included questions regarding night sky brightness and the impact of light at night on sleep and animal behaviour. Complete responses from 462 respondents were analysed. Results: Urban location was, as anticipated, associated with reported brighter night skies, and public lighting was reported as the main source of light at night for urban settings, whilst neighbours’ domestic lighting was the most commonly reported source for rural settings. Respondents from rural settings were more likely to report that light at night impinged on sleep, whilst city dwellers were more likely to report recent changes in wildlife behaviour. Conclusions: Citizen science approaches may be useful in gathering data on public perceptions of light pollution and its impacts. In the current study, this perception was strongly influenced by location, highlighting the importance of assessing experiences and attitudes across a number of geographical settings.Man-made light pollution is an area of increasing concern from a sustainability, ecological and health perspective [1]. A recent global survey of light pollution reveals that at least 80% of the world is exposed to significant levels of artificial illumination at night [2]. Health and ecological concerns centre on the potential of light at night to act physiologically to disrupt homeostatic and behavioural control systems, such as the circadian clock that regulates daily rhythms in activity, physiology and sleep [3]. Circadian rhythms and factors that disrupt them, such as man-made lighting, are recognized as important intrinsic and environmental determinants of health [4]. A number of studies have implicated man-made artificial light at night (ALAN) with health concerns such as increased risk of hormone-dependent cancers [5] and mood disorders [6]. ALAN has also been associated with changes in wildlife behaviour due to light-induced changes in circadian phases, leading to alterations of timing of rest/activity cycles, or direct actions of light on behavioural cycles independent of circadian effects [7]. Recent evidence has indicated that, for humans, the physiological response to nocturnal light may show very high inter-individual variability, with some people responsive to very low illuminance levels [8].Aside from the potential for direct physiological effects, the subjective perception of artificial light at night may also impact on quality of life and health-related behaviours, and may be shaped by psychological processes, such as social amplification [9]. Further, sleep disorders such as insomnia, or even subclinical poor quality sleep, have been associated with sleep attentional biases, wherein greater attention is drawn to sleep-salient factors in the environment [10]. As such, the presence of environmental light pollution has the potential to differentially and detrimentally impact on those with already poor quality sleep or sleep disorders; therefore, subjective perceptions of light pollution may shape how impactful it is for individuals’ sleep health. To date, there are surprisingly few reports in the literature on the subjective perception of ALAN and its impacts. One such study from Finland reported that light pollution was considered a nuisance for outdoors recreation, with over half of respondents reporting that light pollution reduced the overall quality of life of their neighbourhoods [11].In this study, we report the results of a citizen science survey of experiences of light pollution in Ireland, and examine geographic and demographic features that may influence such perceptions. This study addresses an important gap in the literature regarding public perceptions of the prevalence and intrusiveness of artificial light at night.Between March and June 2018, a brief 12 item questionnaire on light pollution was circulated via the citizen science initiative at “The Irish Times”, a national newspaper with a broad circulation (https://www.irishtimes.com/news/science/citizen-science/help-scientists-understand-the-influence-of-light-on-the-environment-1.3416898). As such, the sampling method applied was convenience sampling, and generalisability of findings from this sample was not assumed. The items on the survey asked about the nature of the home location, age, gender, a question about sky brightness at night, a question about the main source of man-made light, and five questions scored on a 7-point Likert-like scale (from “strongly disagree” to “strongly agree”) relating to perceptions of recent increase in light at night, the impact of light at night on sleep, changes in the timing of bird song, changes in the night time behaviour of animals and changes in the number of bats seen (Supplementary Materials for the full questionnaire used). The questionnaire was developed collaboratively by the authors to reflect their combined interests in light pollution, sleep health and ecology, and was designed to be appropriate for a citizen science approach. Data were fully anonymised at the point of collection, and geolocation data were not collected.For data analysis, responses that indicated “no opinion/not applicable” were removed and Likert-like responses on 7-point scales (1 = strongly disagree; 7 = strongly agree) were analysed as ordinal data using non-parametric tests. As appropriate, pairwise comparisons between groups were conducted with Bonferroni-adjusted two-sided Mann–Whitney U scores. Correlational analysis was conducted using Spearman’s Rho for ordinal variables. Associations between categorical responses were tested with Pearson’s chi-square test. p < 0.05 was interpreted as indicating a statistically significant effect. All data were analysed in SPSS (IBM Corporation, Armonk, NY, USA). The statistical approach employed is exploratory, and not hypothesis testing.A total of 464 respondents completed the survey; brief demographics of the study sample are presented in Table 1.Only 7.6% of respondents reported that the sky in their locale was completely dark, with 33.8% reporting regular visibility of the Milky Way; 38.5% reported visibility of only a few stars and not of the Milky Way, and 18.2% reported visibility only of the moon and the brighter planets (Figure 1). There was a significant large effect of location on the self-reported darkness of the night sky, with urban location being associated with subjectively brighter skies (Pearson’s chi-square = 314, df = 16, p < 0.001; Figure 1A). With regards to the brightest light source near to the residence, 61.9% that it was public lighting, 18.6% reported that this was their own domestic lighting, 12.8% that it was their neighbours’ lighting, 5.2% that it was commercial lighting, and 1.5% that it was passing traffic. There was a strong association of location with the source of lighting (Pearson’s chi-square = 190, df = 16, p < 0.001; Figure 1B), with own domestic lighting and neighbours’ lighting being important sources only in rural settings, and public lighting being the most important reported source across all settings. No statistically significant associations were found between gender or age group and the sources of light near the residence.For items scored on a Likert-like scale relating to perceptions of possible impacts of light pollution, there was an overall neutral response to the item “The level of lighting near my home at night has increased over the past three years” (median response = 4); a disagreement with the statement “If light enters my bedroom at night it does not affect my sleep” (median response = 2); and neutral responses to “Birds sing at night (the dawn chorus starts earlier than it used to)” (median response = 4), “The natural night-time behaviour of insects/bats/foxes, etc., remains the same as in previous years” (median response = 4) and “The number of bats I see has increased recently” (median response = 3). We then examined these ratings across three groups for location (i.e., rural, town and city). There was no effect of location on ratings of the item “The level of lighting near my home at night has increased over the past three years” (Kruskal–Wallis H = 2.65, p = 0.266; Figure 2A). For the item “‘If light enters my bedroom at night it does not affect my sleep”, there was an effect of location (Kruskal–Wallis H = 7.74, p = 0.021; Figure 2B), with city dwellers endorsing this statement more strongly than rural dwellers. For the item “‘Birds sing at night (the dawn chorus starts earlier than it used to)”, there was a significant effect of location (Kruskal–Wallis H = 44.3, p < 0.001; Figure 2), with city inhabitant endorsing this statement most strongly. Likewise, there were statistically significant effects of location for the item “The natural night-time behaviour of insects/bats/foxes, etc., remains the same as in previous years” (Kruskal–Wallis H = 44.3, p < 0.001; Figure 2D, city inhabitants endorse this item the least) and a marginal effect of location for the item “‘The number of bats I see has increased recently” (Kruskal–Wallis H = 6.1, p = 0.049; Figure 2E, city dwellers endorse this item the least).When age group was examined as an independent variable, there were significant effects of age on the item “The natural night-time behaviour of insects/bats/foxes, etc., remains the same as in previous years” (Kruskal–Wallis H = 9.3, p < 0.01; those over 55 endorse this statement most strongly) and for the item “The number of bats I see has increased recently” (Kruskal–Wallis H = 7.6, p < 0.023; those over 55 endorse this statement most strongly), but not on other items.Examining inter-relatedness of the above items through simple linear regression, there are a number of statistically significant weak-to-moderate relationships (Table 2). Most notably, there is a moderate inverse relationship between “The natural night-time behaviour of insects/bats/foxes, etc., remains the same as in previous years” and “The number of bats I see has increased recently” (r = −0.304). There is also a positive relationship between “The level of lighting near my home at night has increased over the past three years” and “‘Birds sing at night (the dawn chorus starts earlier than it used to)” (r = 0.251). There were no strong associations between any of the items.Finally, we assessed interest in night events by asking “Which of the following statements best describes your attitude towards night-themed events (such as dark sky festivals, night walks/runs, etc.)?”. Of those that expressed an opinion, 24.2% expressed a negative attitude to such events, 55.1% a neutral attitude and 20.8% a positive attitude. There were no differences across these three groups on any of the Likert-like items relating to light at night impacts, and chi-square analysis indicates that those that express an interest in night events were not more likely to live in city/town/rural settings (p = 0.16), but older respondents (55 or older) were more likely to have a positive attitude towards night events (p = 0.006).Citizen science approaches have previously been deployed successfully in the measurement of night sky illuminance [12], speaking to its utility for exploring other aspects of light pollution. There is a clear evidence gap in the literature around perceptions of, and beliefs about, ALAN and its impacts. A previous survey of >2000 Finnish respondents reported that light pollution was perceived to decrease the recreational amenity of outdoor spaces, that experience of dark skies was not common, that public lighting was the most commonly identified source of light pollution and that commercial lighting was the most annoying source of light at night in a predominantly urban and educated sample with a high level of interest in astronomy [11]. Our current results echo some of these findings, in that no city dwelling respondents report experiencing completely dark skies, and that public lighting was identified as the main source of light pollution for city and town dwellers, and was an important source alongside own/neighbours’ domestic lighting for rural respondents. However, our current results also do not present evidence for a perceived recent increase in the level of ALAN in rural, town or city settings. This may reflect the limited level of switches to high-illuminance LED public lighting in the past three years in Ireland and may be subject to change as such lighting becomes more prevalent.Regarding the perceived impacts of ALAN, we find evidence that respondents endorse that light entering the bedroom impacts on sleep (a finding mostly strongly reported in rural dwellers), a finding that corresponds to finds that ALAN is associated with poorer sleep and other health outcomes in older Japanese adults [13,14], and suggests that in Ireland light pollution may be an important environmental factor to consider for sleep health. However, it should be noted that it is not presently clear whether environmental light pollution is of sufficient intensity at the incident level of the retina to produce physiological effects, or whether sleep-health impacts of ALAN may be primarily mediated through psychological mechanisms such as sleep attentional biases [1,10].The perceived effects of light at night on animal behaviour appears to be most pronounced in city setting, findings that accord with recent reports on light pollution effects on directly observed animal behaviour [15], and is most likely reflective of greater light pollution associated with urbanisation [2]. Given that the level of exposure of wildlife to light pollution may be significantly greater than humans’ (who will mostly be indoors during the night; [1]), the magnitude of direct behavioural effects of ALAN on wildlife behaviour may be greater than that on human behaviour. Another possible link between urban setting and perceived alterations in wildlife behaviour is noise pollution, which may interact with light pollution in altering bird behaviour [16]. Perceptions of the impacts of light pollution did not vary according to interests in night-themed events, suggesting that the reported effects are not a result of general increased awareness amongst respondents with particular interests in dark skies issues.There are a number of important caveats for the current study. Firstly, the sample is unlikely to be representative of the Irish population; rather, the sample was self-selected from those responding to the survey article in the Irish “newspaper of record”, and as such is likely to be biased towards those with pre-existing interests in ALAN, sleep and/or ecology. Second, given the nature of the survey, we did not collect either direct measures of light at night, nor other detailed demographics; nor did we collect objective measures of sleep or wildlife behaviour; future work might usefully address the associations between such subjective reports and objective records of behaviour. Thirdly, perceptions of ALAN may be shaped by other psychological constructs and beliefs; for example, such perceptions may be influenced by social amplification [5]. Fourthly, given the strictures of a citizen science project, we did not use established psychometric scales for assessing sleep or other domains, given the imperative for brevity in the survey design. Fifthly, future work might usefully address whether chronotype (preference towards earlier or later timing of sleep/wake) influence perception of ALAN. Overall, the current study indicates that experience of man-made light at night is common in Ireland, but varies by geographic location and age, with older age and city location associated with the greatest perceived effects on sleep and animal behaviour.This study indicates that citizen science approaches may be useful in gaining insight into public perceptions of man-made lighting. Further, this study indicates that there are differences in perceptions of the presence and impact of light pollution across rural and urban settings and that perceptions of some light-associated effects (e.g., on wildlife behaviour) are more commonly reported in older respondents. Future studies will be needed to assess the relationship between subjective self-reported levels of light at night and objectively assessed levels in order to replicate the current findings and to broaden the scope of factors examined.The following are available online at https://www.mdpi.com/1660-4601/17/15/5628/s1, ALAN Questionnaire.Conceptualisation, G.M., A.G. and B.E.; data curation, A.N.C., M.F. and B.E.; formal analysis, A.N.C., M.C.-G. and M.F.; investigation, A.N.C., M.C.-G., A.G. and B.E.; methodology, M.C.-G., A.G. and B.E.; project administration, A.N.C., G.M., A.G. and B.E.; writing—original draft, A.N.C.; writing—review and editing, M.C.-G., A.G. and B.E. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The authors declare no conflict of interest.Number of responses for items on (A) dark skies and (B) strongest light source near home by self-reported residence location.Box and violin plots for responses on Likert-like items relating to individual questions on perceived impacts of light pollution (A–E) compared across three groups of self-reported residence location. * indicates p < 0.05 for Bonferroni-adjusted pairwise comparison by Mann–Whitney U test; *** denotes p < 0.005.Demographics of the Study Sample.Inter-item correlation matrix for Likert-like items on perceived impacts of light pollution.

Med-MDPI/ijerph_5/ijerph-17-16-05629.txt

@@ -0,0 +1 @@
+Background: Social trust, assessed by the trustworthiness of one another in a community, is known to have beneficial effects on health outcomes. However, the impact of social trust on metabolic syndrome (MetS) is unclear. Methods: The study subjects were extracted from the Korean National Health Insurance Service, and social trust was obtained from the Korean Community Health Survey (KCHS). Previously healthy participants were followed up from 1 January 2010 to 31 December 2011, and again from 1 January 2012 to 31 December 2013 for waist circumference, blood pressure, fasting blood glucose, triglycerides and high-density lipoprotein cholesterol (HDL-C). Multivariate logistic regression was used to calculate the adjusted odds ratios (aORs) with 95% confidence intervals (CIs) for newly developed MetS according to social trust quintiles. Stratified analyses were performed to determine the relationship between lifestyle behaviors and social trust. Results: Compared to the participants within the first quintile of social trust, those in the remaining quintiles had lower risks of developing MetS. The aOR with the 95% CI was 0.88 (0.79–0.98) in the 5th quintile group of social trust. Among the diagnostic criteria for MetS, waist circumference and HDL-C were statistically significant with aORs of 0.91 (0.84–0.99) and 0.88 (0.80–0.95) in the 5th quintile group. The stratified analyses showed protective effects of positive lifestyle behaviors. The aORs with 95% CIs were 0.85 (0.74–0.99) in never smokers, 0.82 (0.70–0.95) in non-drinkers and 0.87 (0.76–1.00) in the physically active in the highest level of social trust. Conclusions: Higher social trust was associated with a lower incidence of MetS. Therefore, building community with psychosocial support may be helpful in improving public health. Social capital has been considered an important factor in determining health status since its introduction in the 1990s [1]. Although many dimensions of social capital exist, it has been generally accepted as an asset for promoting beneficial economic, social and health outcomes [2,3]. Among the components of social capital, social trust, as a cognitive component, has been known to facilitate social interaction and the exchange of information [4]. Social trust is usually assessed by the question, “would you say that people can be trusted?” This question evaluates the trustworthiness of one’s neighborhood, which may impact one’s behavior in the community [5]. Because social trust can also influence health behavior, one’s metabolic profiles may change depending on the level of social trust. While abundant investigations have been performed to study the association between social capital and various health outcomes, such as all-cause mortality [6] and depression [7], there is a dearth of information on the relationship between social capital and metabolic syndrome (MetS). With its increasing incidence and predictive value for cardiovascular risks and diabetes [8], MetS has become an important global health issue. Therefore, determining the effects of social trust on MetS may be crucial for public health implications.Evidence for a significant association between social capital and MetS is lacking and only studies that used proxy measures for MetS are available. While some studies showed a positive correlation between social trust and cardiovascular diseases [9] and obesity [10], others showed null or negative results [11]. A study in Australia found a higher objective crime rate associated with a higher risk of MetS in men, and a higher perceived crime rate associated with a higher risk in women. Both total and violent crime rates were associated with MetS in men, but only the perceived crime rates of neighborhoods were associated with MetS in women [12]. Another study in Canada found a negative correlation between network social capital and waist circumference. It did not, however, find a significant association between social trust and waist circumference [13]. Because the association between MetS and social trust has not yet been established, elucidating the relationship may help improve public health outcomes. As MetS has become such a syndemic [14], the primary prevention of MetS requires more societal and structural changes.Therefore, the objective of this study was to elicit the association between community-level social trust and MetS using a national cohort study of Korean population data. It was hypothesized that a higher level of social trust was related to a lower likelihood of developing MetS.This was a longitudinal, retrospective study that collected health information from existing data. The study population was extracted from the Korean National Health Insurance Service–National Sample Cohort (NHIS-NSC). In South Korea, the NHIS is a universal healthcare system for all Korean citizens, collecting health service utilization records for insurance claim purposes that include outpatient and inpatient hospital visits, health screening examinations, diagnostic and treatment-related procedures and pharmaceutical prescriptions. The health screening exams contain a self-reported questionnaire on lifestyle behaviors, anthropometric measurements and urine and blood tests biannually for enrollees aged 40 years or older. Parts of these data are available for research purposes, and many large-scale epidemiological studies have used the NHIS database. The validity of the database is described in detail elsewhere [15,16].Among 211,125 participants aged 40 years or older in the NHIS-NSC, 724 enrollees who died before the index date were excluded. Two thousand seven hundred and seven and 836 were excluded for missing values for covariates and MetS criteria, respectively. Those who did not answer social capital-related questions (2069 people) and who were already taking statins, hypertension medication or diabetes medication (116,000 people) were removed from the study. This study only included participants who lived in the metropolitan area. A total of 47,289 participants were excluded from the study for living in rural areas. The participants meeting the inclusion criteria were followed from 1 January 2010 to 31 December 2011, and again from 1 January 2012 to 31 December 2013 for waist circumference, blood pressure, fasting blood glucose, triglycerides and high-density lipoprotein cholesterol (HDL-C).The study was approved by the Seoul National University Hospital Institutional Review Board (IRB number: E-1806-076-951). Because the NHIS-NSC database is anonymized according to strict confidentiality guidelines prior to distribution, the requirement for informed consent was waived.Social trust values were measured using the Kawachi method and the details and validity of it have been described in previous papers [17]. The Korean Community Health Survey (KCHS) had a question to assess social trust, which was conducted by the Korean Centers for Disease Control and Prevention in 2011. It is a nationally and district-representative community-based cross-sectional survey that contains community-level information according to administrative district sites [18]. Social trust was assessed by the statement, “the people in my neighborhood can trust one another,” and the responses were categorized into two answers: trusting and non-trusting. Social trust was calculated by determining the proportion of those who answered “yes” to the social trust question for each administrative district site. A total of 253 district sites, with a mean (standard deviation) land area of 55.1 (79.9) km2, covers the entire South Korea land mass. The social trust values were then merged with NHIS-NSC according to each participant’s residential district (a total of 253 districts). Rural areas were then excluded and only residents from 74 districts were included in the study. Participants were then categorized into five groups evenly according to the level of social trust, the 1st quintile having the lowest level of social trust and the 5th quintile having the highest level of social trust.All Korean citizens have universal healthcare access managed by the NHIS, which covers nearly all health care services and biannual health screenings for people 40 years and older. The healthcare database contains waist circumference, fasting blood glucose, HDL-C, and triglyceride levels and blood pressure.The definition of MetS was derived from revised the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) criteria [19]. It requires at least three of the following components: (1) abdominal obesity (waist circumference ≥90 cm for men, or ≥85 cm for women); (2) triglycerides ≥150 mg/dL and/or drug treatment for elevated triglycerides; (3) HDL-C < 40 mg/dL for men or <50 mg/dL for women; (4) systolic blood pressure ≥130/85 mmHg or antihypertensive medication treatment and/or a history of hypertension; and (5) FSG ≥ 100 mg/dL and/or treatment with medications for type II diabetes mellitus.Multivariate logistic regression was used to determine the adjusted odds ratios (aORs) with 95% confidence intervals (CIs) for MetS as a composite outcome and each component of MetS (waist circumference, fasting blood glucose, HDL-C, triglyceride levels and blood pressure). The incidence of developing MetS was calculated compared to the 1st quintile of the social trust group. Social trust was divided into five groups, with the lowest being the 1st quintile and the highest being the 5th quintile. The covariates considered included age (categorical, 40–49, 50–59, 60–69 and ≥70 years), sex (categorical, male and female), household income (categorical, 1st, 2nd, 3rd and 4th quartiles), residence (categorical, capital city and metropolitan area), smoking behavior (categorical, never smoker, past smoker and current smoker), drinking behavior (categorial, none, 1-2 times per week, 3–4 times per week and ≥5 times per week) and Charlson Comorbidity Index (CCI), (continuous). Household income was derived from the insurance premium. The algorithm for calculating CCI using claims data was derived from elsewhere [20]. These covariates were adjusted at three different levels. Model 1 adjusted only for age, income and residence, while model 2 adjusted for smoking, drinking and physical activities in addition to model 1. In model 3, CCI was also considered.The stratified analyses were performed for lifestyle behaviors—smoking, drinking and physical activity. Fully adjusted model 3 was used to determine the effects of each lifestyle behavior on the incidence of MetS. Multivariate logistic regression was also used to calculate the aORs with 95% CIs.Statistical significance was defined as a p-value of <0.05 in a two-tailed manner. All data collection and statistical analyses were conducted using STATA 15.0 (StataCorp, College Station, TX, USA).This study was approved by the Seoul National University Hospital Institutional Review Board (IRB number: E-1806-076-951). The requirement for informed consent was waived as the NHIS-NSC database was anonymized according to strict confidentiality guidelines prior to distribution.Table 1 depicts the descriptive characteristics of the study population. The ranges of social trust for each quintile are 42–53%, 54–59%, 59–61%, 61–68% and 69–88%, respectively. There was no significant difference among the groups, except for the location of residence. There were no capital city dwellers in the 5th quintile of social trust group.The aORs for MetS for the total population and male and female subgroups are shown in Table 2. A lower aOR for the incidence of MetS was shown in the 5th quintile group when compared to the 1st quintile of social trust in total and both sexes. Compared to the 1st quintile of social trust of the total population, the 2nd quintile group has an adjusted odds ratio of 0.88 (95% CI 0.80–0.96), the 3rd quintile 0.97 (0.88–1.07), the 4th quintile 0.87 (0.79–0.95) and the 5th quintile 0.87 (0.78–0.97) in model 1. The numbers did not differ significantly in models 2 and 3. In the case of males, the adjusted odds ratios with 95% CIs were 0.95 (0.83–1.05) in the 2nd quintile group, 0.96 (0.85–1.08) in the 3rd quintile group, 0.89 (0.79–1.01) in the 4th quintile and 0.88 (0.76–1.01) in the 5th quintile in all three models. The female population exhibited aORs with 95% CIs of 0.79 (0.66–0.95), 0.97 (0.82–1.15), 0.80 (0.66, 0.96) and 0.82 (0.68–0.98) in the 2nd, 3rd, 4th and 5th quintile groups, respectively, when compared to the 1st quintile group of social trust.Table 3 shows the aORs of each MetS component when adjusted for age, residence, income, smoking, drinking, physical activity and CCI. Among the components of MetS, only waist circumference reduced the aOR of new incidences MetS in a statistically significant manner. The aOR for HDL-C was statistically significant only in the 5th quintile group of social trust. The aOR with 95% CI for waist circumference for the 5th quintile group of social trust when compared to the 1st quintile group was 0.92 (0.85–0.99) for model 1. The aORs with 95% CIs for model 2 and model 3 in the 5th quintile group were 0.91 (0.84–0.99) and 0.91 (0.84–0.99). When the 5th quintile group of social trust was compared to the 1st quintile group, the aOR with 95% CI was 0.88 (0.81–0.96) in model 1. For models 2 and 3, the aORs with 95% CIs were 0.88 (0.80–0.96) and 0.88 (0.80–0.95), respectively.Lastly, stratified analyses on the association between social trust and MetS, taking into consideration smoking, alcohol intake and physical activity, are shown in Table 4. Never smokers and non-drinkers reduced the incidence of MetS. The aORs with 95% CIs in never smokers when compared to the 1st quintile group of social trust were 0.85 (0.75–0.97), 0.95 (0.84–1.09), 0.84 (0.74–0.95) and 0.85 (0.74–0.99) in the 2nd, 3rd, 4th and 5th quintiles groups. On the other hand, the aORs with 95% CIs for past and current smokers were 0.93 (0.81–1.06), 1.00 (0.87–1.16), 0.91 (0.79–1.05) and 0.89 (0.76–1.04) in the 2nd, 3rd, 4th and 5th quintile groups of social trust when compared to the 1st quintile population. In non-alcohol drinkers, the aORs when compared to the 1st quintile group of social trust were 0.89 (0.77–1.02), 1.03 (0.90–1.18), 0.82 (0.71–0.94) and 0.82 (0.70–0.95) in the 2nd, 3rd, 4th, and 5th quintile groups of social trust. The alcohol drinkers showed aORs with 95% CIs of 0.88 (0.77–1.00), 0.92 (0.80–1.05), 0.92 (0.81–1.05) and 0.93 (0.80–1.07) from the 2nd to the 5th quintiles of social trust. Then the physically active group and inactive group were also compared, and the protective effect of physical activity on MetS was not significant. The physically inactive group had aORs with 95% CIs of 0.81 (0.71–0.95) and 0.88 (0.75–1.03) in the 4th and 5th quintiles of social trust groups, respectively, when compared to the 1st quintile, while the physically active group had aORs with 95% CIs of 0.91 (0.80–1.04) and 0.87 (0.76–1.00) in the 4th and 5th quintile groups.This population-based, longitudinal study examined the association between social trust and MetS. The beneficial effect of social trust on reducing the incidence of MetS persisted even after taking into account differences in age, income, area of residence, lifestyle behaviors—smoking, alcohol drinking and physical activity—and CCI. In stratified analyses with lifestyle behaviors, smoking, alcohol intake and physical activity all showed a statistically significant impact on MetS incidence in a previously healthy population. To our knowledge, this is the first longitudinal study to demonstrate that district-level trust was associated with a lower incidence of MetS in individuals, using nationally representative cohort data.Previous studies have investigated the association between social trust and proxy measures of MetS. In a Canadian paper that investigated the causes of health inequality, Indigenous people with higher social support were associated with a lower cardiovascular disease risk score [21]. Another study conducted with Americans aged 50 years and older, found a statistically significant association between higher perceived social cohesion and a lower incidence of stroke [22]. On the contrary, a nationally representative study conducted in China in 2017 showed that higher social trust was associated with a lower likelihood of obesity, and harmonious social relationships were correlated with higher chances of becoming obese [3]. Most of these investigations were cross-sectional and could not prove causality, while this study was longitudinally designed to capture the effect of social trust on the incidence of MetS. We also used the direct measure of MetS and its components rather than proxy measures. It was determined that higher social trust was associated with a lower incidence of MetS. Furthermore, by adjusting out age, income, area of residence, lifestyle behaviors and CCI, we tried to eliminate the confounding factors that were not pre-determined in the study design. The aORs of developing MetS remained lower in higher social trust groups even after adjusting for covariates.Different mechanisms have been proposed to explain the association between social trust and MetS. First, people with higher social trust are likely to have a higher sense of security, which may help in the exchange of valuable information or instrumental support within society and in absorbing health-promoting behaviors [17,23]. Second, in societies with higher social support and network groups, people have easier access to transportation systems and healthcare [24]. Furthermore, when residents live in a safer neighborhood, they are more likely to exercise [12]. Another explanation is collective efficacy. Members of a community may act together to promote health-promoting behaviors and against harmful behaviors, such as collecting signatures for a smoking-free zone [25]. Lastly, psycho-social pathways also help explain the association between social trust and MetS. A lower level of social trust may increase social anxiety and stress, which in turn may elevate blood cortisol levels. The stimulation of the hypothalamic-pituitary-adrenal (HPA) axis can cause inflammation and diseases, such as cancer [26] and cardiovascular accidents [27].In this study, the aORs of MetS incidence were found to be lower in women than in men. This may be explained by women having more a trusting and pro-social nature than men. Women’s tendency to adopt communal and interpersonal facilitative behavior may work together towards healthful behavior in a community [28]. In addition, women tend to relay information among members of a community more frequently than men. Men rely more on the information communicated with their spouses than with other community members. Moreover, the aORs were statistically less significant in the middle quintile groups of social trust. Social trust may need to be at the extreme ends to exert influence on people’s lifestyle behaviors. Generally, higher social trust was associated with positive health outcomes that can be explained by the abovementioned mechanisms. However, the relationship did not prove to be as significant in fasting blood glucose, blood pressure or triglyceride levels. These three components of MetS are more closely related to eating habits, which this study did not consider. The members of a community may share similar diet patterns and different diets affect metabolic profiles differently [29].Several limitations must be considered when interpreting the results of this study. Social trust was measured at one point in time, and changes were not considered. Additionally, the participants were only followed up for a short period of time because HDL-C levels were only collected in 2009. It may have been insufficient to determine the effects of social trust on the development of MetS. However, social trust is usually influenced by the environment, which does not change rapidly. Because social trust is closely knitted into the lives of community members, one year may have been enough to exert influence over the members’ health outcomes. In addition, although we adjusted for household income and area of residence, we could not fully take into account the effects of the neighborhood environment, education level and friends on health outcomes. The education level and diversity of friends were associated with chances of becoming obese in previous studies [30,31]. Lastly, we excluded the samples from rural areas due to population biases towards older adults and higher levels of social trust. This study tried to be more representative of the general population of the country. In rural areas, social trust is high and MetS incidence is low. Further analyses may be necessary to determine the influence of social trust on MetS in rural adults.In conclusion, higher social trust decreased the likelihood of developing MetS. Quitting smoking, drinking in moderation and being physically active also reduced the risk. Therefore, it is important to create a community where healthy lifestyles are encouraged among members of society. Since it is known that reducing MetS requires collective effort as a society, public health policy should aim to create health-conducive environments by increasing social trust through building recreational facilities and creating community memberships.H.P., conceptualization, methodology, software, validation, formal analysis, investigation, writing—original draft, writing—review and editing; S.C., conceptualization, methodology, software, writing–review and editing; K.H.K., writing—review and editing, methodology, project administration; E.K., writing—review and editing, validation, investigation; A.K., writing—review and editing, project administration; S.M.P., software, resources, data curation, supervision, conceptualization, funding acquisition. All authors have read and agreed to the published version of the manuscript.This research was supported by the Korean Centers for Disease Control and Prevention (grant number: 2018P330400). S. Choi received grants from the BK21-Plus Education Program from the National Research Foundation of Korea.None of the authors reported disclosures.Descriptive characteristics of the study population.p-value calculated with chi-squared test for categorical variables and ANOVA for continuous variables. Abbreviations: MVPA, moderate to vigorous physical activity; CCI, Charlson Comorbidity Index.Adjusted odds ratio (95% confidence intervals) for metabolic syndrome based on the NCEP ATP III criteria by social trust quintiles.Criteria for metabolic syndrome was defined as meeting three or more of the following conditions, as suggested by NCEP ATP III: (1) Impaired Fasting Glucose (≥100 mg/dL), (2) Elevated WC (>90 cm for men and >85 cm for women), (3) High Blood Pressure (SBP: ≥130 mmHg and DBP: ≥85 mmHg), (4) High Triglycerides (≥150 mg/dL), (5) Reduced HDL-cholesterol (<40 mg/dL for men and <50 mg/dL for women). Data presented are N (%) and aOR (95% CI). Logistics Model 1: adjusted for age, income and residence. Logistics Model 2: adjusted for age, income, residence, smoking, drinking and physical activities. Logistics Model 3: adjusted for age, income, residence, smoking, drinking, physical activities and Charlson Comorbidity Index. Abbreviations: aOR, Adjusted Odds Ratio; CI, Confidence Interval; WC, Waist Circumference; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; HDL, High Density Lipoprotein; NCEP ATP, National Cholesterol Education Program-Adult Treatment Panel III. Note: Each category of social trust quartile is compared to the 1st social trust quartile (reference).Adjusted odds ratio of (95% confidence intervals) for each metabolic syndrome criterion by social trust quintiles.Metabolic syndrome diagnostic criteria include (1) Impaired Fasting Glucose (≥100 mg/dL), (2) Elevated WC (>90 cm for men and >85 cm for women), (3) High Blood Pressure (SBP: ≥ 130 mmHg and DBP: ≥ 85 mmHg), (4) High Triglyceride (≥150 mg/dL), (5) Reduced HDL-cholesterol (<40 mg/dL for men and <50 mg/dL for women). Data presented are N (%) and aOR (95% CI). Logistics Model 1: adjusted for age, income and residence. Logistics. Model 2: adjusted for age, income, residence, smoking, drinking and physical activities. Logistics. Model 3: adjusted for age, income, residence, smoking, drinking, physical activities and Charlson Comorbidity Index. Abbreviations: aOR, Adjusted Odds Ratio; CI, Confidence Interval; WC, Waist Circumference; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; HDL, High Density Lipoprotein. Note: Each category of social trust quartile is compared to the 1st social trust quartile (reference).Stratified analyses on the association of social trust with metabolic syndrome, taking into consideration smoking, alcohol intake and physical activity.Fully adjusted model includes adjustments for age, residence, household income and Charlson Comorbidity Index. Adjusted odds ratios were calculated by multivariate logistic regression after adjustments for age, residence, household income and Charlson Comorbidity Index.

Med-MDPI/ijerph_5/ijerph-17-16-05630.txt

@@ -0,0 +1 @@
+We examined the factors associated with human immunodeficiency virus (HIV) screening and developed a HIV screening prevalence surface map using spatial interpolation techniques to identify the geographical areas with the highest and lowest rates of HIV screening in Mozambique. We analyzed the cross-sectional 2015 Mozambique AIDS Indicator Surveys with an analytic sample of 12,995 participants. Analyses were conducted on SPSS-21, STATA-14, and R freeware 3.5.3. We adjusted for the sample design and population weights. Results indicated that 52.5% of Mozambicans had undergone HIV screening. Mozambicans with these characteristics have a higher probability of undergoing HIV screening: females, those with a primary education or higher, urban dwellers, residents of wealthy households, having at least one lifetime sexual partner, and dwelling in these provinces—Niassa, Tete, Manica, Sofala, Inhambane, Gaza, Maputo Provincia, and Maputo Cidade. The spatial map revealed that the national and regional estimates mask sub-regional level estimates. Generally, zones with the highest HIV screening prevalence are found in southern provinces while the lowest prevalence was found in the northern provinces. The map further revealed intraregional differences in HIV screening estimates. We recommend that HIV screening be expanded, with equitable screening resource allocations that target more nuanced areas within provinces which have a low HIV screening prevalence. Despite decades of global concerted efforts to combat the spread of human immunodeficiency virus (HIV) and mitigate its effects on the global population, HIV infections still rank among some of the most important public health concerns across the world. According to the Joint United Nations Programme on HIV/AIDS (UNAIDS), an estimated 1.7 million new HIV infections were recorded worldwide in 2018 alone [1]. Epidemiological research shows the geographical and contextual differences in infection rates across different regions of the world, with countries in the global south being the most affected [2,3]. For example, sub-Saharan Africa has consistently topped the list of regions with the highest infection rate [2], with nations in the southern part of Africa bearing the brunt of the HIV epidemic [4]. Coupled with the myriad of other problems including poverty, tropical diseases such as malaria, rampant corruption, and armed conflicts that have bedeviled the continent for decades, the HIV epidemic continues to wreak havoc on the life expectancy, development, and quality of life of the African population.To combat the HIV burden, research scientists and epidemiologists often recommend a ramping up of screening or testing as a crucial first step. Screening is done in pre-symptomatic and/or asymptomatic populations to identify and treat those who are yet to show signs [5] and is crucial for gaining an important early situational awareness about the disease’s severity, tendencies and transferability. These, in turn, are important in predicting the likely effectiveness of interventions [5,6,7]. Rapid screening has also been shown to be the main means for individuals to be informed about their HIV status. Therefore, screening is a critical first step in the global fight against the HIV epidemic.In 2014, UNAIDS and partners launched the 90–90–90 targets. The aim was to have 90% of all HIV-positive persons diagnosed, provide antiretroviral therapy (ART) for 90% of those diagnosed, and, for 90% of ART recipients, achieve viral suppression by 2020 [8]. The most important aim is step one, to identify 90% of all HIV positive people. However, in a recent systematic analysis of national HIV treatment cascades from 69 countries by Levi and colleagues [9], none of the countries had achieved the 90–90–90 targets. They reported that diagnosis (target one—90% of all HIV-positive people diagnosed) ranged from 87% (the Netherlands) to 11% (Yemen) [9].Mozambique ranks among countries with the highest burden of HIV, with a prevalence of 12.6% in adults between the ages of 15 and 49 years. In 2018 alone, there were 150,000 new infections and 54,000 HIV deaths in the country [1]. Globally, this small south-eastern African country ranks as the fifth highest HIV infected country in the world according to 2015 estimates [10]. Over the years, researchers have listed relatively little knowledge about HIV, poor health systems, and inadequate access to HIV preventive and therapeutic services among a host of other factors as key contributors to the country’s high HIV prevalence [11,12,13,14]. However, factors that influence HIV screening in Mozambique and other sub-Saharan African countries have not been well studied.In the face of the evidence and in response to recommendations, the government of Mozambique, with support from its aid partners, have invested heavily in the procurement of HIV screening and diagnostic testing equipment [7]. To support these positive steps and their consequent gains, insights into the specific trends and discrepancies in screening prevalence and associated predictors are necessary for providing empirical guidance for the continuous evaluation, prioritization and direction of ongoing HIV response activities. However, such insights are largely missing in the body of literature on HIV diagnosis and prevalence in Mozambique. Therefore, this study aimed to elucidate differences in the screening prevalence within provinces in Mozambique using the most current nationally representative AIDS Indicator Survey data.The study used publicly available data from the cross-sectional Mozambique AIDS Indicator Survey (M-AIS) conducted in 2015. The study had two main objectives, to identify the characteristics of those who are more likely to be tested for HIV, and then to identify the geographical areas where testing rates varied. The M-AIS data is collected based on a two-stage sampling design. A master sample plan consisting of 307 primary sampling units (PSUs) was randomly selected from the 2007 Mozambique General Population and Housing Census sample frame. The 307 PSUs were stratified by the locality of residence, with 134 urban PSUs and 173 rural PSUs. From each PSU, twenty-four (24) households were selected, resulting in a total household sample size of 7368 households. The estimated number of individuals to be enumerated were 14,343, but a total of 13,032 individuals was reached, a response rate of 90.86%. Thirty-seven (37) cases did not have complete information on the outcome and the explanatory variables, therefore, they were removed from the dataset. The dataset used for this study contained 12,995 cases.The outcome variable for the study is HIV screening. In the dataset, the variable with information on HIV screening is labelled as “Ever been tested for HIV” with a “Yes” or “No” response. The response “Yes” was coded as “1” and “No” as “0”.The following ten sociodemographic variables were selected from the dataset as potential predictors: gender, age, education, marital status, religion, total lifetime sexual partners, history of sexually transmitted infections (STIs) 12 months before the survey, household wealth index, rural–urban place of residence, and region of residence. The household wealth index was already computed by the DHS program and available in the dataset. DHS used a principal component analysis (PCA) to assign weights to each asset in each household and cumulative scores were calculated from the assigned weights. From the PCA results, households were categorized into five quantiles: poorest, poorer, middle, rich, and richest. The computation used the following information: household characteristics (source of drinking water, type of toilet, sharing of toilet facilities, the main material for the roof, walls and floors, and type of cooking fuel amongst others household characteristics) and household possessions and assets (ownership of television, radio, vehicle, bicycles, motorcycles, watches, agricultural land, farm animals/livestock, and bank account, amongst others).Descriptive statistics and a test of association were performed in SPSS, and the multivariable analysis was performed in STATA. The analyses adjusted for the sample design (PSU and sample strata) and weights. The statistical significance threshold was pegged at the 5% level of significance (p ≤ 0.05). SPSS-21, STATA-14, and the R freeware 3.5.3 were used for the analyses.The first objective of this study was to identify the characteristics of those who are more likely to be tested for HIV. In SPSS, the steps for adjusting for the complex sample procedure is reported elsewhere [15]. Sample characteristics were described using frequencies and percentages. A contingency table with chi-square test of independence was used to measure the association between the outcome variable and each explanatory variable. Explanatory variables that were significantly associated with the outcome were included and assessed in a multivariable model. We assessed the interaction effect of gender on the relationship between each study covariate and HIV screening among Mozambicans by computing, for each covariate, a complex samples Poisson model. A Wald chi-square test was performed to assess the significance of the interaction term.In STATA, the “svy” command in a default mode was used to set up the analytic environment before analysis. The default “svy” computes the standard errors by using the linearized variance estimator called the first-order Taylor linearization. This procedure eliminated the incorrect estimation of the standard errors (SEs) associated with the confidence intervals of the regression coefficients. A Poisson regression was used to estimate the prevalence ratios. We checked the assumption of the overdispersion of the data using the nbreg command, and found that the likelihood-ratio (LR) test of alpha = 0 and the p-value of the LR test was greater than 0.05, implying that the conditional variance is equal to the conditional mean, making Poisson an appropriate model to fit for the outcome.The second objective of this paper was to understand the HIV epidemic and inform programs and interventions at a lower geographical-level. The 2015 M-AIS has information on the HIV screening prevalence for 306 clusters. HIV screening prevalence at a cluster level through a spatial interpolation technique was performed using cluster-level geolocation data (longitude and latitude). The prevR package in the R freeware was used for the spatial interpolation [16]. The package is designed to determine the prevalence of outcomes from surveys with a stratified two-stage sample design [16]. Using functions available in the prevR package came with preinstalled functions which we used to generate the surface map for the HIV screening prevalence for Mozambique by adopting the Gaussian kernel estimator approach, with adaptive bandwidths for an equal number of persons surveyed [16]. A comprehensive methodology on how the prevR package can be used to obtain the surface map is reported elsewhere [16]. We also use the following packages in R for the analysis: foreign, maptools and ggplot2. We also generated a provincial HIV screening prevalence stratified by gender using the Quantum Geographic Information System (QGIS).The Ethical Review Committee of the Mozambique Ministry of Health’s National Institute of Health and the Institutional Review Board of ICF International reviewed and approved the 2015 Mozambique AIS protocol [17]. The 2015 Mozambique AIS data is publicly available upon a simple registration-access request, so we did not seek for further ethical clearance. The data can be obtained from the DHS website at https://dhsprogram.com/data/dataset_admin/index.cfm.The proportion of the respondents who had ever tested for HIV was 52.5% (Table 1). There were more women (59.5%) than men (40.5%) in the study population (Table 1). The majority of the respondents had attained a primary level of education (52.8% (Table 1)). The majority of them were currently married (64.9%), Muslims (39.3%), and residing in rural areas (63.3% (Table 1)).A chi-square test of independence and bivariate logistic regression analyses were performed to ascertain the relationship between the variables and ever screened for HIV. The results revealed that all the variables considered in the study were significantly associated with being tested for HIV. The proportion of females (61.3%) who had ever undergone a HIV screening was more than males (39.6%), and the proportion of urban dwellers (63.6%) screened was more than rural dwellers (46.1% (Table 1)). The proportion of post-secondary education holders (89.1%) who had undergone HIV screening was higher than those with a secondary (64.7%), primary (49.1%), and no education (43.3%) (Table 1). More people had undergone HIV testing in the richest households (69.6%) compared to the richer (61.5%), middle (48.5%), poorer (40.7%), and the poorest households (36.8% (Table 1)). A detailed description of the chi-square test of association results is presented in Table 1.The adjusted complex samples Poisson regression revealed that the following sociodemographic factors are statistically significant predictors of HIV infection in Mozambique: sex, age, education level, marital status, total lifetime sexual partners, household wealth index, urban/rural residence, and region of residence (Table 2).Mozambicans with these characteristics have a higher probability of undergoing a HIV screening: females, those with primary education or higher, urban dwellers, residents of wealthy households, having at least one lifetime sexual partner, and dwelling in these provinces—Niassa, Tete, Manica, Sofala, Inhambane, Gaza, Maputo Provincia, and Maputo Cidade.Gender modified the association between HIV screening and the following study covariates: respondent’s age (p < 0.001), education (p < 0.001), marital status (p < 0.001), religious affiliation (p < 0.001), total lifetime sexual partners (p < 0.001), STI status in the past 12 months (p < 0.001), household wealth (p < 0.001), rural/urban residence (p < 0.001), and province of residence (p < 0.001) (Table 3).The drivers of HIV testing are largely the same for both males and females in Mozambique. However, there are key differences in the way certain factors relate to HIV testing between men and women (Table 4). For instance, the effect of the total number of lifetime sexual partners on HIV testing, though the relationship goes in the same direction, was much bigger for women than for men (Table 4). Another important factor is education. The results indicate that the effect of education on HIV testing is slightly bigger for men than for women, although the relationship goes in the same direction (Table 4).The national HIV screening prevalence for Mozambique was 52.5%. We stratified the HIV screening prevalence by province and discovered that the national estimate masked the prevalence at the provincial level (Figure 1; Table 1). For instance, Gaza’s estimate of 72.4% is higher than the national estimate and Nampula’s estimate of 38.4% is lower than the national estimate (Figure 1; Table 1). We further stratified the provincial HIV screening prevalence by gender represented with a pie graph on the map (Figure 1). The results indicate that the proportion of women who had undergone HIV screening was more than the proportion of men in all the eleven provinces (Figure 1 and Table S1).Overall, the HIV screening surface map revealed that national and regional level estimates mask sub-regional level estimates (Figure 2). The general observation is that the zones with the highest HIV screening prevalence are found in southern provinces and zones with the lowest prevalence are found in the northern provinces (Figure 2). Furthermore, the surface map revealed intraregional level differences in the HIV screening estimate. For instance, there are areas within Gaza with higher screening prevalence than others.HIV screening is critical to treatment, care and prevention. Adequate HIV screening is integral to meeting the 90–90–90 targets made by the UNAIDS [8]. Our study sought to estimate the prevalence of HIV screening and its associated factors among Mozambicans in their reproductive age using the 2015 M-AIS dataset. To eliminate the masking of sub-regional estimates that characterized national and regional level analyses, we developed an HIV/AIDS screening prevalence surface map using spatial interpolation techniques to identify geographical areas with the highest and lowest HIV screening rates in Mozambique. Our results emphasize the importance of geographical-level variations and the impact of factors including gender, marital status, age, education, wealth index and place of residence on HIV screening. We found that being female, being presently or previously married, having at least a primary education, living in an urban area, and coming from a wealthy home were linked to a higher likelihood of being screened for HIV. Although some of these factors have previously been linked to HIV screening [18,19], their geographical-level distributions and contributions have not been studied in Mozambique.We found that a little over half (52.5%) of Mozambicans at reproductive age had undergone HIV screenings. The screening rate found in our study was higher when compared to the same population in Chad (42%), Burkina Faso (41%), and Sierra Leone (34%) but were lower than those found in Malawi (89%), Rwanda (95%) Zambia (82%) and Zimbabwe (89%) [20]. The high rates of HIV screenings in these countries were attributed to the mandatory HIV testing at prenatal clinics and mobile clinics for HIV screening [20]. However, our spatial interpolation revealed that there are areas in Mozambique with extremes of HIV screening prevalence compared to the national and regional estimates. For example, in Niassa province, the regional HIV screening prevalence ranged between 42.1% and 51.6%, however, the spatial analysis identified some clusters to be as low as 20% and some as high as 70%. Our findings also confirm the importance of applying spatial interpolation in population-based studies to unmask hidden details which can help design targeted interventions.The spatial heterogeneity in HIV testing can be attributed to both the offer of HIV testing, i.e., the accessibility, and the demand for HIV testing (i.e., knowledge, attitudes, and beliefs regarding HIV and testing). HIV testing happen in healthcare facilities in Mozambique, however a greater disparity exists in terms of the access to healthcare facilities across the eleven provinces in a walking-to-healthcare facility scenario [21]. About 90% of Mozambicans were not within a walking distance of 60 min. For Maputo city, about 70% of the area are within an hour’s walking distance to a healthcare facility. However, the situation is worse for Tete, Cabo Delgado, and Gaza, with over 90% of the area outside of an hour’s walking distance to a healthcare facility [21]. Besides, a comprehensive knowledge of HIV varied across the provinces in Mozambique [22], and lower HIV knowledge and a fear of HIV related stigma prevented Mozambicans from accessing HIV testing services [23].We also found that, compared to all age groups, adolescents between ages 15 and 19 years tend to have the lowest likelihood of getting screened for HIV. This result is particularly important because of the high rate of new infections among this age group [24] and the broad opportunity for implementing HIV prevention strategies during this critical period in life. Addressing adolescents’ lack of HIV testing to improve treatment and prevention requires an increase in community-based approaches to testing, youth friendly services, integrating HIV testing with other health services, such as reproductive health services, and building differentiated service models amongst others [25,26]. Additionally, a plethora of evidence suggests that school-based sex education interventions is linked with a sound HIV knowledge which is put into practice among adolescents and youths [14,27,28]. Therefore, school-based HIV knowledge, attitudes, and practice interventions can be pursued in Mozambique to promote HIV testing among adolescents.The higher rates of HIV screening among women in the sample are possibly related to the mandatory screening during a prenatal clinic visit to prevent mother-to-child transmission [29]. Similarly, the higher rates of HIV screening among currently and previously married people could be explained by mandatory screening being required by some religious groups, especially Christians, in some parts of sub-Saharan Africa before performing marriage ceremonies [30,31,32]. These mandatory HIV screenings have been previously reported to help the uptake of HIV testing [29,32]. This suggests that a collaboration between public health officials and religious organizations may improve the uptake of HIV screening which could help prevent the spread of the virus. Besides, the higher rates of female screening compared to males also raises questions as to the relative impact/influence that HIV screening education campaigns could be having on the sexes. By stratifying our multivariate model by gender, our results highlight that risk perception regarding having multiple lifetime sexual partners is likely lower in men than women. Our findings basically confirmed a well-documented fact that men and boys in general have a lower likelihood than women to screen for HIV, know their status, and access and adhere to HIV treatment when they test positive [33]. This challenge has been attributed to certain gender norms that make the adoption of safer sex practices and seeking and accessing health services seem unmanly. These results suggest that the government and nongovernmental organizations in Mozambique must adopt policies and programs that encourage men to utilize health services frequently. Regarding HIV testing, community-based testing and counselling, a focus should be made on door-to-door service provisions, mobile outreaches, couples or male partner testing, and integrating HIV testing into existing sexual, reproductive and other health services [33].Additionally, we found that people who are more educated, live in urban areas and who are from wealthy homes are more likely to get screened. The role of socioeconomic factors as a driver of HIV testing has been quite well-documented in sub-Saharan African countries [34,35]. These findings underscore the importance of attracting the most vulnerable and at-risk populations to HIV screenings. This could be done through targeted outreach programs, such as mobile clinics, and integrating HIV screening into routine healthcare services, improving home-based and self-screening and subsidizing the cost of the screening. Community leaders, such as chiefs, should also be involved in promoting HIV testing.The study made use of a large, nationally representative survey dataset that is grounded in standardized methodology for analyses. Secondly, the study employed spatial interpolation techniques that have advantages over standard statistical techniques to identify the geographical variations of HIV screening prevalence in Mozambique. Additionally, our study uncovered the at-risk population and location of low HIV screening areas in Mozambique. These findings could serve as a framework for public health officials to design targeted interventions to increase HIV screening.Our findings, however, are subject to limitations that must be taken into consideration. It is important to note that all the variables in this study, including HIV screening history, were self-reported. Like all self-reported data, some of the responses might be subjected to a social desirability bias. Additionally, we were not able to ascertain the actual reasons for participant’s prior HIV testing and therefore this was not accounted for in our analysis. Despite these limitations, this study has provided profound insights from a population-level survey analysis, as well as a spatial analysis of HIV screening prevalence in Mozambique for informed public health action.In conclusion, this study could help public health and health policy officials to develop an effective intervention in Mozambique by showing where and towards which populations HIV screening resources should be allocated. We found that 52.5% of the population in Mozambique have been screened for HIV—with females, those with a primary education or higher, urban dwellers and those from wealthy homes more likely to be screened for HIV. An expansion of HIV screening, outreach through mobile clinics, home-based approaches and self-testing, wide-ranging coverage through outreach programs, community-based approaches, and integrating opportunities to be screened during regular medical care are critical for reaching all HIV-positive persons in sub-Saharan Africa with lifesaving treatments.The following are available online at https://www.mdpi.com/1660-4601/17/16/5630/s1, Table S1: Regional HIV Screening Prevalence (%) by Gender (n = 12,995).Conceptualization, P.A., J.J.N., H.O.D.; methodology, P.A.; software, P.A.; validation, H.O.D.; formal analysis, P.A.; investigation, P.A., J.J.N., E.D., H.O.D., P.A.D.; resources, J.J.N.; data curation, J.J.N.; writing—original draft preparation, P.A., R.K.A., J.J.N., E.D., H.O.D., R.K.A., P.A.D.; writing—review and editing, P.A., J.J.N., E.D., H.O.D., P.A.D., R.K.A., K.B.; visualization, P.A., J.J.N.; supervision, J.J.N., K.B.; project administration, J.J.N.; funding acquisition, J.J.N. All authors have read and agreed to the published version of the manuscript.This work was supported by University of California, San Francisco Population Health and Health Equity Scholar program, and School of Nursing.We would like to thank the DHS program for permitting us to use the dataset and also a special thank you to the University of California, San Francisco Population Health and Health Equity Scholar programs for supporting this study.The authors declare no conflict of interest. Provincial HIV screening prevalence by gender.HIV screening prevalence in Mozambique estimated by the kernel estimator approach.Weighted summary statistics and chi-square test of the independence between covariates and the outcome (n = 12,995).a column percentage reported; b row percentage reported.Complex samples Poisson regression results of predictors of ever testing for HIV (n = 12,995).PR: prevalence ratio; APR: adjusted prevalence ratio; CI: confidence intervals.Covariates of HIV screening among Mozambicans by gender (male/female).Complex sample Poisson regression results of the predictors of having ever tested for HIV stratified by gender.

Med-MDPI/ijerph_5/ijerph-17-16-05631.txt

@@ -0,0 +1,14 @@
+The Green-for-Grain program (GGP) is the largest environmental restoration program in China. It is effective in controlling land desertification but at the same time is highly affected by regional differences. Ecosystem health, as an important indicator of ecosystem sustainability, can effectively assess the ecological impact of the GGP and provide a basis for follow-up actions. As a typical agro-pastoral ecotone along the Great Wall, the Xilin Gol League has seen increasing land-use intensity, thus, it is crucial to understand the ecological conditions of the region in order to deploy a policy of the GGP in accordance with local conditions. In this study, using remote sensing images and social statistics data from 1990–2015, land-use transformation and the turning point of vegetation coverage was determined. Based on the pressure-state-response (PSR) model, an ecological health evaluation system was constructed to quantify the temporal and spatial variation of ecosystem health. Then, the spatial correlation between the changes in forest and grass coverage, as well as the changes in the ecosystem health index (EHI), was evaluated using GeoDa software. The results showed that (1) grassland was the primary land-use/land-cover (LULC) in the Xilin Gol League. Since 2000, land-use transfer types changed significantly and grassland degradation weakened; landscape connectivity increased, and vegetation coverage increased. (2) Over the past 25 years, the ecosystem in the study area was at a subhealthy level and showed a trend toward a healthy level. (3) The spatial correlation between △Area% (change in forest and grass coverage) and △EHI (change in ecosystem health index) was positive between 2000 and 2015 and the correlation gradually increased, indicating that the GGP did enhance the health of the ecosystem of Xilin Gol. This study provided a specific reference for the evaluation of ecosystem health in the agro-pastoral ecotone of China and a theoretical basis for the implementation of sustainable management policies in the study area.The Grain-for-Green program (GGP) is one of China’s largest and best invested environmental projects [1]. Since 1999, China has gradually implemented ecological restoration projects centered around the GGP [2]. The GGP reduces soil erosion and increases carbon sequestration through increased vegetation cover [2,3]. Moreover, the GGP guarantees ecological sustainability and promotes harmonious development between humans and nature [4]. Although the GGP is one of the most effective measures in preventing and treating serious environmental problems [5], its effectiveness is limited by a number of socioeconomic factors, such as food production, local economic structure and farmer employment [6]. Some experts believe that the GGP is effective in improving ecosystem health [2]. Conversely, other experts argue that fencing natural grasslands reduces livestock farming and thus the livelihoods of pastoralists, which could affect the ecological and social development of the region [7]. Therefore, how to effectively assess the ecological impact of the GGP and how to formulate policy for the subsequent program according to local conditions is a key issue that China needs to solve immediately.Ecosystem health is a comprehensive, multiscale measurement for ecosystem vigor, organization, and resilience [8]. A healthy ecosystem should be stable and sustainable [9]. However, ecosystems in a sick state are often in the process of gradually declining and irreversibly collapsing [10] with a high probability of problems such as soil erosion, desertification, or salinization of the land. Therefore, ensuring ecosystem health is seen as the foundation and driving force for environmental sustainability. As an important method of assessing ecological condition, quantifying ecosystem health through objective indicators provides us with clear goals for managing the complex system of the ecosystem [11]. There are two main approaches to quantify ecosystem health: the indicative species method and the indicator system approach [12]. The indicative species method requires a large amount of measured data (such as biomass, productivity, structural indications, and functional indications for endangered species, long-lived species, environmentally sensitive species, etc. of the ecosystem) on a species, whereas the indicator system approach is not limited by the number, type, and source of data on ecosystems [13]. The indicator system approach is generally used for ecosystem health assessment.The pressure-state-response (PSR) is one of the most widely used models to guide the construction of the indicator system for ecosystem health evaluation [14]. PSR is based on the logical relationship between causes and consequences of problems and measures [15]. This model is often used to analyze selected environmental processes, identify relevant indicators, and ultimately provide a quantitative or qualitative description of causal relationships to understand the relationship between human activities and environmental impacts [16]. To address the shortcomings of the PSR model, namely, the subjectivity in indicator selection and weighting determination [17], we revised the model. First, we broadened the selection of indicators. The ecosystem service value was added in the evaluation indicators. In addition, the entropy weighting method was chosen to calculate weights to reduce human interference. More importantly, we used the PSR model in a 1000 m × 1000 m grid unit. Compared to previous studies of ecological health at the county level, we enhanced the analytical precision of PSR and it provided additional data for subsequent quantification of the effects of the GGP on ecosystem health.An ecotone is a transition zone between adjacent ecosystems [18] where the temporal and spatial changes are rapid and that is sensitive to external changes [19]. The health of ecotones is important for biodiversity conservation, ecosystem management, and restoration and functional zoning of nature reserves. The Xilin Gol League, located in the middle of Inner Mongolia in China, belongs to the agro-pastoral ecotone along the Great Wall, and is an essential ecological barrier, boding a significant strategic position in China’s social and economic development [20]. The complicated terrain of the Xilin Gol League leads to plant growth difficulty. Furthermore, the rapid development of the population and economy in recent decades intensified the human–land conflict. In these circumstances, the ecological environment in the Xilin Gol League has severely deteriorated and gradually become one of the primary sources of wind and sand in North China. To solve environmental problems, the GGP was launched in the Xilin Gol League in 2000, which set five counties as project pilots. From 2000 to 2003, the whole region implemented the GGP project. To date, few studies had used the GGP as a control to assess temporal and spatial changes in ecological health. The analysis of the impact of the GGP on regional ecosystem health, however, is a key factor in understanding the ecological conditions of the region and developing the right reforestation measures. The purpose of this study is to (1) calculate the rate and direction of the land-use/land-cover (LULC) change in the Xilin Gol League by using the related land-use model; (2) use the PSR model to build an indicator system to evaluate the health status of the agro-pastoral ecotone in the Xilin Gol League between 1990 and 2015; (3) analyze the spatial correlation between the GGP program and the spatial and temporal changes in the health of the ecosystem health, aided by spatial analysis methods. The results provide data support and scientific guidance for the utilization of the ecosystem in the agro-pastoral ecotone and provide information regarding the sustainable development of agriculture and the follow-up to the GGP.The Xilin Gol League (43°02′–44°52′ N, 115°13′–117°06′ E) is known as one of the four dominant natural grasslands in China; it borders the People’s Republic of Mongolia to the north (Figure 1). With a northern temperate continental climate, the Xilin Gol League features an average temperature of 0 °C to 3 °C, an average precipitation of 295 mm, an average relative humidity below 60%, and evaporation rates ranging from 1500 mm to 2700 mm. Annual sums of sunshine duration correspond to around 2800 h to 3200 h. The Xilin Gol League landform types include the shallow mountain/hill farming-pasturing interlaced zone, Hunshadake sandy land, and a degraded and desertified grassland zone. The Xilin Gol League covers an area of 2.026 × 109 hm2, of which 1.796 × 109 hm2 is grassland, 1.847 × 105 hm2 is the arable land area, and 1.436 × 105 hm2 is for the sowing of food crops. The permanent population of the Xilin Gol League is 1.055 million [21]. Agriculture is the main production source and grassland farming exists to supplement the agriculture.The land-use data used for the study area were based on remote sensing images in the years 1990, 1995, 2000, 2005, 2010, and 2015 [22,23]. These images were produced by the Chinese Academy of Sciences Resource Environmental Data Center (RESDC) based on Landsat TM-/ETM+, Landsat 8 OLI and GF-2. According to Liu’s paper, they constructed a national land-use database for 1990–2015 via human–computer interaction using a high-resolution remote sensing–UAV–ground survey observation system, based on geographic knowledge [22]. The spatial resolution of land-use dataset is 1000 m × 1000 m. According to National Standard Land-Use Classification of China, land-use types in the study area are classified into 6 primary types and 13 subtypes. The primary classifications are farmland, forest, grassland, water, build and unused land. The comprehensive evaluation accuracy of the first level of land use is >93% and that of the second level is >90%, which can meet user mapping accuracy demands at a scale of 1:1 million [23]. Annual normalized difference vegetation index was based on the bimonthly dataset synthesized by Global Inventor Modeling and Mapping Studies (GIMMS) with a spatial resolution of 8 km (http://ecocast.arc.nasa.gov/data/pub/gimms/3g.v0/). The time series was from January 1990 to December 2015. The population data, per capita, and the annual average precipitation and temperature were derived from the Resource and Environment Data Cloud Platform, Institute of Geographic Sciences and Natural Resources Research, China Academy of Science (http://www.resdc.cn/). The spatial resolution was 1000 m × 1000 m. The data of soil erosion and wind and sand fixation from 1990 to 2015 came from the National Earth System Science Data Center. The spatial resolution was 1000 m × 1000 m. Socioeconomic data (i.e., total water resources and total agricultural output value) from 1990 to 2015 came from the Inner Mongolia Statistical Yearbook.① Land-Use Dynamics DegreeThe formula of the comprehensive land-use dynamic degree [24] is
+          (1)LC=∑i=1nΔLUi-j2∑i=1nLUi×1T×100%
+          where LUi is the area of i-th land-use cover at the beginning of monitoring (hm2); ΔLUi-j is the area of land-use cover i transforming to land-use cover j for the entire study region; T is the length of monitoring time (year).② Landscape Pattern IndexIn this paper, patch density (PD), the interspersion and juxtaposition index (IJI), patch richness (PR), the contagion index (CONTAG) and Shannon’s evenness index (SHEI) were selected. These indices can reflect the landscape structure [25]. The landscape pattern indices were analyzed by Fragstats 4.2 software [26].③ LULC Change Direction ModelThe LULC change direction model (LCDM) [27] is used to explore the direction and significance of LULC changes. The formula is
+          (2)LCDM=∑i=1n[Aij×(DJ−DI)]A×100%
+          where i is the i-th land-use cover; j is j-th land-use cover transformed from the i-th in a specific period; Aij is the area of i-th land-use cover transforming to j-th; A is the total transformed area of all land-use types in the entire study area during this period; Di and Dj represent the ecological level of the LULC types before and after transformation (Table 1). A higher LCDM value indicates better ecosystem functions, whereas the lower the LCDM value, the more negatively the ecosystem is functioning.Based on the PSR model and taking the regional characteristics into account, the ecosystem health evaluation indicators were selected from the three layers of pressure, state and response [28]. The exact composition is shown in Table 2. The weights of the indicators were calculated by the entropy weighting method [29]. The raw judgment matrix [30] of the evaluation indicators was normalized using the extreme difference method. The formulae for calculating weight are as follows:(3)wi=1−Him−∑i=1mHi(0≤wi≤1,∑i=1mwi=1)
+          (4)Hi=−k∑j=1nfijlnfij,i=1,2,3,…,m
+          (5)fij=rij∑j=1nrij,k=1lnn
+          where Wj characterizes the extent to which this indicator affects ecosystem health, i ∈ [1, n], j ∈ [1, m]; i is the i-th evaluation indicator, j is the j-th evaluation object; r is the statistical value of the j-th evaluation object on the i-th evaluation indicator.Ultimately, the ecosystem health index (EHI) [31] was calculated. The range of EHI is from 0 to 1. For a more precise representation of the ecosystem health status, drawing on research results of experts [32] and combining the value of the ecological health index of the Xilin Gol League, the values of ecosystem health indices were classified into five levels: very healthy (0.8–1), healthy (0.6–0.8), sub healthy (0.4–0.6), unhealthy (0.2–0.4), and sick (0–0.2). The formula of the EHI model is as follows:(6)EHI=∑i=1nωi×bij
+          where EHI reflects the ecosystem health index, i ∈ [1, n]; i is the i-th evaluation indicator; wi is the weight of the i-th indicator; bij is the normalized value of the i-th indicator.① Annual Normalized Difference Vegetation Index (NDVI)To ensure consistency in the resolution of indicators, we fitted an Empirical Orthogonal Teleconnections (EOT) model [33] to resample GIMMS NDVI to a regular 1-km grid. According to Florian Detsch, the EOT algorithm performed reasonably well across space [34]. The maximum value composites (MVC) was used to synthesize monthly NDVI data. Based on monthly data, the average value of NDVI in the growing season from April to October was calculated as the annual NDVI using the mean value method [35]. The formulae are as follows:(7)MNDVIi=MAX(NDVIi1,NDVIi2)
+          (8)NDVIa¯=17∑410MNDVIi
+          where i is the i-th month, i ∈ [4,10]; MNDVIi represents the maximum NDVI value of the i-th month; NDVIi1 and NDVIi2 represent the NDVI value for the first and second half of the i-th month; NDVIa¯ represents annual NDVI for year a, a ∈ {1990, 1995, 2000, 2005, 2010, 2015}.② Ecological ResilienceEcosystem resilience [36] was calculated from changes in vegetation type to determine the range of resilience forces or the size of the variable margin of an ecosystem. The formula is as follows:(9)ER=∑i=1mSi·Pi
+          where ER reflects the value of ecosystem resilience; i is the i-th land-use cover; Si is the area ratio of each landscape type; Pi is the resilience value of the i-th landscape type (Table 3).③ Ecosystem Services ValueEcosystem services value is the benefits that humans receive directly or indirectly from an ecosystem. Based on the characteristics of the study area, drawing on the research experience of Li et al. [37], who improved Xie’s service value calculation formula applicable to the national level [38] and derived formulae suitable for calculating the ecosystem services value on regional scale, we decided to use the following formulae:(10)ESV=∑[Ak×VCk×NPPsNPPcn×(2×GDPmsGDPm1+e2.5−1En)]
+          (11)v=1.05Pre1+(1+1.05PreL)2
+          where ESV is ecosystem services value; Ak is the area of land-use type k; VCk is the value of ecosystem services per unit area of land-use type k in China (yuan/ha) [37]; NPPs and NPPcn represent the net primary productivity of natural vegetation in the study area and in Cshina(t/ha/a); GDPm is China’s GDP per capita in 2002 (yuan/person); En is the Engel coefficient for the study area in 2002; v is actual annual evapotranspiration (mm); Pre is annual precipitation (mm); L is yearly mean evapotranspiration (mm).In GeoDA software [39], Moran’s I [40] index is utilized to represent spatial correlation. It reveals the correlation between the values of a certain property of adjacent patches, ranging from −1 to 1. The bivariate local indicator of spatial association (LISA) displays the clustering of the two properties in every patch. Applied to the spatial analysis of ecosystem health, the spatial correlation pattern between △Area% (change in forest and grass coverage) and △EHI (change in ecosystem health index) can be shown by visualizing the distribution map and reveals the pattern of spatial differentiation of ecological health in the study area.NDVI was used to assess grassland productivity and quantify LULC changes. The NDVI in the Xilin Gol League featured a stripped distribution, which gradually declined from the east to the west. The annual NDVI for the entirety of the Xilin Gol League in 1990, 1995, 2000, 2005, 2010 and 2015 were 0.509, 0.499, 0.450, and 0.458, 0.473 and 0.456, respectively. Over the past 25 years, the trend of NDVI in the Xilin Gol League had been “gradually dropping–gradually rising–slowly dropping”. The NDVI dropped from 1990 to 2000 and, in 2000–2015, the vegetation gradually improved compared to the 1990s. Figure 2 presents the NDVI spatial changes in the Xilin Gol League from 1990 to 2015. Over the past 25 years, negative NDVI was concentrated in the southwest of the Xilin Gol League, where the most serious cases were the Sonid Left and Right Banner. Positive changes mainly appeared in Dong Ujimqin.To evaluate the direction and rate of LULC change in the Xilin Gol League, we calculated the land-use dynamic degree and made a transfer matrix diagram (Figure 2). The trend of change in the land-use dynamic degrees between 1990 and 2015 was “rapid increase–rapid decrease–slow decline”. The land-use transition map revealed that the conversion between grasslands with different coverage was the most dominant land-use transfer type from 1990 to 2015. There was a significant increase in dynamic degrees after the implementation of the GGP. In terms of the spatial scope, grasslands transition was concentrated in the western part of the study area. The overall vegetation status in the study area started to take a turn for the better in 2000.The conversion between farm, forest, and grasslands is bound to affect ecosystem functions. The LULC change direction model was developed to assess directional changes in different LULC types. We calculated the LCDM values based on formula (2). The results are as follows: LCDM1990–1995 = −0.0113%, LCDM1995–2000 = 0.0255%, LCDM2000–2005 = −0.1036%, LCDM2005–2010 = 0.0084%, and LCDM2010–2015 = 0.0590%. Between 1990 and 2015, the LCDM value showed an “up–down–up” trend. Negative values were recorded in 1990–1995 and 2000–2005. The changes in the landscape structure of the Xilin Gol League from 1990 to 2015 are shown in Table 4. It could be seen that when the values of patch density (PD) and patch richness (PR) increased, Shannon’s evenness index (SHEI) decreased then increased. The values of the interspersion and juxtaposition index (IJI) decreased and then increased, while the contagion index (CONTAG) increased.From 1990 to 2015, the spatial changes in the landscape pattern of Xilin Gol are shown in Figure 3. IJI, PR, and SHEI all increased in Plain and Bordered White and Taibus Banner, which is located in the south of the study area. In Sonid Right Banner and the Dong Ujimqin, there was no obvious change in IJI, while PR and SHEI decreased significantly. CONTAG increased in West Ujimqin, Sonid Right Banner, Duolun and Taibus Banner, and declined in Plain and Bordered White. Generally, IJI, PR, and SHEI changes were not visible, with slight rises in some areas. CONTAG fluctuated in most parts of the research area and the rising trend was more pronounced and the distribution wider. These plaques were mainly distributed in the south of the study area.During 1990–2015, the ecosystem health index (EHI) in the Xilin Gol League ranged from 0.48 to 0.50 (Table 5). From 1990 to 1995, the study area showed a small increase in EHI values. From 1995 to 2000, there was a significant decline in EHI. During the decade 2000 to 2010, EHI values showed an upward trend until 2010, when the study area showed a small decrease in EHI values. However, the overall EHI of the Xilin Gol League increased from 2000 to 2015. Overall, the EHI in the study area showed an upward trend over the 25 years. Observing the calculation results of each standard level, the change trend of the total score in the pressure layer was “rising–falling–rising”, and the score of the Xilin Gol League increased over the 25 years. The total score in the state layer continued to decline in the early period, and it only picked up as of 2005. The total score in the response layer declined from 1990 to 2000 and began to rise in 2000. This was consistent with the start of the GGP in the study area.The EHI spatial distribution was calculated based on the PSR model in the study area from 1990 to 2015 (Figure 4). The level of EHI was higher in the northeast of the Xilin Gol League than in the southwest. The change in the ecological health of the northeast was not significant. In contrast, a significant improvement occurred in the southeast when EHI increased in areas and decreased in areas with low EHI over time.During 1990–2000, ecosystem health significantly deteriorated in the western part of the Xilin Gol League, especially in the Sonid Right Banner. From 2000 to 2015, the areas with a dropping EHI in the Dong Ujimqin Banner (located in the south of the Xilin Gol League), Sonid Left Banner, Sonid right Banner, Xilin Hot, and Bordered Yellow (located in the southwest of the Xilin Gol League) gradually expanded, meaning that the EHI of these league cities steadily dropped after 2000. However, the EHI of Dong Ujimqin Banner and Xilin Hot in the east of the study area improved when EHI values increased by more than the area where the EHI value decreased.Univariate Moran’s I indexes for 1990, 1995, 2000, 2005, 2010, and 2015 were calculated as 0.279, 0.230, 0.226, 0.258, 0.274, and 0.169, respectively, which showed positive spatial correlation. To test whether Moran’s I was significant, Monte Carlo was used to simulate 999 tests in Geoda. All the results passed the significance test of 0.001, indicating that the spatial autocorrelation was significant at a 99.9% confidence level. From 1990 to 2000, Moran’s I index decreased, implying that the EHI spatial correlation gradually declined. From 2000 to 2010, Moran’s I index steadily increased and the EHI spatial distribution started to show a noticeable trend of clustering in the same direction. During 2010–2015, the EHI spatial autocorrelation in the study area suddenly declined.The local indicator of spatial association (LISA) (Figure 5) showed that high–high clustering (p ≤ 0.01) prevailed in the northern area of the study area. The most significant cluster appeared in Dong Ujimqin (p = 0.001). In the southwest of the Xilin Gol League, most of the areas showed insignificant correlation.Bivariate local Moran’s I indexes for 2000–2005, 2005–2010, and 2010–2015 were 0.183, 0.270, and 0.313, respectively, which meant that △EHI (change in ecosystem health index) was positively spatially correlated with △Area% (change in forest and grass coverage), and the correlation strengthened within 15 years of the GGP implementation. In the correlation between △Area% and △EHI in 2005–2010, the Dong Ujimqin Banner, West Ujimqin Banner, and Abag showed a high–high correlation (p = 0.05), while Xilin Hot showed low–high correlation and the correlation in other counties were not significant. In 2010–2015, Dong and West Ujimqin Banner showed a high–high correlation (p ≤ 0.05), XilinHot showed low–high correlations and the significance increased (p = 0.01). Particularly, a significant low–low correlation appeared in Plain and Bordered White (p = 0.001).Grassland was the main LULC type in the Xilin Gol League, constituting 83.87% of the total LULC from 1990 to 2015 and existing mainly in the form of high and medium coverage. Furthermore, while forest and arable land gathered in the northeast and south of the study area, the utilized land was scattered among the grassland, and the construction land was interspersed among the arable land. Since the implementation of the GGP in 2000, apparent changes have occurred in the Xilin Gol League’s land-use, with arable land slowing down the pace of increase, and the area of grassland changing from a rapid decline to a slow deterioration and even a rise. As for land-use transition types, the transition between different grassland types was the primary change type in the 1990s. After 2000, the transition within different grassland types gradually decreased and the transition of arable land and unutilized land appeared. In terms of the spatial scope, the grassland change areas were concentrated in the west of the study area and the transition distribution of the unutilized area was extensive. Analyzing the changes in land use from the perspective of landscape pattern, the changes in SHEI and IJI values showed a turning point in 1995, indicating that the landscape diversity and connectivity in the study area have declined. However, this situation has gradually improved since 2000.According to NDVI analysis of the Xilin Gol League from 1990 to 2015, the vegetation coverage of the study area changed in 2000, which could be attributed to reclamation. By comparing the distribution and area changes in NDVI in different thresholds over the past 25 years, we confirmed that vegetation coverage changed. The vegetation growth worsen from 1990 to 2000, while in 2000–2015, the vegetation gradually improved. Combining relevant policies, the change mainly benefited from the GGP in the study area since 2000. A significant influence of the implementation of the GGP increased vegetation coverage. Over recent decades, climate conditions in the study area have been stable. The LULC changes were consistent with the time nodes and spatial distribution of NDVI changes. Therefore, we concluded that NDVI changes were mainly caused by reclamation.The changes in LULC and NDVI indicated that the policy of the GGP in the Xilin Gol League was promoted step-by-step and region-by-region between 2000 and 2015. From 2000 to 2005, the study area banned grassland restoration to effectively curb grassland reclamation. During this period, forest and grass coverage annually increased and areas with a low value reduced. Forest and grass coverage in most areas were at a medium level. From 2005 to 2010, the transition of arable land and unutilized land to grassland appeared, and grassland increased by 535.8 km2, suggesting a significant increase in forest and grass coverage. Areas with high NDVI reached their most senior over 25 years. From 2010 to 2015, the implementation of the GGP loosened. At the same time, the dynamic degrees of land utilization was only 0.6027%. In small areas of the southwest, high coverages of grassland were replaced by low grassland coverage and the annual average NDVI decreased. LCDM results from 1990 to 2015 showed that LULC changed and was affected by the return of arable lands to forests and grasslands. The gradual increase in LCDM values after 2005 suggested that LULC changes had beneficial effects on ecosystem functions in the Xilin Gol League, providing preliminary evidence of the positive effects of the GGP on ecosystem health.The results of ecosystem health showed that the Xilin Gol League ecosystem remained in a sub-healthy state. In general, the health of the ecosystem of the study area improved (Table 5). The increase in the total value of the pressure layer proved that with the development of economy and society, population pressure and economic pressure were increasing. At the same time, with the global resource shortage and climate deterioration, the land and water resources in the study were gradually decreasing. Analyzing the results of the state and response layer, the improvement effect of the GGP launched in Xinlin Gol in 2000 on the function of regional ecological structure only began to appear in 2005, while the government and the people’s response was immediate. The Wj value calculated by the entropy method can characterize the impact of the indicators on ecosystem health. The higher the value of Wj was, the more significantly the indicator impacted ecosystem health. It can be seen from Table 2 that the Wj values of arable land area per capita and grassland area per capita were the largest in the pressure layer, indicating that these two indicators had a major impact on ecological health. Both are processed based on LULC data with a resolution of 1 km, indicating that unreasonable land-use was the greatest pressure exerted on the ecosystem and the main driving factor to reduce EHI.Moran’s I index for all six periods showed positive values around 0.25, indicating that the spatial distribution of EHI values did not exhibit complete randomness, but rather there was some degree of clustering. Across the study area, large EHI areas were highly–highly clustered (Figure 5), suggesting that the spatial correlation in the Xilin Gol League was characterized with higher EHI areas adjacent to each other. This meant that a healthy ecosystem could drive the surrounding ecosystem in a positive direction. From 1990 to 2000, the number of not-significant-type areas increased and gathered in the southern part of the study area, indicating that the ecosystem in the study area in the 1990s tended to be fragmented, and spatial differences in EHI emerged. From 2000 to 2010, the percentage of areas featuring the high–high clustering rose to 9.58%, which suggested that more and more areas within the Xilin Gol League tended to be healthier in recent years. These areas had also driven the health of the surrounding areas.In the 1990s, the health of the ecosystem gradually worsened because large areas of land were transitioning to arable lands in order to develop the local agriculture, which was adapting to population growth. The forest and grass coverage also dropped, leading to increased pressure in the research area. The worsening areas of ecological health were most obvious in Dong Ujimqin Banner and Duolun County, where arable lands were concentrated. From 2000 to 2010, the GGP was comprehensively implemented in the Xilin Gol League, which effectively alleviated grassland degradation. The decreasing per capita arable land area at the pressure layer and increasing NDVI at the state layer were mainly attributed to the rising EHI of the study area over 10 years. The state of health deteriorated from 2010 to 2015 because the Xilin Gol League failed to vigorously implement the GGP, which resulted in an increasing reclamation rate of lands and a decreasing NDVI. The degradation of grasslands with high and medium coverage in Sonid Right Banner increased plaque density, decreased landscape diversity, and destabilized the study area’s system structure.With the launch of the GGP, EHI changes and areas of forests and grasslands were in direct proportion to each other. The correlation strengthened annually. After the implementation of the GGP, large areas of arable land and unutilized land were transferred to forest and grassland, leading to an increasing landscape abundance of the forest and grass ecosystem, and the interconnection of adjacent plaques. This meant that the GGP improved the health state of the ecosystem in the research area. The Xilin Gol League turned farmland to forest in 2000, and in 2003, turned farmland to grasslands. In the late period of the GGP project (2010–2015), emphasis on the GGP implementation weakened, resulting in a recovery speed slowdown of forestland and grassland. In five years, the EHI demonstrated negative changes in the whole area and was responsible for the spatial relations of the low–low clustering area between the forest and grass coverage changes and the EHI indices in the clustering diagram.In areas with relatively poor ecological conditions, it is necessary to adopt effective measures to curb the trend of grass degradation and realize the balance among “humans, livestock, and grass” by protecting grassland environmental security. The environmental protection red line of grassland resources should be demarcated to protect the grassland areas and types within the environmental protection red line. Abuse of grassland resources should be banned to ensure the sustainable development of grasslands. As for other areas, protection of grassland with high and medium coverage should also be strengthened to prevent the regression of grassland to low coverage. Grassland should be restricted from being reclaimed into arable land. The GGP should be carried out on wasteland without too much utilization value. The GGP should be tailored to different geographical locations, climatic, and stand conditions. There should no longer be a limit on the ratio of restored ecological grasslands to economic grasslands. The focus should shift to increasing vegetation cover and mobilizing peasantry enthusiasm, making the GGP a conscious action of the general public to protect the ecological environment and improve production. Therefore, a subsidy policy may be established based on the prescribed task of the GGP. The local government should strengthen the advertising among the pastoralists and enhance herdsmen’s awareness of the social, economic, and cultural values of grassland resources, as well as the role of grassland in protecting the environment. The local government should also improve the operation means and economic efficacy of the animal husbandry to gradually upgrade the traditional internal value of grassland to landscape ecology. Meanwhile, the local government should enhance the monitoring of grassland resources and improve grassland investment mechanisms.We propose the following management measures according to the EHI of the Xilin Gol League:The basic grassland types in the agricultural area of Siqi County, located in the south of the Xilin Gol League, should be divided. Taipusi Banner, Zhengxiangbai Banner, Sonid Right Banner, and Zhenglan Banner should refer to the advanced experience of Duolun County with favorable ecosystem health in order to finish and subcontract the grassland division. This would guarantee personnel, equipment, and funds, as well as synchronize on-the-spot dotting and information collection, and standardize archive management.The non-shepherding project should be launched in the north of the Xilin Gol League. The shepherding area should be strictly controlled in ecologically vulnerable areas such as the Ujimqin Sand to gradually realize the withdrawal of sheep raising in the whole area. The grassland ecological subsidy and reward policy should be launched to strictly control the grazing capacity and the scale of grazing in winter on natural grassland. The issuance of subsidies and rewards should be linked with the responsibilities of agricultural and pastoral households to ensure the realization of the ban on shepherding in the shepherding area and also to understand equilibrium in the equilibrium area.Xilin Hot is the seat of the Xilin Gol League government and is situated precisely to the south of Beijing. Xilin Hot governs the environmental civilization construction of the Xilin Gol League. It is necessary to strengthen problem awareness and objective orientation, constructing a better environmental security barrier. Major environmental engineering projects, such as the Beijing-Tianjin Sandstorm Source Phase II Project, the Wulagai River Water Conservation Forest Phase II Project, and other major environmental engineering projects should be implemented to protect grasslands. Meanwhile, specific implementation plans or regulations should be promulgated to guide departments to lower levels of grassland management.The environmental protection red line of the whole Xilin Gol League area should be completed. Representative areas should be chosen from the north, center, and south of the Xilin Gol League, including Dong Ujimqin Banner, Sonid Left Banner, and Plain Blue Banner. These three areas should be selected as pilot areas for red line demarcation and boundary settlement. The grassland dynamic detection and evaluation should be strengthened, and the strictest control system should be established to realize non-grazing in the non-grazing area and prevent overload in the equilibrium area.In order to better achieve ecosystem restoration and healthy ecosystem development in agro-pastoral ecotone, in this study, using the Grain-for-Green program (GGP) as a control for the temporal and spatial variation of ecological health indices, we assessed the process of the reforestation project and its impact on the ecological health of the agro-pastoral ecotone to propose more locally appropriate policies for reforestation and restoration. When evaluating the ecological health of the study area, the PSR model was modified to address its shortcomings. We adopt the principle of large-scale and multifactor indicator selection to improve the accuracy of the results with innovatively introducing ecosystem service values in the state layer. The entropy weighting method was used to calculate the weights, eliminating human interference. During the period of 1990–2000, land-use change was characterized by the reclamation of cultivated land and degradation of grasslands. The NDVI gradually decreased and the landscape tended to be fragmented, while it was effectively controlled during the period of 2000–2015 (after the implementation of the GGP). The study area had been in a subhealthy state for the past 25 years, with high EHI values mainly found in areas with high forest and grassland cover, such as the northern part of the Xilin Gol League. In response to the GGP, the rational use of land led to a decrease in grassland degradation and an increase in vegetation cover in the Xilin Gol League. From 2000 to 2015, the rate of change in forest and grassland area was positively correlated with the change in EHI value, and the area of high–high clustering was dominant, proving that the correct implementation of the policy of returning farmland to forest can effectively improve the ecological health of the mixed farming and pastoral areas. The spatial distribution of EHI was largely influenced by land-use. Up to now, it can be seen that the GGP was carried out in a step-by-step and zoned manner in the Xilin Gol League. Initial results were achieved in reforestation and combating desertification. In recent years, however, it can be seen that the policy of the GGP has slackened. Thus, more attention should be paid to achieving a high degree of integration of ecological, economic and social benefits while at the same time maintaining and consolidating the existing ecological benefits, so as to achieve sustainable development of the regional economy. Reasonable land-use practices are crucial for the sustainable development of the agro-pastoral ecotone. Understanding the rational and efficient use of land resources and developing relevant environmental policies are key issues for the future.Conceptualization, Z.W. and L.G.; methodology, Z.W. and Q.Y.; software, Z.W.; formal analysis, Q.Y. and L.G.; investigation, Z.W.; resources, Z.W., Q.Y. and L.G.; writing—original draft preparation, Z.W.; writing—review and editing, Q.Y. and L.G.; project administration, L.G.; funding acquisition, L.G. All authors have read and agreed to the published version of the manuscript.This research study was supported by the key research project of Chinese Ministry of science and technology (2017YFC0505601) and innovation team project of the Chinese Nationalities Affairs Commission, grant number 10301-0190040129.We are grateful for the comments of the anonymous reviewers, which greatly improved the quality of this paper.The authors declare no conflict of interest.The location of the Xilin Gol League.(a) Changes in the normalized difference vegetation index (NDVI) data from 1990 to 2015; (b) land-use transfer types from 1990 to 2015. (HDG: high-density grassland; MDG: mid-density grassland; LDG: low-density grassland).Spatial change in the landscape index in the Xilin Gol League between 1990 and 2015. (1: Dong Ujimqin; 2: West Ujimqin; 3: Xinlin Hot; 4: Abag; 5: Sonid Left; 6: Erenhot; 7: Sonid Right; 8: Bordered Yellow; 9: Plain and Bordered White; 10: Plain Blue; 11: Duolun; 12: Taibus Banner).Ecosystem health index of the Xilin Gol League from 1990 to 2015. (1: Dong Ujimqin; 2: West Ujimqin; 3: Xinlin Hot; 4: Abag; 5: Sonid Left; 6: Erenhot; 7: Sonid Right; 8: Bordered Yellow; 9: Plain and Bordered White; 10: Plain Blue; 11: Duolun; 12: Taibus Banner)Significant areas of local Moran’s I for the ecosystem health index (EHI) of the Xilin Gol League from 1990 to 2015.Ecological level of different land-use [27].HDG: high-density grassland; MDG: mid-density grassland; LDG: low-density grassland.Index system of the ecosystem health evaluation and their weights.Ecosystem resilience coefficient of different land-use [36].R: ecosystem resilience; HDG: high-density Grassland; MDG: mid-density grassland; LDG: low-density grassland.Changes in the landscape pattern index in the Xilin Gol League.PD: patch density; IJI: interspersion and juxtaposition index; SHEI: Shannon’s evenness index; PR: patch richness; CONTAG: contagion index.Results of ecosystem health assessment in the Xilin Gol League from 1990 to 2015.

Med-MDPI/ijerph_5/ijerph-17-16-05632.txt

@@ -0,0 +1 @@
+The rate of Caesarean section (CS) without medical indication has increased markedly worldwide in the past decades. This study reports the incidence of CS and identifies the determinants of elective and emergency CS as separate pregnancy outcomes in a cohort of Iranian women. Mothers (n = 700) of healthy, full-term infants were recruited from five maternity hospitals in Shiraz. The association between maternal socio-demographic and biomedical factors with mode of delivery was explored using multivariable, multinomial logistic regression. Most mothers underwent either an elective (35.4%) or emergency (34.7%) CS. After adjustment, women were more likely to deliver by elective CS than vaginally if they were older (≥30 year) compared to younger mothers (<25 year) (Relative Risk Ratio (RRR) 2.22; 95% Confidence Interval (CI) 1.28, 3.84), and had given birth at a private hospital (RRR 3.64; 95% CI 1.79, 7.38). Compared to those educated to primary or lower secondary level, university educated women were more likely to have undergone an elective (RRR 2.65; 95% CI 1.54, 4.58) or an emergency CS (RRR 3.92; 95% CI 2.27, 6.78) than a vaginal delivery. Similarly, overweight or obese women were more likely than healthy weight women to have undergone an elective (RRR 1.91; 95% CI 1.27, 2.87) or an emergency CS (RRR 2.02; 95% CI 1.35, 3.02) than a vaginal delivery. Specialist education of obstetricians and midwives along with financial incentives paid to private hospitals to encourage natural delivery may help in the reduction of unnecessary CS in Iran. In addition, to increase their childbirth knowledge and self-efficacy, pregnant women need to have the opportunity to attend purposefully designed antenatal childbirth preparation classes where they receive evidence-based information on natural childbirth and alternative methods of pain control, as well as the risks and indications for CS.When used appropriately, Caesarean section (CS) is an important intervention in reducing maternal and perinatal mortality and morbidity. A recent systematic review [1] reported that at the population level CS rates higher than 10% are not associated with reductions in maternal and newborn mortality rates. It is unclear, however, what the optimal rate of CS should be to minimise maternal and infant morbidity outcomes associated with complicated pregnancies [1]. Iran has one of the highest rates of CS in the world, and in 2009, at 47.9%, it was higher than other countries in the Eastern Mediterranean region [2].The high rates of CS in Iran and other countries suggest that many women undergo a CS for reasons that are not justified on medical grounds. A large proportion of CS in Iran, particularly in private hospitals, are elective CS and performed due to the preferences of the health care provider and/or mother [3,4]. Caesarean sections are not without risk and, as with any surgery, are associated with short- and long-term dangers to the health of mothers and neonates, and may have a negative effect on future pregnancies [5,6].Advanced maternal age, grand multiparity, obesity, and gestational diabetes are biomedical risk factors associated with an increased likelihood of CS in general [7,8,9,10], while in Iran and other low and middle-income countries, socio-demographic determinants include maternal education, workforce participation, and socio-economic status [11,12,13,14,15,16].This study reports the incidence of CS and identifies the socio-demographic and biomedical factors associated with delivery by CS of a cohort of healthy, full-term Iranian infants. To our knowledge, this is the first Iranian study to explore the determinants of elective and emergency CS as separate pregnancy outcomes.This study is a secondary analysis of data collected in the Shiraz, Infant Feeding Survey, a prospective cohort study conducted in Shiraz, Iran, between June 2014 and March 2015. The design of this study has been reported previously [17], but briefly, a cohort of mothers was recruited within 48 h of giving birth in three public and two private maternity hospitals. Participants were followed-up for six months when they attended their local Maternal and Child Health (MCH) clinic for routine postpartum care at 1, 3, 4, and 6 months.Women were eligible to participate if they were Iranian-born, 18 years of age or older, resided in Shiraz, and had delivered a healthy, full-term (≥37 weeks) infant, weighing 2500 g or more and who had not been admitted to the neonatal intensive care unit (NICU) for 72 h or more. Women were recruited sequentially until the target sample size of 700 was achieved, and recruitment from each hospital was proportionate to the number of infants delivered by the hospital in the preceding year Table 1.The data used in this cross-sectional analysis were collected from mothers at the time of recruitment via structured face-to-face interviews conducted by a trained research assistant with some data extracted with permission from the participant’s medical record (fasting plasma glucose (FPG) levels and pre-pregnancy weight and height). The questionnaire used by interviewers was based on a questionnaire previously used to study infant feeding practices in a number of Islamic countries [12,18].The study was approved by the Curtin University Human Research Ethics Committee (HR 31/2014) and the Local Research Ethics Committee of the Shiraz University of Medical Sciences (209/2014). Participation was voluntary and all women provided signed informed consent.The dependent outcome variable in this analysis was self-reported method of delivery (vaginal, elective CS, emergency CS). No distinction was made as to whether an elective CS was requested by a woman or recommended by her obstetrician. The independent explanatory variables identified in the literature as being associated with CS included maternal age (<25 year, 25–29 year, ≥30 year), educational level (primary to lower secondary, high school, university), pre-pregnancy employment status (employed, unemployed), parity (1 child, 2 children, ≥3 children), pre-pregnancy body mass index (BMI) (normal weight-BMI <25 kg/m2, overweight or obese-BMI ≥25 kg/m2), a diagnosis of gestational diabetes or pre-pregnancy diabetes (FPG ≥ 92 mg/dL) according to WHO diagnostic criteria [19], attendance at antenatal classes (yes, no), and infant birth weight (2500–2999 g, 2000–3499 g, ≥3500 g).Data were analysed using IBM SPSS Statistics for Windows, Version 24.0 (IBM Corp. Released 2013. IBM Corp: Armonk, NY, USA). The chi-square test was conducted to examine the bivariate relationship between method of delivery and each of the explanatory variables. As the outcome variable has more than two categorical levels, a multivariable multinomial logistic regression analysis was conducted to investigate the independent association of the explanatory variables and method of delivery. The reference category for the outcome variable was vaginal delivery. All explanatory variables were entered into the initial full model, which was then reduced by using a backward stepwise elimination procedure (cut-off probability for retainment is p < 0.05). The model fit was assessed by the likelihood ratio test. The exponentiated results of the multinomial logistic regression are reported as adjusted relative risk ratios (RRRs), and 95% confidence intervals (CIs) were generated. A p-value less than 0.05 was considered statistically significant.A total of 1571 mothers who delivered in the participating hospitals were invited to participate. Of these, 719 were eligible and 700 (97.4%) agreed to participate. Of the 852 women who failed to meet one or more of the eligibility requirements, 394 resided outside of Shiraz, had delivered an infant that was <37 weeks gestation (n = 129), <2500 g (n = 280), or were admitted to the NICU for ≥72 h (n = 304) (Figure A1).The majority of mothers had delivered by CS (70.1%) either elective (35.4%) or emergency (34.7%). The overall rate of CS was higher in the private (86.3%) than public hospitals (68.1%) (Table 2). The chi-square tests revealed that method of delivery was associated with maternal age, level of education, pre-pregnancy employment status, pre-pregnancy BMI, and type of hospital where delivery occurred. No association was found between method of delivery and infant birth weight, parity or whether a woman had attended an antenatal class, or had gestational diabetes or pre-pregnancy diabetes.The multivariable multinomial logistic regression analysis revealed a number of significant independent associations between maternal characteristics and the mode of delivery, with four of the nine explanatory variables remaining in the final model (Likelihood ratio test p = 0.001) (Table 3). For instance, women were more likely to deliver by elective CS than vaginally if they were older (≥30 year) compared to younger mothers (<25 year) (RRR 2.22; 95% CI 1.28, 3.84) and had given birth in a private hospital (RRR 3.64; 95% CI 1.79, 7.38), but no such significant associations were found with the likelihood of delivering by emergency CS. Compared to those educated to primary or lower secondary level, university educated women were more likely to have undergone an elective CS (RRR 2.65; 95% CI 1.54, 4.58) or an emergency CS (RRR 3.92; 95% CI 2.27, 6.78) than a vaginal delivery. Similarly, overweight or obese women were more likely than healthy weight women to have undergone an elective CS (RRR 1.91; 95% CI 1.27, 2.87) or an emergency CS (RRR 2.02; 95% CI 1.35, 3.02) than a vaginal delivery. Iran has one of the highest rates of CS in the world and inappropriate CS represents a major health system problem [20]. This study revealed that seven out of every ten mothers of healthy, full-term infants delivered by CS, which is comparable to an incidence of 62.2% reported in a contemporaneous study conducted in Shiraz in 2013 [21]. Just over a third of deliveries in this study were purported to be emergency CS, a rate that far exceeds the acceptable rate of 10 to 15% of deliveries recommended by the World Health Organization (WHO) [5], suggesting that many of these may not have been medically indicated. CS rates are highest in Tehran and developed provinces [2], and the increase in CS over the last three decades has been attributed to socioeconomic development and ‘modernisation’ [11]. In this study, approximately half of all CS were elective operations, and a steady increase in elective CS explains, at least in part, the steady rise in overall CS rates [2].Maternal obesity and gestational diabetes are associated independently with an increased risk of a range of adverse pregnancy outcomes affecting both the mother and the offspring in the short- and long-term [8,9]. Both are associated with an increased incidence of CS [8,9] and macrosomia [9,22]. Therefore, an elective CS may be suggested by an obstetrician or requested by a mother with one or both of these conditions as a means of reducing the maternal fear and pain, and obstetrical difficulties and risks associated with delivering a large birth weight baby vaginally. In this study, there was no independent association between mode of delivery and infant birth weight or a diagnosis of gestational diabetes or pre-pregnancy diabetes; however, compared to healthy weight women, those who were overweight or obese were more likely to undergo either an elective or emergency CS than to deliver vaginally.Similarly, advanced maternal age is associated with a variety of adverse maternal and perinatal pregnancy outcomes [10], and these risks may inform an obstetrician’s decision to perform an elective CS. In this study, while there was no association with maternal age and emergency CS, compared to younger mothers aged less than 25 years, women aged 25 years and older were more than twice as likely to undergo an elective CS than deliver vaginally. This finding was independent of parity and is consistent with other Iranian studies, which have reported that the rate of CS increases steadily with maternal age [11,13,21].The rate of CS, was higher in the private hospitals studied where almost nine out of every 10 women delivered by CS compared to just under seven out of every 10 women in the public hospitals. Higher rates of CS have been consistently reported for private hospitals compared to government run hospitals in Iran, and this difference is explained primarily by a higher rate of elective operations in private hospitals [2,11,21,23]. In this study, just over seven out of every 10 women delivered by elective CS in private hospitals compared to three out of every 10 women in public hospitals. While earlier studies have not always differentiated between the determinants of elective and emergency CS, in this study, women delivering in a private hospital were close to four times more likely to have an elective CS than a vaginal delivery, but there was no significant association with emergency CS and type of hospital.While women were not asked in this study if they had requested a CS, other Iranian studies have reported that women delivering in private hospitals are more likely to request an elective CS than those delivering in public hospitals [23]. For instance, Ghotbi and colleagues reported in a study of public and private hospitals in Tehran that, overall, 20.8% of women requested a CS; however, the rate of maternal request was much higher in private hospitals (42.4%) than in public hospitals (6.1%). Older, employed, and better educated women were the most likely to request a CS [23].In this study, women educated to primary or lower secondary level were less likely to undergo either an elective or emergency CS than university educated women. This association has been reported consistently in other Iranian studies [11,13,21], and there are a number of probable reasons why less educated woman are more likely to deliver vaginally. Less educated women may come from more traditional backgrounds where vaginal delivery has been the norm in their family and community and, therefore, they may have more positive attitudes towards natural childbirth, or alternatively, they may be fearful of surgery. While a vaginal delivery is free in public hospitals, there is a fee for a CS operation and, therefore, less well-educated women, who are more likely to be of low income, will be less likely for financial reasons to request an elective CS. Conversely, there is a cultural perception of social prestige attached to being able to afford the costs associated with having a CS [4], which helps explain the higher rates among better educated and employed women [23].Fear of childbirth is universal and a major reason why women elect for a CS [24,25]. Iranian women do not receive health education about the reproductive system and childbirth in school [23,26] and, in general, have poor knowledge about childbirth and various modes of delivery [23]. As a consequence, women’s attitudes and beliefs about childbirth are influenced by normative beliefs [4], and their preference for CS is largely informed by their mother, relatives, and friends who may share negative natural childbirth stories and positive CS stories [26]. Several qualitative studies have investigated the reasons why Iranian women request CS, and these primarily include fear of childbirth (labour pain, injury to mother or infant) and the belief that CS is a safer mode of delivery and that vaginal delivery carries increased risk of complications (vaginal prolapse, urinary incontinence, sexual dysfunction) [4,13,26]. These same views are often held and promoted by obstetricians who describe vaginal deliveries as time consuming and unpredictable, with obstetricians preferring to perform a CS because they perceive them to be safer and that they have more control over the process [3,27].While CS can be lifesaving under emergency medical conditions there are a number of adverse consequences associated with unnecessary CS. Planned CS have been associated with an increased risk of death and a number of postpartum complications including wound hematoma, hysterectomy, major puerperal infection, anaesthetic complications, and prolonged hospital stay [28]. In addition, a CS is more costly to perform than a vaginal delivery; hence, when performed without medical indication, diverts much needed money from other sectors of the health care system [2]. Finally, CS is associated negatively with a variety of breastfeeding outcomes [18,29]. There was evidence of a negative impact on breastfeeding in this study, and compared to women who delivered vaginally, those who delivered by either elective or emergency CS were less likely to initiate breastfeeding within the first hour and their infant was more likely to receive infant formula during their hospital stay [30], with both these practices being associated with reduced duration of breastfeeding [31].Research has shown that awareness of the benefits of natural childbirth is associated with a decreased preference for CS [32]. Structured antenatal classes have been shown in a Danish study to increase childbirth self-efficacy, which is associated with lower levels of anxiety and pain [33]. Despite universal access to antenatal care, childbirth education programs are uncommon in the Iranian antenatal care system [34]. A lack of suitable facilities, audio-visual equipment, and education materials are reasons given by health care providers for why Iranian women hardly ever receive group childbirth education [35]. High workloads and staff shortages mean that health care providers have only a few minutes with each woman during antenatal visits to provide advice and answer a woman’s questions [35].In this study, only one in every eight women attended an antenatal class, and of these, more delivered vaginally (16.3%) than by elective CS (9.3%, p = 0.069); however, the numbers were insufficient to detect statistical significance. Nevertheless, a small quasi-experimental study conducted in central Iran provides evidence of the effectiveness of antenatal classes attended by primigravida women [36]. Compared to the control group, there was a significant improvement in the childbirth knowledge and attitudes of the intervention group, and significantly more vaginal deliveries.The high rate of CS in Iran is recognised as a major health system problem requiring a multi-strategy solution [27]. To this effect, the Iranian Ministry of Health has implemented the Promotion of Natural Childbirth (PNC) program, which reportedly has started to have a modest effect in reducing the rate of CS [20]. In addition to the strategies included in the PNC, further reductions in CS may be seen with the specialist education of health professionals aimed at improving attitudes towards natural delivery and changing their behaviours [3], and if private hospitals were to receive financial incentives for natural delivery [20]. More needs to be done, however, to change the childbirth attitudes and behaviours of mothers [34]. Routine antenatal childbirth preparation classes should be conducted for pregnant women in which the risks and benefits associated with different modes of delivery and methods of pain relief are discussed, and where they can ask questions and express their concerns. If made mandatory, antenatal classes could not only have an immediate effect on CS rates, but a cumulative effect on CS rates potentially would be felt over the long-term as more women have positive natural childbirth experiences and subsequently report these experiences to other women through their family and social networks.There are several limitations to consider when interpreting the results of this study. The primary limitation is that this study recruited only mothers of healthy, full-term infants. The majority of women deemed ineligible to participate had delivered infants who were either premature, of low birth weight, or had health problems requiring them to be admitted to the NICU for an extended period. Therefore, the incidence of CS in the hospitals studied is likely to be higher than that reported here. Findings are based on self-reported data, and as this was a secondary analysis of data collected as part of infant feeding, information on the reason for the CS and whether it was medically justified and prior history of CS in multiparous women or history of gynaecological diseases was not obtained from the mother, nor were the opinions and medical judgements of the obstetricians available. Despite these limitations, the findings of this study are consistent with those of contemporaneous studies in Shiraz [21] and Tehran [23].This study reveals that rates of elective and emergency CS were four to five times higher than that recommended by WHO. CS were more commonly performed at private than public hospitals, and the high and increasing rate of CS suggests that many women undergo a CS for reasons that are not justified on medical grounds. There is an urgent need for a multi-strategy approach to halt and reverse this trend, which includes financial incentives for private hospitals, specialist health professional education, and childbirth education for pregnant women. To increase their childbirth knowledge and self-efficacy, women should have the opportunity to attend group education antenatal classes where they receive evidence-based information on natural childbirth and alternative methods of pain control, as well as the risks of and indications for CS.Conceptualization, M.Z., J.A.S. and C.W.B.; methodology, M.Z., C.W.B., J.A.S., Y.Z.; formal analysis, M.Z., Y.Z. and J.A.S.; data curation, J.A.S., M.Z.; writing—original draft preparation, M.Z., J.A.S., C.W.B.; writing—review and editing, M.Z., Y.Z., C.W.B. and J.A.S. All authors have read and agreed to the published version of the manuscript.This research received no external funding.We would like to thank the mothers who participated in this study for their assistance so willingly given and the staff of the hospitals involved in the project and the staff of Shiraz University of Medical Sciences, Shiraz, Iran.The authors declare no conflict of interest.Flow Diagram of Participant Recruitment and Follow-Up.Number of participants recruited from each hospital.Characteristics of participants and association with mode of delivery.a Column percentage; b Row percentage; c Chi-square test; d Healthy weight = BMI < 25 kg/m2, Overweight or obese = BMI ≥ 25 kg/m2; e FPG ≥ 92 mg/dL.Maternal characteristics associated with Caesarean (elective or emergency) delivery.a Adjusted for mother’s pre-pregnancy employment status, attendance at antenatal classes, parity, gestational diabetes or diabetes in pregnancy, and infant birth weight. RRR adjusted odds ratio, CI confidence interval.

Med-MDPI/ijerph_5/ijerph-17-16-05633.txt

@@ -0,0 +1,21 @@
+Human-gait-phase-recognition is an important technology in the field of exoskeleton robot control and medical rehabilitation. Inertial sensors with accelerometers and gyroscopes are easy to wear, inexpensive and have great potential for analyzing gait dynamics. However, current deep-learning methods extract spatial and temporal features in isolation—while ignoring the inherent correlation in high-dimensional spaces—which limits the accuracy of a single model. This paper proposes an effective hybrid deep-learning framework based on the fusion of multiple spatiotemporal networks (FMS-Net), which is used to detect asynchronous phases from IMU signals. More specifically, it first uses a gait-information acquisition system to collect IMU sensor data fixed on the lower leg. Through data preprocessing, the framework constructs a spatial feature extractor with CNN module and a temporal feature extractor, combined with LSTM module. Finally, a skip-connection structure and the two-layer fully connected layer fusion module are used to achieve the final gait recognition. Experimental results show that this method has better identification accuracy than other comparative methods with the macro-F1 reaching 96.7%.In recent years, robotic exoskeleton has become an emerging technology in medical, living, industrial and military applications. Among them, lower extremity exoskeleton has important research value in the medical field, its main potential is to enhance the patient’s ability to move in rehabilitation therapy, and enhance physical function after receiving treatment, and hope to improve their quality of life as much as possible. Among them, gait recognition technology is an important technical guarantee for the robot to process a large amount of instantaneous time series data, which is one of the most important features to display the posture and phase of each specific patient [1]. Therefore, there is an urgent need to accurately judge the gait phase of the human lower extremity state change in order to enhance the consistency and coordination of human-computer interaction [2]. In medical disease-diagnosis and rehabilitation research, effective analysis of gait phases has also achieved remarkable results, which has been used in clinical treatment plans for stroke, Parkinson’s disease, brain trauma and other diseases [3,4]. Note that traditional walking analysis is expressed by detecting different gait phases based on motion information (e.g., angle, speed or acceleration) of knees, ankles and hips while walking or running. For example, Fino et al. [5] used abnormal gait phases to detect concussion or mild head injury. Mathieu et al. [6] proposed a novel adaptive dynamic time warping based on Hidden Markov to analyze gait for identifying persons with physical disabilities and provide them with appropriate alerts by monitoring walking. In order to overcome the problem of poor adaptive ability caused by pure mechanical structure, some researchers have begun to recognize the phase of the lower limbs of the human body through programming and algorithms to achieve the purpose of controlling wearable auxiliary devices [7]. For instance, Ruiming [8] mentioned gait subphase recognition of high-quality is significant to the control of lower-limb powered exoskeletons.According to previous studies, gait-phase-recognition methods can generally be divided into two categories. The first type is the threshold method, which determines the corresponding phase information by setting the corresponding threshold [1]. However, this type of algorithm is too rough and difficult to deal with complicated situations. In recent years, with the development of artificial intelligence technology, many researchers began to input different types of sensor data into deep learning models to achieve the purpose of detecting gait phase. For example, Mukherjee, etc. [9] proposed a deep-learning method using machine vision to detect pedestrian gait phase in real time. However, the camera is susceptible to interference from the external environment when capturing images. Moreover, this method is susceptible to the limitation of the use space. Ryu et al. [10] proposed an SVM method to process the electromyography (EMG) data collected during gait to identify the four sub-gait phases of pedestrians, thereby improving the above-mentioned problem of being easily interfered by the environment. However, commercial sEMG acquisition equipment is bulky, expensive and extremely inconvenient to wear, and it is easily affected by sweat stains during the collection of EMG data. In order to overcome the shortcomings of the above-mentioned technology, Ding et al. [11] proposed using the proportion-based fuzzy algorithm to process foot pressure signals to realize gait-phase-recognition, but the plantar pressure is also more susceptible to the wearer’s weight, shoe size and load. What is worse, the failure rate of pressure sensors is also relatively high, making it difficult to be widely applied in reality.In recent years, many researchers have begun to study the use of inertial sensors (IMU) to achieve gait-phase-recognition methods. This is mainly because more abundant information of human movement can be obtained by using a small number of IMUs. Moreover, the IMUs are non-invasively installed on relevant parts of the body, which will not cause harm and too much inconvenience to the wearer [12]. Simultaneously—in the process of collecting IMU information—the signal is difficult to be interfered by the wearer’s own weight, load, and sweat on the wearing part. Compared with the detection method of plantar pressure or muscle electrical signal, the IMU detection method has obvious advantages. In addition, the cost of inertial sensors is relatively low [13], and the inertial information of certain parts of the body during the movement of the human body also has periodic characteristics. Many researchers place the IMU on the instep, calf and thigh. For example, Yan et al. [1] designed a voting-weighted integrated deep learning algorithm, and by inputting acceleration signals on the instep, calf and thigh into the model, successfully detected four subphases of pedestrians, and achieved very good results. Identify the effect. Zhen et al. [14] combined LSTM and DNN models to design an LSTM–DNN deep learning algorithm, and also detected four sub-phases of pedestrians by accelerating the acceleration signals on the instep, calf and thigh. Of course, using more IMUs can collect more phase information, but the requirements on the equipment will be very high, and more input signals will also cause a greater calculation burden on the recognition algorithm model. In addition, too many sensors will have a greater impact on the wearer. If there are multiple external sensor signal inputs, the model should first process the signals simultaneously before processing to ensure that the input signals are sent at the same time, which is undoubtedly very difficult for researchers. Therefore, the researchers gradually shifted their attention to the scheme of using a single IMU to detect the gait phase. For example, Manchola et al. [15] positioned the inertial sensor on the instep, used hidden Markov model as the phase recognition model, and obtained better results than the threshold-based algorithm. In view of the previous research, this article considers placing the IMU in the position of the lower leg, through continuous optimization algorithm model to obtain excellent recognition effect. Gohar et al. [16] Using the inertial information of the chest to realize the identification of personnel. It is important that this sensor application has little impact on personal activities and daily life, and is easy to wear, which is helpful for clinicians to judge early intervention, treatment effects and patient rehabilitation progress.Although we have always been interested in IMU technology, there is still a lack of reality for actual research. This study uses multiple wearable sensors for verification to monitor walking status and gait stability in actual clinical practice. In a multi-IMU-based method, a sufficiently intelligent and diverse gait phase pattern recognition model with multi-sensor data preprocessing analysis and information fusion technology is the most critical problem to be solved for gait of the lower limbs to detect various wearable motion systems or rehabilitation Exoskeleton. To achieve this goal, some statistical learning or machine learning methods are used to calculate the spatiotemporal and biomechanical parameters of walking gait. The mainstream solution is to build a variety of shallow structural models, including hidden Markov models [15], Boosting [17] and support vector machines (SVM) [18]. These models are carefully analyzed based on physical and statistical analysis by selecting the threshold parameter. The raw data or processed data are used to divide the gait stage. These methods construct feature engineering to adaptively learn model parameters and obtain hidden relationships and information between historical data. However, gait phase detection is still a challenging problem because the high sampling frequency data collected from sensors always contains the complex nonlinear relationship with multiple components makes it impossible to apply traditional models to analyze sensory data and distinguish walking information in real time.The emerging deep learning technology has excellent ability to detect complex time relationships. Thanks to breakthroughs in the design and training of model architectures with complex structures composed of multiple processing layers or nonlinear transformations, unprecedented improvements have penetrated many aspects of intelligence, including large-scale visual classification, natural language processing and time series forecasting. Such rapid research progress has also attracted the attention of relevant researchers and companies to build software and hardware to identify the gait stage of snapshots in real life. In particular, convolutional neural networks (CNN) and recurrent neural networks (RNN) have been used to extract the motion characteristics of time-series time data obtained from IMU’s accelerometers and gyroscopes. For example, Omid [19] and others designed a deep convolutional neural network (DCNN) to extract discriminative features from the 2D extended gait cycle and jointly optimize the recognition model in a discriminant manner to complete accurate recognition of the gait phase. Due to their ability to process two-dimensional signals (such as images), most CNNs must convert time-series inertial data into energy images or visually segmented data. This does help to utilize the characteristics of spatial relationships in the gait-phase-recognition, but when processing sequential time series data captured by the IMU sensor, it will obviously ignore the time law and periodic changes, which is often difficult to measure the continuous motion trajectory and extract The quality characteristics of the lower extremities are unrestricted. Therefore, RNN and its improved models including long short-term memory (LSTM) network and gated recurrent unit (GRU) network believe that the current output layer captures a high degree of nonlinearity and time through the time recording sequence and the parameters of the previous hidden layer. Sequential relationship-IMU serial data have attracted wide attention of researchers. Gao et al. [20] used RNN to accurately collect acceleration data to complete the control of the prosthesis. Khokhlova et al. [21] used the LSTM model to classify Normal and Pathologic gait. In addition, CNN can also be combined with traditional machine learning methods, such as SVM [22]. In [22], CNN is used as a feature extractor, and then the extracted features are classified by SVM. At the same time, researchers have found that CNNs tend to ignore temporal continuity when processing only a single time stamp of data [23]. Therefore, it is more common to introduce RNN. The time information of the recording order is passed to the current hidden layer by passing the previous hidden layer state. The LSTM network [24] and GRU network [25] are all RNN. [26] recommended to combine LSTM and CNN for high-dimensional feature representation and character detection. In this way, CNN is usually used as a feature extractor, and then LSTM is used to further process the gesture features extracted by CNN [27]. Gait-phase-recognition was introduced in the comparison of DNN, CNN and RNN in many application scenarios [28].CNN can use convolutional feature maps to process inertial data, while RNN can also process them as time series. However, research on how to combine these two types of neural networks to achieve better gait-phase-recognition is still lacking. LSTM can mine the timing information in the information better, while CNN can mine the spatial information in the signal better. Therefore, this article combines LSTM and CNN to identify the gait phase. However, if the two are simply combined together, the gradient disappears because the network is too deep, so this paper designed the skip-connection structure and batch normalization layer to alleviate this phenomenon.Existing research and results show that the FMS-Net model proposed in this paper is effective in gait-phase-recognition. However, most of the data for these works were collected under road conditions under specific walking conditions. Identifying gait phases under complex conditions is still challenging and requires further research.The rest of the study is organized as follows: The second section introduces data sources and preprocessing techniques and then introduces the gait-phase-recognition model constructed in this paper in detail. Third, evaluate the experimental results of the model through related comparison methods and conduct related discussions. Finally, Section 4 presents our conclusions.In terms of experimental data, 16 volunteers with body weight ranging from 46 kg to 70 kg and height ranging from 158 cm to 177 cm were selected to collect IMU data. The height and weight distribution of the subjects is shown in Figure 1. The subject’s legs or feet did not have any diseases that could affect normal walking.With the advancement of sensor processing technology and algorithms, this study selected one IMU modules to collect the corresponding inertial information. Input data in this work only include Lower leg calf acceleration signals. The hardware characteristics required for signal acquisition will be introduced next. To collect lower limb calf acceleration signals, the JY901 nine-axis angle sensor (the type is Uxin Electronics Co., Ltd., Gansu, China) with Kalman filtering algorithm is used in this paper, as shown in Figure 2. There are two communication modes that can be selected: serial port communication and I2C communication. In order to cooperate with the microcontroller, serial port communication is selected for this topic. The TX, RX, VCC and GND pins corresponding to the serial communication are used to connect to the microcontroller. The microcontroller selected is STM32C8T6, which is a 32-bit microcontroller based on the ARM Cortex-M core STM32 series, the program memory capacity is 64 KB, the required voltage is 2 V–3.6 V, and the operating temperature is −40 °C–85 °C; the operating frequency is 72 MHz.The inertial sensor module is placed outside the lower leg. The arrangement of acceleration sensors for calf monitoring lower limb movement is shown in Figure 3. The system flow of the entire experimental process contains the data collection, processing and application parts. The acceleration resolution of the nine-axis inertial sensor module (MPU9250) used in the experiment is 0.0005 g, the stability of the attitude measurement is 0.05°, and the transmission baud rate is set to 115,200 bps.During the experiment, all participants were required to walk normally on the same treadmill at a speed of 0.78 m/s, 1.0 m/s and 1.25 m/s for at least 120 s. All participants were asked to normally walk 3 times at each speed. All participants have the same sports environment in the same state. In order to prevent participants from affecting the later movement gait due to continuous exercise, the experiment requires all participants to rest for 2 min after completing the designated walking test each time to alleviate the possible impact of exercise fatigue on walking gait. In addition, when collecting data, it should be noted that we only start saving data after the running speed of the treadmill reaches the set speed. When the treadmill starts to slow down, we stop collecting data and complete the data collection process.Each input vector contains three acceleration data along the x, y and z directions. Let two sequences of gait data a→ be the input of the network, which is expressed as:(1)ax=(ax,1,ax,2,…,ax,T)
+        (2)ay=(ay,1,ay,2,…,ay,T)
+        (3)ax=(az,1,az,2,…,az,T)
+        (4)a→=(ax,ay,az)  
+        where ax,ay and az respectively represent the inertial acceleration in the X, Y and Z directions, and T is the length of the input sequence.According to the above experimental settings, we can obtain the acceleration data curves in the X, Y and Z directions collected by the inertial sensor, as shown in Figure 4.The human walking process is a cyclical movement, the complete gait cycle is from one-sided heel landing-to-landing again [29]. Although, two phase model recognition systems are sufficient to control active knee orthosis [30]. However, the most widespread method currently relies on four-phase identification technology [31], which are represented as Flat Foot (FF), Heel-Off (HO), Heel Strike (HS) and Swing phase (SW). This gait four-phase detection model has been successfully used to drive ankle-foot orthosis robots [32].According to previous studies and the scientific nature of the gait phase division, this article also divides the walking cycle into HS, FF, HO and SW phases. During normal walking, the acceleration signal on the calf has a strong periodicity. Studies have shown that the swing phase segment accounts for about 40% of the entire gait cycle and the standing phase accounts for about 60% of the entire gait cycle. According to the previous analysis [1], the schematic diagram of gait cycle division is shown in Figure 5.The next step is to design an algorithm model to identify the relevant human gait phase through the input acceleration time series vector signal. Since the input acceleration signal is a time series signal, its current signal will have a strong correlation with the previously generated signal, so we need a network model that can mine the internal time series information. Among them, thanks to the design of the hidden layer, the LSTM network can handle timing signals very well. However, LSTM can only mine the timing information in the signal, and it is easy to ignore the spatial information of the signal. Therefore, this paper proposes the fusion of spatiotemporal neural networks. In order to alleviate the gradient disappearing phenomenon in the transmission process, this paper introduces batch normalization (batch norm) layer and skip connection structure. This design can reduce the use of the dropout layer, which can further improve the performance of the network. The entire structure is showed in Figure 6.As shown in Figure 6, the FMS-Net model for gait-phase-recognition is composed of CNN, LSTM and multiple fully connected layers. CNN can extract important spatial features in the data and reduce the number of network parameters through parameter sharing. In addition, LSTM, as a feature extractor, can obtain the corresponding timing features well. Then, they are followed by two complete fully connected layers, which are used as classifiers. What we need to emphasize is that the “skip-connection structure” proposed in this paper is that the input vector of the first fully connected layer is the superposition of the output vector of the LSTM and the input vector of the entire network, not the addition. The expression is shown in Equation (5).In addition, in order to prevent the problem of gradient disappearance, batch normalization is used in this paper. Its expression is shown in Figure 6. Learning a deep network is a complicated process. As long as the input layer of the network changes slightly, the network parameters of the subsequent layers will be accumulated and amplified. Once the distribution of input data in a layer of the deep network changes, then this layer of network needs to adapt to learning this new data distribution. During the training process, in order to improve the situation where the data distribution of the middle layer of the network changes, Ioffe et al. [33] introduced batch normalization. The process of batch normalization is shown in Equations (6)–(9). In order for our network to learn to recover the feature distribution that the original network would learn, this learnable reconstruction parameter γ,β is introduced.
+        (5)xFC1=concat  (olstm,x)=[olstm   x]
+        (6)μBN←1m∑i=1mxi
+        (7)σBN←1m∑i=1m(xi−μBN)2
+        (8)xi∧←xi−μBNσBN2+ε
+        (9)BNγ,β(xi)←γxi∧+β
+        where xFC1,olstm, x, m, μBN and σBN represent the input vector of the first Fully connected layer in the FMS-Net network, the output vector of the LSTM network, the input vector of the FMS-Net network, batch size, batch mean and batch variance, respectively and γ,β are learnable reconstruction parameters.In this paper, LSTM and CNN is selected as the combined classifier, and some of the network parameters are shown in Table 1. In addition, num_units in the LSTM network is 36, forget_bias is 0.7 and Activation is Relu. The focus of research is the design of the neural network structure. CNN can share weight information through convolution kernel and reduce network parameters. The latter’s LSTM network structure can enhance the learning effect of the network. Finally, specify the final learning rate and set it to 0.05. Finally, the neural network outputs the classification results through the Softmax regression layer.The FMS-Net algorithm is a model for multiple classification tasks. However, the output of the neural network does not conform to the probability distribution, so it is necessary to convert the output of the neural network into a probability distribution through the Softmax function. The expression of the Softmax function is shown in Equation (10). Then calculate the classification result by Equation (11). Finally, the cross-entropy loss of the model is calculated by Equation (12) and the model parameters are optimized by the gradient descent method.
+        (10)softmax(qi)=eq′i∑i=1neq′i
+        (11)O=max(q)
+        (12)l=−∑i=14yilog(qi)
+        where, yi denotes the indicative variable (0 or 1), if the category is the same as the sample category, it is 1, otherwise it is 0; qi denotes the predicted probability that the observation sample belongs to category i.For each input vector x, the predicted output of the network is q=(q0,q1,q2,q3). After Equation (10), the value of qi is between 0 and 1, and the larger the value, the greater the probability that x belongs to the real label. Based on the output qi, we can get the class label as O.As can be seen from Equation (12) that, the cross entropy is a positive number. When the probability value of the true label qi in the vector q is smaller, larger difference between qi and yi will result in a larger cross-entropy value. This property will help the convergence of the network in the training.In order to avoid overfitting, we chose 70% of the sample set for training and 30% of the samples for testing. After using the same training set to train different models 10,000 times, use the same test set to test the trained model, and record the classification accuracy, macro-F value and macro- of each classifier after testing the classification model with the test machine accuracy (AUC). Then evaluate the performance of all models based on these three indicators.In order to prove the classification performance of the proposed FMS-Net network, we need to draw the corresponding conclusion through corresponding indicators. As we all know, Accuracy is a good comprehensive indicator, which is widely used in evaluation indicators. However, in the classification, it is difficult to characterize the performance of a certain model simply by relying on Accuracy and we must choose other indicators to comprehensively characterize the classification performance of a certain model. In classification problems, commonly used classification performance indicators also include precision, recall and F1. Among them, precision and recall are widely used in the field of information retrieval and statistical classification. These two indicators are used to evaluate the quality of the model results. precision is used to measure the accuracy of the retrieval system. recall is used to measure the recall of the retrieval system. Of course, we hope precision and recall results are as high as possible. Generally speaking, if both precision and recall are high, we can conclude that this model performs well in this classification task. We hope that there is an indicator that can represent the performance of the model in both precision and recall. F1 comprehensively considers the influence of P and R and can comprehensively measure P and R. However, this study studies a multiclassification task and cannot directly use F1. The most direct method is to calculate macro-F1 [34]. Accuracy reflects the ratio of correctly classified samples to total samples. The above evaluation factors are shown in Equations (13)–(18), where TP, TN, FP and FN represent true positive, true negative, false positive and false negative, respectively.In multi-classification tasks, we also have an indicator that is often used to measure classification performance. The more famous is the area under the ROC. The ROC chart was first publicly proposed by Spackman (1989) when performing machine learning and he proved the important role of ROC curve in model evaluation. [35]. In recent years, it was widely used in the fields of machine learning and deep learning. People also realize that simple classification accuracy cannot measure the comprehensive performance and performance of the designed model [29]. We can make the conclusions more reliable by comparing AUC.
+        (13)Accuracy=TP+TNTP+FP+TN+FN
+        (14)Pi=TPTP+FP
+        (15)Ri=TPTP+FN
+        (16) macro-P=1n∑i=1nPi
+        (17)macro-R=1n∑i=1nRi
+        (18)macro-F1=2×macro-P×macro-Rmacro-P+macro-RAs shown in Figure 7, Figure 8 and Figure 9, the confusion matrix provides the performance of visual gait sub-phase recognition. The vertical axis of the matrix represents the actual classification category of the test and the horizontal axis represents the corresponding predicted classification category. In addition, in the confusion matrix diagram, “0.0” represents the “HS” phase, “1.0” represents the “FF” phase, “2.0” represents the “HO” phase and “3.0” represents the “SW” phase. These nine matrices are the average recognition results of all subjects under different walking pace conditions. The value in the main diagonal is the proportion of samples correctly classified. As shown in Figure 7, all the confusion matrices, except for the HS phase, perform quite well. The HS phase is mostly incorrectly classified as the FF and SW phases. In order to verify the effectiveness of the proposed recognition model, we implemented two other gait-phase-recognition methods, namely LSTM and LSTM + CNN. The corresponding confusion matrix is shown in Figure 8 and Figure 9. It can be drawn from Figure 8 and Figure 9 that the LSTM and LSTM + CNN models cannot identify the HS phase and directly categorize most of the HS phases into adjacent FF phases by mistake. The LSTM and LSTM + CNN models have achieved good recognition effects on other phases.From the confusion matrix, we can get Table 2. As shown in Table 2, the macro-F1 of the four groups (HS, FF, HO, SW) of the FSM-Net model differ greatly. When the pace is 0.78 m/s, F1 is, respectively It is 63.7%, 96.4%, 97.6% and 98.8%; when the pace is 1.0 m/s, macro-F1 is 76.6%, 97.6%, 98.2% and 99.0%; when the pace is 1.25 m/s, macro-F1 is It is 54.9%, 97.1%, 97.4% and 98.1%. It can be seen from the above data that FF, HO and SW perform best and obtain a better recognition effect, exceeding 96%). The performance of HS phase recognition is the worst. As for the recognition accuracy of each sub-stage, the SW phase performed best, with the maximum value being the group (99.0%) at a pace of 1.0 m/s. The recognition effect of FF phase and HO phase is also quite good. Obviously, the performance of the HS phase recognition effect is the worst, none of the macro-F1 values reaches 80% and the macro-F1 with the lowest HS phase recognition is only 54.9%. For the LSTM and LSTM + CNN models, macro-F1 for HS phase is 0, showing the worst recognition effect for HS phase; for SW phase recognition, the minimum value of macro-F1 is 97% and 98%.We also evaluated the performance of the proposed algorithm model and the other two algorithm models in terms of ROC curve. The results are shown in Figure 10, Figure 11 and Figure 12. Through the ROC curve, we can calculate the corresponding macro-AUC and the additional accuracy and macro-F1 statistics to Table 3. It should be noted that “NO-skip” in Table 3 means that the “skip connection” structure is not added to the FMS-Net network. By comparing the FMS-Net network before and after adding the “skip connection” structure, it can be concluded that the skip connection structure has a certain improvement effect on the recognition performance of the FMS-Net network. It is also worth noting that our method obtained a higher macro-F1 value and accuracy, higher than all other groups method.Gait analysis provides an opportunity to assess walking behavior. Gait analysis can be used for various applications, such as rehabilitation, clinical diagnosis and physical activity [36]. The acceleration signal generated when the human calf is walking on a flat ground is a regular signal, and this information can be extracted by using the IMU. Although the FMS-Net has shown certain effectiveness in the detection of gait events, it still needs further optimization in the future. In this study, the IMU needs to be placed on the designated position of each subject’s calf. However, due to each subject’s height, weight, gender, walking habits, etc., the sensor cannot be accurately placed at the designated location, and can only be installed at an approximate designated location, which requires further study.The IMU used in this article is the JY901 sensor. The JY901 sensor uses Kalman filtering to filter the collected data better, filter out redundant noise and ensure the quality of the transmitted signal. In addition, the data output rate of the JY901 sensor is 200 HZ, which can ensure the real time nature of the later signal transmission and avoid the occurrence of repetitive signals at the signal receiving end. The measurement range of acceleration is between −16 g and 16 g, and this range can fully meet the needs of this study. It can be seen by observing Figure 4 that the acceleration data collected by the experiment is between −1 g and 2 g, which is completely within the measurable range of the JY901 sensor. 2.4-G wireless communication adopted between JY901 sensor and host computer. The signal collecting terminal is separated from the signal receiving terminal to avoid the signal cable from having an additional effect on the coordination of the body.In order to increase the variability of the experiment, this study asked each subject to perform three different pace experiments on a treadmill and let them walk three times at each different pace. Although the collected data have certain differences, there is still a big gap from the complex walking state in reality, and further research is needed.In reviewing the literature, regarding gait events, the ANN algorithm model was used to achieve an 82.2% recognition accuracy rate for the IMU under different walking conditions [37]. When using the IMU to classify the five gait phases, an accuracy of 82% can be achieved [34]. In addition, gait-phase-recognition is very important for the development of calf assist devices because they are strongly related to gait events [35]. In order to propose an acceleration data acquisition system that can be applied to the masses, we tested the recognition effect of the FMS-Net algorithm proposed in this study on unlearned acceleration signal data. This study found that the proposed FMS-Net can successfully predict gait events for test set, and the accuracy of phase recognition for HS, FF, HO and SW is up to 96%, and based on acceleration signals, detection of HS, FF, HO and SW phase seems to be reliable.The core technology of gait-phase-recognition system is the design of recognition algorithm model. This paper proposes the FMS-Net and uses it to detect the HS, FF, HO and SW phases. According to the results obtained in this study, the acceleration signal has relatively high stability when walking on the ground, which satisfies the research in this study. In this study, the FMS-Net algorithm combines LSTM, CNN, skip-connection and other structures. The algorithm model has certain complexity. Although good results were achieved, it should continue to be optimized in the direction of lightweight networks in the future.When walking on flat ground, the acceleration signal on the human calf is a typical time-series signal. LSTM is a classic algorithm model for processing time-series signals and CNN is a typical algorithm for extracting spatial information from signals. LSTM + CNN directly combines the two, but it is easy to cause the gradient to disappear as the depth becomes larger. This study takes full advantage of the advantages of LSTM in processing timing signals and CNN’s convolution operation in extracting spatial features and uses skip-connection structure and batch normalization to solve the problem of deep gradient disappearance and design FMS-Net algorithm model. From the results of Figure 7, Figure 8 and Figure 9, the three models show good recognition effect on the FF, HO and SW phases, but LSTM and LSTM + CNN cannot accurately identify HS. Although LSTM + CNN is still unable to identify the HS phase, it is superior to LSTM in the recognition of the other three phases compared to the LSTM algorithm. The FMS-Net algorithm by adding skip-connection structure has been further improved compared with LSTM + CNN. Through Figure 9, we can see that the FMS-Net algorithm has further improved the phase recognition of HS and can recognize most of the HS phase. Although the FMS-Net algorithm can identify part of the HS phase, it is still lower than 80%, so further optimization and improvement are still needed. By observing Figure 10, Figure 11 and Figure 12, it can be seen that the AUC of the HS phase of the FMS-Net algorithm is also superior to LSTM and LSTM + CNN. It can be seen from Table 1 that the LSTM and LSTM + CNN recognition of HS phase F1 is 0, while the average F1 of FMS-Net algorithm for HS phase recognition at three paces is 65.1%, which may be better than HS During the transmission of phase data, the gradient disappears. It can be seen from Table 2 that the performance of the three is not much different in accuracy, FMS-Net is the best, LSTM + CNN is the second and CNN is the worst. However, the performance of macro-F1 is quite different. The average macro-F1 of FMS-Net at three paces is 89.5%, while the average macro-F1 of LSTM + CNN and CNN at three paces is 72.4% and 71.4%. In terms of AUC, FMS-Net also performs best. As can be seen from Table 3, the recognition performance of FMS-Net when walking at a speed of 1.0 m/s is the best among the three. The reason for the result may be that 1.0 m/s is relatively close to the normal walking speed, but this conclusion still needs to be proved by adding more control groups. In the future, we need to set more walking speed control experiments to obtain more reliable conclusions.Even if FMS-Net shows its usefulness in classifying acceleration signals detected by gait events, other deep-learning methods need to be used for further evaluation. Future work should improve classification accuracy by improving feature extraction and gait-phase-recognition algorithms. In this study, it is considered acceptable to install a wearable inertial sensor module only on the lower leg compared to other wearable sensors. However, in fact, when the subject wears the sensor for a long time, it may have a potential impact on the subject’s gait, which requires further exploration. In the future, we will try to use new gait functions (for example, gait dynamic images [38,39]) instead of neural network input original x, y, z to verify its effect on gait-phase-recognition, which may be one of the future jobs.In order to meet the application of gait-phase-recognition technology in the control of lower extremity dynamic exoskeleton, this study propose a low-cost, easy to implement and efficient IMU-based gait sub-phase recognition system. First, we constructed a wireless calf acceleration signal acquisition device. Then, we preprocess the collected data in order to train the classifier for the subsequent use of the data set. Finally, a novel classifier FMS-Net applying seamlessly combining LSTM and CNN models by applying the skip-connection structure is established to extract acceleration signal features and predict gait sub-phase. Experiments and discussions prove that the FMS-Net method has better classification accuracy with the macro-F1 up to 96%, which is superior to other integrated algorithm models. The results show that the proposed method can effectively perform gait-phase-recognition, which lays a solid foundation for the application of gait-phase-recognition technology in the control of lower extremity dynamic exoskeleton.T.Z. conceived and designed the whole structure of the paper under the supervision of J.-l.K. and L.Y.; J.-l.K. and T.Z. accomplished experimental work and wrote the paper under funding acquisition of L.Y. All authors have read and agreed to the published version of the manuscript.This research was funded by the Fundamental Research Funds for the Central Universities (No. 2015ZCQ-GX-03), the National Key Research and Development Program of China (No. 2017YFC1600605) and Beijing Municipal Education Commission (No. KM201910011010).The authors declare no conflicts of interest. The research complies with Institutional Research Ethics Board regulations.Information about volunteers participating in this experiment.Schematic diagram of inertial sensors (IMU) connected to single chip microcomputer.Human gait-information acquisition system.Acceleration data collected under the calf. a_x, a_y and a_z represent the acceleration data in the X-axis, Y-axis and Z-axis directions collected by the experimental equipment, respectively.Phase-division diagram of gait.Proposed network (FMS-Net) architecture.Confusion matrix obtained from LSTM classifier at three different paces: 0.78 m/s (a), 1.0 m/s (b) and 1.25 m/s class (c).Confusion matrix obtained from LSTM+CNN classifier at three different paces: 0.78 m/s (a), 1.0 m/s (b) and 1.25 m/s class (c).Confusion matrix obtained from FMS-Net classifier at three different paces: 0.78 m/s (a), 1.0 m/s (b) and 1.25 m/s class (c).ROC curve performance derived from LSTM classification under three pace settings: 0.78 m/s (a), 1.0 m/s (b) and 1.25 m/s classes (c).ROC curve performance derived from LSTM+CNN classification under three pace settings: 0.78 m/s (a), 1.0 m/s (b) and 1.25 m/s classes (c).ROC curve performance derived from FMS-Net classification under three pace settings: 0.78 m/s (a), 1.0 m/s (b) and 1.25 m/s classes (c).Parameter setting of FMS-Net structure.Summary of classification performance of different models at unsynchronized speed.Summary of classification performance for different training functions.

Med-MDPI/ijerph_5/ijerph-17-16-05634.txt

@@ -0,0 +1 @@
+SARS-CoV-2 virus infections in humans were first reported in December 2019, the boreal winter. The resulting COVID-19 pandemic was declared by the WHO in March 2020. By July 2020, COVID-19 was present in 213 countries and territories, with over 12 million confirmed cases and over half a million attributed deaths. Knowledge of other viral respiratory diseases suggests that the transmission of SARS-CoV-2 could be modulated by seasonally varying environmental factors such as temperature and humidity. Many studies on the environmental sensitivity of COVID-19 are appearing online, and some have been published in peer-reviewed journals. Initially, these studies raised the hypothesis that climatic conditions would subdue the viral transmission rate in places entering the boreal summer, and that southern hemisphere countries would experience enhanced disease spread. For the latter, the COVID-19 peak would coincide with the peak of the influenza season, increasing misdiagnosis and placing an additional burden on health systems. In this review, we assess the evidence that environmental drivers are a significant factor in the trajectory of the COVID-19 pandemic, globally and regionally. We critically assessed 42 peer-reviewed and 80 preprint publications that met qualifying criteria. Since the disease has been prevalent for only half a year in the northern, and one-quarter of a year in the southern hemisphere, datasets capturing a full seasonal cycle in one locality are not yet available. Analyses based on space-for-time substitutions, i.e., using data from climatically distinct locations as a surrogate for seasonal progression, have been inconclusive. The reported studies present a strong northern bias. Socio-economic conditions peculiar to the ‘Global South’ have been omitted as confounding variables, thereby weakening evidence of environmental signals. We explore why research to date has failed to show convincing evidence for environmental modulation of COVID-19, and discuss directions for future research. We conclude that the evidence thus far suggests a weak modulation effect, currently overwhelmed by the scale and rate of the spread of COVID-19. Seasonally modulated transmission, if it exists, will be more evident in 2021 and subsequent years.A novel coronavirus, thought to have made a zoonotic transition from bats, infected a human host in Wuhan, Hubei Province, China [1]. By late January 2020, the virus, newly named SARS-CoV-2, and the disease it causes, COVID-19, had spread to 18 other Chinese provinces, and to Japan, South Korea, Taiwan, Thailand, and the USA. On the date of submission of this review (15 July 2020), there were 13,331,879 confirmed cases, in virtually every country worldwide (213 countries and territories, Figure 1). At the time, it was reported that 577,825 people infected with the virus had died; both numbers have subsequently risen. The only comparable acute respiratory disease pandemic was Spanish Influenza (H1N1), transmitted from birds to people in 1918, which lasted until 1919 and killed an estimated 50 million people worldwide. In the current highly interconnected world, the impact of the COVID-19 pandemic is likely to be felt for many years [2,3,4]. It is therefore crucial that all potential determinants of the rate and location of the pandemic spread receive careful consideration in order to make appropriate plans for its management.Epidemiological models have been used worldwide to guide the imposition (or not) of policy and regulatory intervention [5,6]. These models can be modified to include aspects of social characteristics of the infected populations; and they can also be adapted to reflect the modulating effect of environmental influences on the processes that determine transmission. Many related respiratory diseases show a connection between climate variables and the dynamics of the disease. It is thus plausible that such a dependency could exist for SARS-CoV-2 (reviewed in Section 4). Given that the COVID-19 outbreak began in mid-winter in the northern hemisphere, where it was (at the time of writing) peaking toward the middle of the boreal summer, and that the opposite scenario seems to be playing out in many southern hemisphere countries, it is tempting to associate this pattern with climate seasonality, as many publications have suggested. However, it is also plausible that the association is spurious, related simply to coincidental spatial connectivity between countries. It is necessary to critically assess the evidence for environmental sensitivity, in both the virus and the disease, before arriving at conclusions that may have significant implications. In terms of a response to the pandemic, we need to understand whether and how environmental variables influence the infection rate. This knowledge provides clues for policy and practice to reduce the spread of the virus and potential for treatment options. For example, if analyses show that absolute humidity is strongly associated with reduced infection rates (e.g., influenza [7,8]), artificially raising indoor absolute humidity during periods of low ambient humidity may be an effective intervention. Second, if environmental variables do influence the trajectory of the pandemic, the seasonal progression of the disease will lead to different implications across the globe, varying by hemisphere, region, and climatic zone. In the extratropical northern hemisphere, there would be a real possibility of a second wave appearing during the next winter [9]. Conversely, there is a danger that the initially slow pace of the epidemic in the southern hemisphere could be misinterpreted to mean that proactive management has supressed the disease spread. Given that in the south, where the peak of COVID-19 incidence is likely to coincide with the winter peak of other endemic respiratory illnesses, complicating diagnosis and placing additional strain on the health systems, missing the environmental drivers of COVID-19, if they exist, would be profoundly damaging. As we will argue, many southern hemisphere countries are particularly vulnerable (they are in the developmental ‘Global South’ as well as the geographical south). For these regions especially, clarifying the environmental sensitivity will assist the prioritisation of resources.Third, for longer-term management of the disease, we need to understand whether the seasonal effect will manifest as it does in established or endemic respiratory viruses, in the absence of being able to predict in what period of time (in years) the virus will be eliminated [10].In this review, we consider all the pertinent studies relating to the effect of a range of specific environmental and climatological variables on the biology of the virus and the epidemiology of the disease.In Section 2, we develop our reasoning for why southern hemisphere countries can benefit from the lessons learnt in the north, if the application of that knowledge takes heed of particularly southern hemisphere issues. In Section 3, we briefly present the main classes of epidemiological models, since key parameters revealing environmental modulation are derived from such analyses. In Section 4, we explore environmental sensitivity in extant respiratory viral diseases and past epidemics in order to suggest why seasonally coupled environmental influences are also likely to exist for SARS-CoV-2. Section 5 then critically reviews evidence for such signals in the literature that had accumulated to 15 July 2020. Section 6 summarises our findings, and offers suggestions for future analyses of the seasonal modulation of COVID-19.The situation regarding COVID-19 in southern hemisphere is different from that in the north in three ways. First, while the northern hemisphere is moving out of winter at the time of their peak of infections, the southern hemisphere is moving into winter. Second, a much larger proportion of countries in the southern hemisphere are developing countries, with significant resource limitations in their healthcare systems. Third, many of the countries in the southern hemisphere, and on the African continent in particular, have a much higher incidence of pulmonary diseases such as tuberculosis, immunocompromising diseases such as HIV-AIDS, and a higher prevalence of diseases such as cholera and malaria, which may not be recognised as comorbidity risks in COVID-19 but do place coinciding stressors on the health system. To their advantage, the delayed arrival of COVID-19 in much of the southern hemisphere has allowed these countries the time to observe the efficacy of containment and treatment practices in the Global North, and to adapt their healthcare and policy response accordingly. The initial outbreak of COVID-19 in China, early epidemics in Iran, Italy, and later much of Europe and the United States took place during the coldest months of their year, and were distributed within a narrow climatic band [11,12]. During the early period of the outbreak in January and February 2020, few known cases had been recorded in the southern hemisphere, which was experiencing peak summer conditions. This could reflect a climate sensitivity, but could just as plausibly reflect dominant trade and human movement patterns [13]. Thus, the initial relatively low rates of spread and mortality in southern Africa, Australia and some regions of South America may simply be a result of being at an earlier stage in these epidemics. However, in both the northern and southern hemisphere, influenza and other coronavirus diseases peak during their respective winter seasons [14]. Thus, if climate factors do play a role in COVID-19 infection rates, the concurrence of transition of southern hemisphere countries to their winter season with the mid-stages of the disease transmission trajectory is of concern, especially with respect to containment policy and health system resource allocation.The status of healthcare services in the Global South is of concern even without a climatic component to COVID-19. While Australia and New Zealand have healthcare services as good as any in the northern hemisphere [13], much of South America and sub-Saharan Africa struggle with access to quality healthcare. This is associated with poverty and socio-economic inequalities and result in poor health outcomes and financial risk to the state and individuals [15,16,17,18]. The healthcare sectors are understaffed, underresourced, and understocked under normal conditions, which were working at maximum capacity even before the COVID-19 pandemic [19], and will be severely challenged as COVID-19 cases increase [20,21]. Early evidence from China shows a significant correlation between mortality and the healthcare burden in COVID-19 cases [22]. Efforts to model the preparedness of African countries have highlighted concerns relating to the staffing of testing centres, stock for testing, and the ability to implement effective quarantining both inside and outside of healthcare facilities [20]. The prevalence of pre-existing infectious diseases compounds this issue. In the period 2016–2018, 41 African countries have experienced at least one epidemic, while 21 have experienced at least one epidemic per year [23]. South America is currently struggling with outbreaks of measles in 14 countries, and a tripling of the incidence of Dengue Fever in four countries [24]. Recent outbreaks of diphtheria, Zika and Chikungunya have further stretched the healthcare systems [24]. The most prevalent infectious diseases in sub-Saharan Africa include cholera, malaria, viral haemorrhagic fever, measles and malaria [19]. Of particular concern in the Global South is the possibility of comorbidity with HIV-AIDS and tuberculosis (TB). Many TB cases are pulmonary in nature [25], while patients with HIV are significantly immunocompromised [26]. There is considerable TB-HIV comorbidity [27]. Corbett et al. [28] found a 38% incidence of HIV in TB-infected patients across Africa, and for the countries with the highest HIV prevalence, up to 75% of TB patients also tested positive for HIV. Comorbidity has decreased from 33% to 31.8% over the past decade, and over the period 1990–2017, TB incidence, TB mortality rate and HIV-associated TB have declined in a number of southern African countries [26]. South America has much lower cases of both HIV and TB, and a comorbidity of approximately 10% [29]. While results from Spain suggest that HIV-positive patients currently on antiretroviral treatment have no higher risk of severe SARS-COV-2-induced illness [30], the comorbidity of those with a longer HIV history and TB comorbidity, with or without HIV, is unknown. There are further related concerns pertaining to continued HIV [31] and TB [32] care during COVID-19, as social distancing requires people to stay indoors and hospitals are overstretched. Finally, the relatively delayed spread of COVID-19 to the southern hemisphere has allowed these countries to ‘get ahead of the curve’ through evidence-based management derived in the north [33]. Recent experiences of two Ebola epidemics have meant that many countries in sub-Saharan Africa implemented temperature screening at airports long before the first COVID-19 cases were reported [20], and contact tracing and epidemic management plans are in place [19]. South Africa, Kenya, Uganda and Zambia were reported as having all been particularly proactive in planning for their eventual COVID-19 cases [19]. South America has arguably not been as prepared (Rodriguez-Morales et al. 2020). Studies on modelling risk for the African continent are largely related to importation risk [20], which has been capped due to lockdown in many countries. This form of response is important in delaying the peak and “flattening the curve”, but is unlikely to completely avoid extreme pressure on already stressed healthcare systems [16,22].When assembling datasets from many different locations to test the effects of environment on COVID-19 progression, it is essential that the criteria for determining the infection and mortality rates are consistent across sources. The data used to calibrate and validate epidemiological models (e.g., the COVID-19 Data Repository, Center for Systems Science and Engineering (CSSE), Johns Hopkins University) consist of time series of infections, which often include only those with symptoms sufficiently severe that the patients sought medical assistance, and who subsequently tested positive using a PCR-based test for the presence of the SARS-CoV-2 virus [34]. This is known as the ‘case rate’. As the number of tests increases and includes community-based testing, as opposed to testing only those displaying symptoms, the case rate will converge on a true infection rate. PCR testing is accurate (though reporting is often delayed by days to weeks [35]), but if testing is mostly performed on those presenting symptoms and their close contacts, estimates of the true infection rate inevitably include large biases, especially given the high occurrence of asymptomatic or mild cases. Compensating for this bias requires that the sample frame be weighted to be representative of the population as a whole. As antibody-based tests become more widely used, datasets that indicate post facto what fraction of the general population was exposed to the virus will emerge. Antibody tests have variable accuracy, both in terms of false positives and false negatives [35]; nevertheless, their overall accuracy is much better than the guesswork that otherwise goes into estimating the number infected from the medical case rate alone. It is suspected that mildly infected people and even asymptomatic cases can spread the disease [36], but perhaps less effectively than severely ill individuals. It is likely that recovery from SARS-CoV-2 provides subsequent immunity, with initial indications that this may be persitent [37].The models that predict mortality use a time series of recorded deaths. At a minimum, this includes the deaths recorded in hospitals for people being treated for SARS-CoV-2 at the time of death. More complete records are supplemented with data on people who died in the community or in nursing homes, and were inferred from the symptoms they displayed to have died from COVID-19. For severely-affected areas, it is possible to estimate the anomalies between the COVID-attributed death rate relative to the seasonally adjusted expected population death rate, and infer that these additional deaths (‘excess deaths’) were caused by the pandemic [38]. Where this has been carried out, it suggests that the death rate is substantially higher than that initially reported; however, this approach conflates deaths directly caused by SARS-CoV-2, and those that may have resulted from overburdening the health system. Making accurate estimates of transmission rates requires a sufficient number of cases. Often the models are initiated only once 30 or 100 cases have occurred in a location [39] so that the effect of importation of cases due to travelling may be minimised. Therefore, if the area selected for analysis is too small, the number of cases may be inadequate to support the more data-intensive approaches. In most countries, data are collected daily, but the daily data show a lot of noise, partly for stochastic reasons; also, for spurious reasons such as the effect of weekends, laboratory delays, or recoding the date of reporting rather than the date of testing or infection (Section 5.1). Smoothing the data over periods of a week helps to solve irregular daily data patterns [40,41], but this also means that the analyses are unresponsive to events at finer timescales.The need to match the time period for which infections and deaths are recorded and the period over which environmental drivers are integrated is widely accepted. Similar considerations also apply to spatial resolution. COVID-19 outbreaks are apparently highly clustered, often in small areas. Environmental drivers are also spatially heterogeneous, some much more so than others. The resolution chosen for the environmental data needs to be appropriate for both the grain of the infection process and the grain of the environmental variable. Several analytical typologies have been applied to epidemiological models, mostly based on what factors they take into account [42]. Table 1 is a pragmatic classification of the types thus far predominantly used for COVID-19 projections, based on the logic of their construction. Most of these model types can be implemented either deterministically or stochastically for age-structured or non-age-structured populations; for a single, equally-exposed population or for a spatially disaggregated population with transfers between groups; and using frequentist or Bayesian approaches.Simple extrapolation and phenomenological models are suitable for projections of less than one month into the future, whereas the somewhat mechanistic epidemiology models are more robust for projections months or years into the future. The various classes of models can in principle run at any spatial scale and over any time period, but in practice there are data-imposed constraints. Environmental influences can be introduced into the basic model structures at a variety of points (Figure 2). Where they are introduced and what the models are able to say about the relationship between the environmental influences and infection or mortality rates depend on the theoretical basis of the model (Table 1). Models that best capture the functional relationship of confirmed daily cases across time are best suited for revealing environmental drivers. The phenomenological and compartmental models are the strongest contenders here. The raw time series of confirmed infections and deaths can be time aggregated, and time lagged with respect to the environmental factors, to find the best fits, as long as this is performed consistently, and considers the time lags already built into the model structure.One approach is to establish correlations, either over time or across space, between the infection rate at a given time and simultaneous metrics of environmental factors such as temperature, humidity and UV (see Section 5.4). In SEIR and similar models, two metrics are available for this infection rate: R0, the Basic Reproductive Number, and Rt, the Effective Reproductive Number. R0 is defined as the expected number of secondary infectious cases generated by any single average infectious case in an entirely susceptible population. R0 should be largely free from signals attributed to imposed factors that affect human behaviour. It is typically derived from the initial portion of the growth curve when the disease spreads in a population where everyone is susceptible, before control measures have been put in place (i.e., completely ‘natural conditions’ sensu Shi et al. [50]) or herd immunity had been attained. Neher et al. [51] note that “R0 is not a biological constant for a pathogen” (p. 1) but it is affected by factors such as the infectiousness of the virus, susceptibility of the hosts (e.g., due to age or an assortment of comorbidities), duration of the infectious period, density of susceptible people (also population density and the proportion of the population that is urbanised) or the contact rate with them (including aspects of mobility), and environmental influences (as shown in Figure 2). These aspects are subject to localised idiosyncrasies across the globe and must be accounted for in regional or global analyses when calculating or comparing R0.Rt is a measure of observed disease transmissibility, defined as the average number of people a case infects at any time (t) once the epidemic is underway. Rt incorporates changes in a society’s behaviour (self-regulated responses and non-pharmaceutical interventions or NPIs [52]) as the disease becomes widespread, and varies day to day. These effects are typically stronger than the environmental influences, and can easily mask them or generate spurious associations. It is not advised to base assessments of environmental effects on Rt due to the ‘noise’ that the signal will contain, unless there is sufficient information that permits inclusion of the interventions as continuous, time-varying factors. For the compartment models, it is possible to derive the values of the key model parameters by model inversion, in near-real time, and from these, calculate R0. This needs at least one more observation than there are free parameters to be estimated. In practice, accurate estimates of confidence limits require many more data points than parameters. The multiple observations can come from a single-population time series, but this would limit the degree to which changes over time can be resolved within the parameters themselves. If there are multiple time series from different populations, both temporal and spatial variation of the parameters can be obtained. Phenomenological approaches typically use a variety of parametric regression models (see Section 5.4). It is sometimes necessary to fit a piecewise model to accommodate the breakpoint that develops when country-specific NPIs are introduced. It is generally only possible to compare the parameters of the curves across locations (rather than within locations, over time) to determine whether there is a systematic pattern that relates to any environmental predictors. This is because fitting multi-parameter non-linear curves using data from only a part of the curve (in epidemics, usually just the initial part) is notoriously difficult and uncertain. If the effect of the environmental factors on the model parameters was known, they could be used to alter the curve parameters dynamically, and thus the projected outcomes; but the parameters typically have no intrinsic biological meaning.Seasonality of prevalence is a common feature in most persistent and established or endemic respiratory infectious disease [53,54,55,56,57,58,59,60], as well as many other infectious diseases [61,62], in diseases (or endemic tolerated infections) of both humans and other animals. Peaks incidence periods occur during the shoulder seasons or the winter, oscillating globally with the opposing boreal and austral climate. Seasonally varying prevalence has a general latitudinal gradient and is accentuated in highly seasonal temperate and subtropical climates (with some rare exceptions) but is also observed in tropical regions [63]. Seasonality is found in a wide range of viral respiratory diseases (VRDs)—including influenza viruses, para-influenza virus (PIV), human syncytial virus (RSV), rhinoviruses and human coronavirus strains (HCoV) [55,60]. For endemic viruses causing VRDs in humans, seasonal peaks are usually quite predictable, but interannual variability in onset and duration of any season, and the virulence of respective seasonal strains, vary. It follows, therefore, that if VRD prevalence follows this climatological pattern, a mechanism(s) that connects and modulates the viral disease progression with seasonally varying climatological variables in individuals or populations must exist. This sensitivity must occur in at least one location of the SEIR model (Figure 2). In the case of novel viruses, the role of seasonality is more contentious, mainly because they have not existed for enough time for seasonality to be unambiguously established. The seasonal prevalence of pandemic strains of virus is often conflated with the so-called second wave, which may be coincidentally associated with the following winter season, suggesting that there is a climate-based modulating effect on its incidence [10,64]. In the case of SARS and MERS, the attribution of resurgence to climatological drivers, as opposed to secondary circulation dynamics, remains unresolved [65,66]. Novel viruses are much less predictable than established viruses with respect to their persistence, re-emergence in the following years or seasons, and virulence in later outbreaks [64,67]. Until a novel virus becomes endemic and recycles (in its existing form or as mutated strains), its seasonal prevalence is difficult to assess [68]. The magnitude of the current SARS-CoV-2 pandemic is likely to result in an extended period of persistence [69], and thus if seasonal prevalence exists, it should eventually be unambiguously apparent.In the generalised SEIR model shown in Figure 2, environmental modulation can primarily take place at two stages, namely Susceptibility and Exposure. Environmental sensitivity insights can come from two basic sources. The first is observational data and laboratory studies and analyses of the environmental modulation on the SARS-CoV-2 virus biology and the incidence of the disease it causes (as in this review). Second, we can examine data and information from published studies on respiratory viruses and VRDs and related endemic and novel coronaviruses specifically (see [53,55,57,59,60,67,70] for general treatment of this topic). In this section, we examine three sets of hypothetical mechanisms which explain environmental modulation and seasonality of VRDs other than COVID-19: (i) physical environmental variable modulation, (iii) biological and host behavioural modulation, and (iii) viral molecular and biochemical modulation.Physical environmental variable modulation hypotheses focus on the meteorological correlates of seasonality in the diseases [54,58] and comprise the bulk of such studies. These all follow the basic tenet that selected environmental variables (such as temperature or humidity) vary in space and time with the progressing seasons, and if a mechanism that links them with a VRD can be demonstrated, this makes them a suitable candidate for explaining VRD seasonality. There is a lack of clarity in the literature regarding which definition of humidity is best applied as environmental moderator of respiratory viral epidemiology. Studies employ relative humidity (RH), absolute humidity (AH), specific humidity (SH), vapour pressure or dew point (more in Section 5 below). This renders comparisons and conclusions difficult to reach [7]. RH and SH have strong dependence on temperature, which further complicates studies that include both temperature and humidity as predictors.The postulated mechanisms are usually tested in laboratory studies which monitor the persistence of viable viruses in aerosol droplets and on surfaces [71], perform experimental transmission studies in animal models [72], or study the relationship between observed ambient or indoor environmental variability and infection rate, morbidity and mortality, with the assumption of causality (Section 5). Notably, results from temperate and tropical climate zones (or with ranging latitude) are often contradictory. This has led to a suggestion that different seasonality mechanisms are at play in different climate zones: humidity (aerosol droplet transmission) as the key driver in temperate regions, and precipitation driving behaviorally mediated contact transfections in the tropics [73,74,75]. The environmental determinants of virus transmission in aerosol liquid droplets have received substantial consideration. The premise is that, in winter, characterised by relatively lower humidity, pathogen-bearing aerosol droplets (PBADs) are more persistent. PBADs expelled by infected individuals often contain viruses or bacteria, in a mixture of mucus, saliva and dissolved salts, and can travel up 8 m from a simple sneeze [76]. Upon leaving the airway with moisture saturation close to 100%, PBADs are exposed to much drier air which results in evaporation. They can quickly lose up to 90% of their water mass and reduce in size. At an RH of 40–60%, the water loss greatly increases the salt concentration to levels that inactivate viruses. In contrast, for RH < 40%, the dissolved salts precipitate, resulting in a PBAD with low salt concentration and a high number of infectious viruses [77]. PBADs range in diameters 5–20 μm when the ambient RH is 30–60%, whereas below 30%, a PBAD may immediately reduce its size below 0.5 μm, and become a droplet nucleus [78,79]. Thus, conditions of lower ambient RH result in the production of smaller, lighter (longer floating periods), and potentially more penetrative PBADs, thereby elevating the exposure component of the SEIR model [80,81]. The role of temperature in influencing the prevalence of VRDs is more contested and complex. This is partly because temperature and AH together determine RH, which affects the rate of evaporation and thus PBAD dynamics, as argued above [72,82]; and temperature could also have direct effects. Several studies associate temperature with respiratory disease incidence, some by direct association [83,84] and some focussed on the temperature changes (i.e., lowering temperatures rather than lower temperatures [85]). Temperature may also play a mediating role in other ways. The first set of hypotheses consider the direct effect of temperature on respiratory virus survival. There are very few such studies but they show that viruses in general are surprisingly tenacious, with survival periods of days at room temperature for SARS-CoV-1. Effective inactivation occurs at temperatures of above 56 °C [86,87].Another temperature-mediated mechanism with substantial literature involves the fomite viability of viruses [88,89], particularly in public spaces and hospitals, involving endemic coronaviruses and SARS-CoV-2 [90,91]. Some studies explore the role of temperature alone on specific surface types [88], while others look at the combined role of temperature and humidity [90,92]. Respiratory viruses, including human coronaviruses, can remain viable as fomites on a range of surface types, indoors and in sheltered external environments, at room temperature and higher, for periods of hours to days and from days to weeks on refrigerated surfaces at 4 °C. Persistence depends on both the surface type and the temperature and humidity range (see Table 1 [91] for a recent summary). Thus, the risk of infection from fomites (the exposure element of the SEIR model) increases as temperature decreases. The combination of temperature and humidity has been found important for fomite viability in the endemic human coronavirus HCoV 229E (Table 2). Most studies aim to test sterilisation techniques and the efficacy of personal protective gear [91,93,94,95,96]. One hypothesis posits a predominance of surface contact transmission in the tropical climates, versus transmission through PBADs in temperate climates [97]. A range of other physical environmental variables have been cited as moderators of respiratory viral epidemiology. They often co-vary with other causal variables. Wind and wind speed are relatively neglected as physical environmental factors in infectious disease epidemiology. Given that windy seasons occur in many climates zones, wind should not be discarded as a contributing variable [98]. For influenza, wind has been cited in some instances as a factor in transmission of infectious particles from remote locations, as promoting the extended local transmission of PBADs [99,100], with a convincing account in one case of equine influenza [100]. Barometric pressure has also been considered, for example in the case of respiratory syncytial virus, where it was found to have no statistically significant influence [101]. In other studies, it does have an influence, along with temperature [102]. Guo et al. [103] found air pressure to be a predictor of the risk of influenza infection in children in Guangzhou, China, with a differential effect by age.Rainfall seasonality and disease incidence in general are well described [104], but literature on the relationship between rainfall patterns and VRD epidemiology is restricted to tropical climates. Most studies have considered rainfall either at a very local scale, or as part of a set of meteorological variables being tested. Pica and Bouvier [105] comprehensively review the literature on this rainfall and VRDs, and conclude that for a range of respiratory viruses (primarily influenza and RSV), there are as many studies finding some association as there are studies finding no link. With attenuated intraseasonal temperature variation in the tropics, rainfall provides a key differentiator between seasons, possibly explaining the strong associations between rainfall and respiratory illness prevalence there. The mechanism of association is less clear. There is a suggestion that the tropical rainy season causes crowding, and thus increased exposure [106], another suggesting that reduced sunlight is associated with pneumonia incidence [107], and yet another citing diurnal temperature changes [108]. The improvement in air quality and reduction in allergen production following rainfall may be another mechanism [109].Solar ultraviolet radiation (solar UV) varies greatly with season everywhere and is thus an attractive candidate to explain seasonality of VRDs. UV radiation in laboratory settings is a very effective means of deactivating viruses, and there are a plethora of studies of this effect on all kinds of pathogens (including coronaviruses SARS-CoV-1 and MERS-CoV), mainly targeting hygiene and outbreak management in public spaces and hospitals [110,111,112]. Studies that consider the environmental effect of solar UV (a component of sunlight) without confounding effects of other variables are rare. Sagripanti and Lytle [113] state that, for influenza, “the correlation between low and high solar virucidal radiation and high and low disease prevalence, respectively, suggest that inactivation of viruses in the environment by solar UV radiation plays a role in the seasonal occurrence of influenza pandemics” but concede that there are a range of additional factors that need to be considered. Despite UV being regarded by several authors as the “primary germicide in the environment”, its independent effect as a seasonal driver of VRDs remains uncertain (on this point, for influenza, see [114]).A second set of hypotheses for explaining the seasonality of VRDs consider behavioural and physiological responses to changing environmental variables such as temperature [54,55,60], related mostly to the exposure but also the susceptibility component of the model in Figure 2. These include considering the consequences of confining people in sheltered and enclosed spaces during colder weather, with recirculating air and closer proximity to infected co-inhabitants, thus increasing the likelihood of exposure. They also include the idea that exposure to colder and drier air at the cellular level in the respiratory tract results in impaired physical or immune-system defences to infection, and hence increased susceptibility [60,115]. Large (<30 μm) and medium (<10 μm) inhaled PBADs are normally captured in the upper nasal mucosa and upper respiratory tract, respectively, and are transported towards the mouth (and expelled) through a synchronised circular movement of cilia. The combination of the mucosal layer and cilia can effectively clear the particles [79]. However, low ambient RH has been demonstrated to reduce the effectiveness of both mucosal production and cilia action [60,72]. A corroborating study demonstrates that dry air (low RH) impairs host defence against influenza infection in genetically engineered mice with human-like lung tissue, as well as slowing recovery [116]. The third set of hypotheses consider the biochemistry and molecular adaptation of the viral pathogens [55]. These take into account the temperature sensitivities of the various stages of the virus infection cycle, from binding to the host cell, replication of nucleic acids, the stability of secondary structures of viral proteins, and eventual ejection of the virus from the host cell [55]. Given that there is a gradient of temperature within the respiratory tract, and that breathed air can greatly alter conditions in the upper respiratory tract, susceptibility can increase under cold conditions, especially to viruses which are adapted to be most efficiently infectious at temperatures slightly below normal body temperature [55,115]. Falling somewhere between the physical, physiological and biochemical hypotheses in explaining seasonality of respiratory viruses is the change in susceptibility with varying serological levels of vitamin D. Vitamin D synthesis occurs when the skin is exposed to sunshine, which varies seasonally (confounded with UV, temperature and other variables). Vitamin D has been suggested as an important form of defence against microbes, influenza and pneumonia in particular [117,118,119,120]. Shaman et al. [121] attempted to model this effect on influenza prevalence in the USA and concluded that seasonal variability in other factors such as humidity and even the school calendar were better at explaining their results. These considerations are incomplete, with a final abiotic aspect that must be included. Air pollution refers to a wide range of harmful, primarily geogenic (naturally occurring) and anthropogenic particulate matter, chemicals or gasses that cause negative or dangerous physiological responses and effects in humans and biota. It is well known that poor air quality can have direct and indirect impacts on human health, and in particular on the susceptibility of humans to respiratory viral infections as well and a measurable effect on the severity and mortality rates [122]. Gases such as nitrogen dioxide, ozone and especially particulates classified by size (PM10, PM2.5, and PM0.1) have different pathological mechanisms and effects but are all known to be associated with the increases in viral respiratory disease incidence, hospitalisation or attributed deaths, famously during the London fog of 1952 [123] and the 1918 Spanish Influenza Pandemic [124]. Clifford et al. [125], for example, showed that PM10 inhalation exacerbates the response to influenza, and Ye et al. [126] showed that ‘haze’ (a combination of air pollutants) was associated with the spread of respiratory syncytial virus in children. Air pollution is also known to have a strong seasonality, driven by both seasonal economic production activity and also by ranging seasonal metrological conditions which can either concentrate and trap pollutants in surface air or conversely disperse pollutants and improve air quality [127,128,129]. Therefore, it is a further consideration that seasonal variation in air quality and pollution is an additional factor for consideration as a contributor to the seasonality of respiratory viral infections that have been reported.It is most likely that each of these hypothesised mechanisms has some role, either in unison, or independently or that one mechanism dominates in particular conditions [60]. While the precise mechanism that explains the relationship between environmental factors and disease prevalence is important, particularly because it may reveal optimal management interventions (of transmission and for treatment), statistical attribution of a strong correlate may suffice for effective management [8]. Evidence from the many studies on viruses not dissimilar from SARS-CoV-2 suggests that a seasonal and environmentally-mediated signal should be seen in the novel COVID-19 epidemic. What do studies to date tell us?We comprehensively reviewed the preprint and peer-reviewed literature on the topic of environmental influences of SARS-CoV-2 transmission. We used the Boolean search capability of Google Scholar to locate articles with the following keywords in the article title: “(COVID-19 OR SARS-CoV-2) AND (pollution OR humidity OR temperature OR UV OR climate OR weather OR season OR seasonality)”. This returned 287 articles on 8 July 2020. On the same day, additional searches for these search terms were conducted in the title fields on PubMed and the title, abstract and subject fields on the WHO COVID-19 literature database (https://search.bvsalud.org/global-literature-on-novel-coronavirus-2019-ncov/), returning 469 and 170 publications, respectively. All searches were constrained to the year 2020. We selected only those studies on infection rates or similar metrics, excluding studies based solely on mortality rates. The combined set, which contained many duplicates and triplicates due to the intersection of three sets of search results, was screened manually and papers suitable for inclusion in our review were retained. Five reviews in preprint were excluded from our assessment, but we did verify that we included in our analysis all relevant papers cited in these reviews. Since we a priori expected many preprint manuscripts, we did not embark on the review with the intention to be PRISMA compliant (as would be necessary for a meta-analysis and systematic reviews), and hence we did not count the number of duplicates and triplicates, the ineligible studies discarded, or the reasons why they were discarded.The result of our searches yielded 42 peer-reviewed publications and 80 preprint manuscripts (Supplementary Tables S1 and S2). The peer-reviewed publications were subject to normal review scrutiny, and form the main body of this section. We did not assess the outcomes of the preprint papers (i.e., they are not discussed in detail as part of Section 5.5) in order to avoid erroneous conclusions based on unassessed data, results or interpretations; nor did we attempt to apply our own peer-review process.The peer-reviewed research conducted on the role of climatic variables in COVID-19 transmission has been highly interdisciplinary, with authors spanning 25 broad academic backgrounds. The largest number of authors (27) currently work in disciplines of geography, earth and environmental sciences, which incorporate climate science. This is closely followed by the 26 authors working in public health, and 25 authors in disciplines of epidemiology, virology and disease control. A total of 40% of the authors are in fields directly relating to COVID-19 and climate. There is, however, a notable spread of authors in more distal academic and medical fields. Notably, the authorship of 18 papers included nobody with an explicitly medical background. Of the multi-authored papers, only three were by researchers who all come from the same disciplinary background, and for two of these, the backgrounds were epidemiology and medical laboratories. Collectively, the peer-reviewed studies provide only weak evidence that SARS-CoV-2 is more infectious under lower temperatures and lower levels of absolute humidity. Similarly ambiguous relationships for air pollution, UV and wind are reported, with a smaller focus on these variables in the literature. There are considerable differences in the ways in which the relationships have been established, resulting from which co-varying variables were included or not; use of different metrics of viral transmission, and which statistical methods were applied. In many cases, insufficient information is provided on the methods and data used, making it impossible to replicate the analyses.This section is relevant because of the high dependence on spatial variance to provide information at this early stage of the pandemic. The geographical coverage of the literature on the environmental influences on SARS-CoV-2 is heavily weighted to the northern hemisphere. Data from Bolivia, Ecuador, Brazil and Australia were included in only five studies, i.e., one-tenth of the total. Most of the southern hemisphere studies are included in studies claiming to be near global in their sampling. Only eight studies focus specifically on a country in the southern hemisphere, Brazil [130,131,132,133,134,135,136,137], and none of them consider any African country.Environmental variables considered in preprint and peer-reviewed publications as modulators of SARS-CoV-2 transmission rates include mean, minimum and/or maximum daily temperature, and diurnal temperature range; an undefined ‘humidity’ variable, relative humidity, specific humidity and absolute humidity; dew point temperature; rainfall; wind speed or wind power; air pressure; some metric of solar or UV radiation; and ‘air quality’ (Supplementary Tables S1 and S2). These choices are apparently strongly influenced by the literature on other viral respiratory diseases.Which definition of ‘humidity,’ is selected is significant challenge for interpreting and comparing studies. Humidity broadly refers to the amount of water vapour held by air (which effects on the viability of pathogens in exhaled aerosol droplets—see Section 4). Studies must account for the fact that atmospheric pressure and temperature modulate the amount of water that a volume of air is able to hold in a gaseous state. A relatively small amount of water vapour is able to saturate cold air, whereas more water vapour is required to bring warm air to saturation. The studies we reviewed that seek to establish whether humidity is a potential driver of COVID-19 use absolute humidity, relative humidity or specific humidity. Two studies use ‘humidity’ [138,139] without qualifying whether it is relative, specific or absolute humidity. This ambiguous use of the term does not permit reproducibility or meta-analysis. Absolute humidity is defined as the total amount of water vapour held by air, in units of g·m−3. A temperature change will not necessarily change the moisture content; it simply changes the capacity of the volume of air to hold water. Only if temperature drops to saturation point, will condensation occur and water vapour content (but not relative humidity) will drop. If temperature increases, water vapour content will only increase if a moisture source is available from where evaporation can take place, or if a moist air mass moves in to replace the drier one. Relative humidity is the fraction of water vapour, expressed as a %, contained by air relative to the amount of water vapour required to result in saturation of air at a given temperature and pressure. Specific humidity is the amount of water vapour per unit mass of dry air (g·g−1). The distinction between relative and absolute humidity matters less in situations when the seasonal thermal range is constrained to a narrow band, such as at some mid-latitude coastal locations and near the tropics. However, in space-for-time studies—such as are required for global syntheses of seasonality effects—the reliance on absolute humidity should allow the investigator to arrive at plausible conclusions about atmospheric water vapour’s effect on viral transmissibility [140,141,142].Environmental data were obtained from various sources such as ERA interim [143] or local meteorological organisations, and maybe provided as daily data or aggregates on temporal scales from 10 days to months. Some use ‘seasonal climatologies’, i.e., averaged long-term data. Since symptoms first manifest 3 to 14 days after infection, analyses sometimes apply lags between the independent and dependent variables of up to 14 [40] or 21 days [41]. Lags have been accommodated in the reviewed literature by applying moving average filters to the daily time series of environmental variables with a width of 7, 14 or 21 days [41]. Another approach is to base the analysis on 10 day aggregates of environmental data [140]. It is uncertain how such discretised intervals can be aligned with case data that is typically daily, but yet contains various delays. Some studies take the mean of the variable over the analytic time period; for example, Jahangiri et al. [144] who ambiguously use either the mean temperature over the study period or over the year, or Liu et al. [141] and Sajadi et al. [12] who use the mean of the environmental variables over the period for which case incidence data were obtained. Most studies, do not account for lag effects [138], or if they do, fail to adequately explain how lags were accommodated [40].Which metric of SARS-CoV-2 transmission to use as dependent variable is critical in addressing the central question, “do environmental variables modulate the transmission of the virus?” We argue in Section 3.3 that the Basic Reproductive Number, R0, is the best parameter for this purpose since it excludes the effects spontaneous or imposed non-pharmacological control measures implemented to slow the spread of the disease, but which still incorporates the environmental influence of a particular place. The failure to adequately account for non-entrée influences is the Achilles’ heel of many of the studies reviewed. Of the literature we assessed (Supplementary Tables S1 and S2), only six studies base their assessment of the presence or magnitude of environmental influences on R0 as the dependent variable [39,145,146,147,148,149]. Jebril [150], Luo et al. [151], Poirier et al. [142], and Wang et al. [152] used Rt (see Section 3.3) as the response variable. Because Rt is very context specific and sensitive to social factors and interventions, using this parameter to assess the presence and size of environmental influences will in most instances have a low signal: noise ratio. The usefulness of Rt is that it demonstrates how effective NPI measures are in controlling an epidemic, and provides information on how regulators must adapt these interventions over time, based on health and economic goals. The non-environmental ‘noise’ can be filtered out, but this requires a great deal of data regarding the nature of the specific interventions applied, movement patterns, precise knowledge about testing and reporting (which is not necessarily constant), and so forth. None of the Rt based studies to date meet these preconditions, and are therefore not able to remove the non-climatic (social) influences from the rapidly fluctuating Rt values.Another approach that holds merit is to use the growth rate or doubling time estimated from the exponential increase in cases as dependent variable [148,153,154,155,156,157,158,159,160]. Merow and Urban [156] argue that these kinds of metric are robust even if the details of testing and reporting vary from place to place, as long as the detection probabilities at a place remain constant over the estimation period. This argument is equally valid for estimates of R0.Another variation to this theme of estimating growth rate-related parameters as an indication of transmissibility is to take rates as time required to progress from the first reported case to 200 cases [161], or to use the cumulative number of cases reached 28 days after the first reported case [162]. However, these approaches effectively fit a linear model to case vs. time data, which does not account for the accelerating rate of increase in number of cases. Lolli et al. [163] use the daily ICU case anomaly, but this of course entirely excludes all but the most severely ill patients and cannot be seen as being representative of disease transmissibility.Other data-related considerations, particularly in relation to studies that use parameter estimates of the exponential relationship that daily new infections has with time, are that care must be taken to omit both (i) cases that result from the importation of infected individuals from the time series (i.e., new cases must be local transmissions only), and (ii) the case data obtained after the intervention period begins. Requirement (i) can be affected by including only the portion of the time series after a certain minimum number of cases are present, as has been performed by Caspi et al. [154], Merow and Urban [156], and Notari [157]. Requirement (ii) is met by Ficetola and Rubolini [155], Merow and Urban [156], Notari [157], and possibly for Oliveiros et al. [158], although it is uncertain how strictly this was implemented due to their statement that Oliveiros et al. [158] “considered mainly the initial days of the time series” (p. 4). We will comment on the reproducibility of methods in Section 6. Requirement (ii) is implicit in the definition of R0, but the two requirements constrain the usable data to between ‘not too early’ and ‘not too late’.The bulk of the studies in Supplementary Tables S1 and S2 used daily new or cumulative confirmed cases as response variables. This practice is not advised for largely the same reasons given for Rt. Such daily data are likely to carry too many other non-climatic signals to be generally used—unless, of course, analyses included a specific set of controls that would be difficult to extend to the global context.The studies in Supplementary Tables S1 and S2 employed the following statistical methods to evaluate relationships between environmental variables and the transmission rate of SARS-CoV-2: various linear, logistic, or exponential parametric models [39,41,50,130,134,142,146,148,151,158,164,165,166,167], sometimes with the inclusion of non-Gaussian error structures as permitted by Generalised Linear Models (GLMs) [141,157,162,164,168,169,170,171]; Generalised Additive Models (GAMs) [41,134,167,172,173]; distributed lag panel regression models [153]; machine learning such as support vector machines and decision trees [147]; local panel projection estimator within a country-level dynamic framework [174]; Loess smoothers/curves [142]; Bayesian methods [157]; and Pearson’s, Spearman’s, and Kendall’s correlations [40,138,139,140,154,175,176,177].Regression approaches allow functional relationships to be established between the driver (any of the environmental influences) and response variable (a metric of infection rate), allowing the magnitude of the environmental effect can be determined. Robust implementation of a regression approach would include place as a random effect (i.e., as mixed models, also known as panel regressions; for example, [11,153,155,178,179]). This allows the fact that the effect of the environment on viral transmission varies from place to place, for social and historical reasons. Multiple regression allows the simultaneous evaluation of several predictor variables in terms of the influence they collectively or individually have on the outcome [39,180,181]. It is possible to establish which of the drivers, if any, has the greatest contribution to an effect seen in the outcome variable. For example, Mollalo et al. [180] used multiple regression to evaluate the simultaneous contributions of environmental and socio-economic influences on USA county case counts. If parameterised properly, multiple regression can be used to rule out contributions of potentially confounding and multi-collinear variables.Loess smoothers and correlation approaches, although useful for a qualitative assessment for the presence of environmental influences, cannot inform us about the relative importance of environmental modulators versus other location-specific or social influences. Similar non-quantitative approaches that only hint at the presence of relationships include the simple visual mapping of the number of infections in relation to climate zones or latitudes [12,150,182,183,184]. These methods can at best raise an hypothesis that requires further testing.Other approaches worth mentioning include the application of wavelet transforms [185], multivariate analyses [130], and ecological niche models [186,187]. Wavelet analysis, which requires a long time series, provides only a qualitative view of disease dynamics as modulated by weather or climate variables. Ecological niche models are not suited for studies on COVID-19 because disease dynamics are entirely different mechanistically from the principles that govern organisms and ecological systems (as reviewed by Carlson et al. [188]). Multivariate methods are useful for examining environmental variable modulation of COVID-19, since they provide many, if not all, of the benefits of multiple regressions, plus they have other features that confer flexibility and the ability to accommodate a range of data types. They are ideally suited for situations where there are many factors that might contribute simultaneously to the variation of one or many outcome variables. The application of a multivariate approach by Auler et al. [130] uses data on the daily new confirmed cases (see critique above), and for this reason we do not consider the findings of this study further in our review.Dynamic or mechanistic models (predominantly the compartment models of the SEIR family) are useful tools to explore how seasonality may impact on the evolution of the disease, and provide a way to discern the signature of seasonality in near real-time observational data. Such an investigation recently reported on by Baker et al. [189] concluded that under the high infection rates of COVID-19, within the context of almost the entire population being susceptible at the onset of the disease, seasonality effects on the disease evolution will be limited initially. However it cannot be discounted at later stages, if for instance, the immunity gained by recovered patients is temporary, so that they become susceptible again in subsequent years or if herd immunity is not attained before managed abatement of the epidemic (as we are seeing in some countries experiencing resurgences). A similar study by Neher et al. [51] came to similar conclusions.We will now discuss only the findings of those studies that have undergone peer review, have selected appropriate environmental data as influential variables, relied on suitable response variables (such as R0 or parametric estimates) to estimate the local viral transmission rates in the absence of policy control measures, accounted for potential confounding influences, and applied appropriate statistical models.The only peer-reviewed paper that fulfils all of these criteria is that by Yao et al. [39], which undertakes an assessment of the effects that temperature, relative humidity, and UV radiation have on the R0. This study has a relatively narrow geographical focus: it includes 227 Chinese cities. R0 was calculated from data over the period 10 February to 9 March 2020. The authors assert that these data are for the “expected number of secondary cases generated by an initial infectious individual, in a completely susceptible population” [39] (p. 1). All daily environmental data were spatially matched as closely as possible to the cities they represent. Given the large number of cities, each with its unique climate, this kind of study lends itself to a regression-type analysis if each of the daily observations per environmental variable are averaged over the study period duration before relating them to each locality’s R0. This study did not find an influence due to any of the environmental variables studied on the rate of SARS-CoV-2 transmission. A weakness of the study was the failure to account in their multiple regression model for any of a large number of city-level confounding influences.A single published study does not provide robust support for the presence or absence of a climatic influence on SARS-CoV-2 transmission rates. The preprint studies [11,147,152,153,155,156,162,174,190] offer mixed statistical support (none, weak, or strong relationships) for the influence of environmental drivers. Carlton et al. [153] show that that UV radiation affects COVID-19 growth rates, but not temperature or humidity. Merow and Urban [156] offer comparable support for a UV radiation effect. According to Ficetola and Rubolini [155] and Wan et al. [190], COVID-19 transmission is greatest at a temperature of 5 and 6.3 °C, respectively; the former authors further show that transmission peaks at a specific humidity, ~4–6 g·m−3 (peaking implying optimum conditions above and below which transmission rates drop off). Similarly, Leung et al. [162] suggest support for the hypothesis that lower temperature and humidity enhance COVID-19 transmission. Similar responses are seen by Lin et al. [165] and Wilson [174] with regards to temperature, but they also suggest an interaction between temperature and relative humidity [165] and temperature and mobility [174] in terms of modulating infection rates. In contrast, Gupta and Gharehgozli [147] show that higher temperatures enhance the spread of the disease; they also show that viral transmission is enhanced under higher concentrations of PM2.5.This pandemic has rapidly mobilised scientists from diverse disciplines in a possibly unprecedented way. Scientists have helpfully offered insights and analytical methods based on their own disciplines They did so efficiently and swiftly, particularly in those countries most heavily affected by the pandemic early on. The rush to contribute knowledge about the future spread of COVID-19 resulted in a flood of papers appearing on preprint servers [191], which will in due course be peer reviewed and some will be published. The pressure to speed up the peer-review process, in order to address the urgent challenge, may result in a compromise in the quality of both the review process and the science that is thereby published. In our screening process in Section 5, we scrutinised 29 peer-reviewed publications and 23 preprint articles. Of these, we found one published and potentially four preprint studies that offer credible insight into the climate-related SARS-CoV-2 and COVID-19 dynamics and epidemiology with a reasonable degree of confidence and rigour.The general prevalence of climatologically-coupled seasonal signals and environmental variable modulation seen in the majority of other viral respiratory diseases creates the expectation for a similar effect on SARS-CoV-2 and in COVID-19 epidemiology. However, this virus and disease have only been spreading for 8 months. Observational evidence available to date has not yet been analysed sufficiently thoroughly to show that climate-related modulation is indeed a significant factor. The studies reviewed in Section 5 have aimed to find signs for such a signal, but a variety of methodological problems render a definitive conclusion premature.The currently available time series do not capture a full annual cycle at any one location, or globally. The first studies appeared in late January on preprint servers (the majority of these are yet to be formally published as of mid-July 2020). As such, the initial reports looked for spatial variation in infectivity within a region and attempt to explain it in terms of associated variability in temperature, humidity or other environmental factors among these locations. Later studies could have benefitted from the larger datasets and a wider range of variation in the environmental drivers, resulting from the global spread, but became increasingly confounded by co-varying differences among the countries’ socio-economic conditions and pandemic responses. To date, the ‘global’ messages coming from the current body of COVID-19 research in general, and in respect to the environmental drivers of the disease in particular, do not equitably address the specific dynamics and considerations pertaining to the ‘Global South’. This is in part likely due to the slightly later arrival of the disease in the southern hemisphere. Thus, fewer southern hemisphere countries have suffered outbreaks of the same scale and severity (at the stage of assembling this manuscript) as the epidemics in the Far East, Europe and the United States. At the time of writing, the situation in some South American countries (such as Brazil and Peru) was deteriorating quickly. There is also a technical challenge in countries with relatively lower medical health research capacity, such as those in Africa [192]. The upshot is a circumstantially driven bias in the current literature which needs to be corrected, for several reasons. Neglecting the hemispheric disparities in knowledge regarding the role of environmental variables on SARS-CoV-2 and the modulation of the COVID-19 epidemic influences the discussion on the attribution of the reductions in cases. Northern countries are likely to move past peak daily infections coincidentally with the height of summer. It also neglects the urgent consideration of countries which are moving into winter. Importantly, many of the countries in the global south have already-stressed healthcare systems, and accurate modelling is critical in determining policy interventions for control measures to protect the lives of some of the world’s most vulnerable people. The collective global experience can provide a shortcut to knowledge and information regarding the role of environmental variables on SARS-CoV-2 biology and modulation of COVID-19 epidemiology and seasonality, applicable anywhere, by exploiting the latitudinal phasing of seasons to conduct research in all climates zones simultaneously. This leads us to call for global collaboration on this topic.Much of the work we reviewed failed to carefully consider the implications of the choice of available metrics for viral transmission. We deem R0 to be best suited for the purpose of finding environmental sensitivity and seasonal climatic signals; some parametric estimates from regression models can also work, provided that care is taken to constrain the cases to those that result from local transmissions up to the time when NPIs come into play. R0 is closely aligned with the SIR-SEIR model family, and can be derived from the inversion of time series of case rate data using these models (see below).Due to the effects of the incubation period, it may be important to use daily data (rather than data averaged over a several days) and a suitable lag period for both environmental and test-result data incorporated in the analysis. In the case of a highly infectious disease such as COVID-19, manifesting in a densely populated location, the effect of daily weather variations on transmission mechanisms is likely to be overwhelmed by the sheer magnitude of exposure. It may be that environmental modulation is still an important factor in these circumstances, but may reflect in indoor environments rather than outdoor ambient conditions [193]. Once the disease spread begins to approach an equilibrium (Rt~1), the environmental effect may become more apparent.To date, studies that attempted to discern the effects of climate by comparing infection rates across regions with different climates have been compromised by the heterogeneities that exist across locations and times in terms of control measures applied [194], and social, economic and cultural conditions that affect the practise of social distancing. Most studies have omitted variables such as poverty, population size and demographics (particularly age frequencies of the populace), the density of the population and how much high-resolution clustering is present (such as in the informal settlements in many countries of the South), the degree of urbanisation, access to healthcare, mobility and migration, various types of comorbidities (e.g., TB, HIV, malnourishment), the effect of the Bacillus Calmette-Guérin(BCG) vaccine [195], and a plethora of additional influences which are still not well understood with regards to how they influence the unfolding of COVID-19 across the globe. Simple graphing of case numbers across time in relation to some of the potentially influential drivers (as for example permitted by the Our World in Data Coronavirus Pandemic Data Explorer) will help reveal which of the additional variables to admit into the analysis.An important obstacle to finding the seasonal signal in the global COVID-19 data is to find a way to deal with the hemispheric disparity (gradient away from the equator) in out-of-phase climatic signals. Comparing the evolution of COVID-19 for northern hemisphere countries moving from winter to summer to its evolution in southern hemisphere countries moving from summer to winter provides a valuable opportunity to discern the signature of seasonality. However, such a comparison will remain compromised by short time series and can only fully fulfil its potential once both hemispheres have experienced a full annual seasonal cycle.We have concluded that due to high values of R0 exhibited by SARS-CoV-2, seasonal climate modulation should not be relied on to significantly dampen the infection rate even in the midst of the northern hemisphere approaching summer. Should the disease persist several years into the future, however, under the condition of an increasing fraction of the population of a given region having immunity, it is likely that the COVID-19 will exhibit an increasingly clear seasonal cycle as evident in similar endemic human coronaviruses. Such insights will only be apparent after the main pandemic surge in 2020.We suggest some avenues for progress in addressing the environmental sensitivity of the disease. In addition to regression and correlative empirical approaches (Section 5.4), non-linear methods can also be applied. These may include the use of extended Kalman filters and the inversion of compartment models. Extended Kalman filters are commonly used in data assimilation to infer parameters from high-dimensional input data sets. Recently, Pei et al. [196] applied an ensemble-adjusted Kalman filter to infer the differential spatial distribution of COVID-19 infection rates from empirical data collected across different counties in the USA, followed by their application in a SEIR model. It may be feasible to apply this technique to estimate the relative roles of non-pharmaceutical control measures and seasonality in determining the infection rate. Inverse modelling, particularly using SEIR-type models, can infer infection rates from case and testing data, as demonstrated for the Hubai Province in China [46]. Making use of large ensembles that ingest data from many locations and systematically explore various combinations of the forcings can potentially explore the relative sensitivities of infection rates to NPI control measures and seasonality.We recommend the use of regression-type statistical analyses than can be adapted to accommodate many simultaneous driving variables, including both environmental and non-environmental factors, thereby removing confounding influences. These models also readily accept non-Gaussian error terms and can account for autocorrelation in time series. Lags between exposure and when an individual is confirmed as infected can be accommodated by distributed lag non-linear models [197,198]. These techniques rely on Generalised Additive Models (GAMs) for the flexible estimation of smooth responses and parametric terms. The recognition that disease dynamics may differ between locations for a multitude of reasons requires that ‘location’ be specified as random effect (notable examples involving COVID-19 include Carlton et al. [153] and Wilson [174]). Such approaches can be accommodated by longitudinal models (called panel regressions by economists) (sensu Gardiner et al. [199]), which regress the dependant variable (plus covariates and constraints) as a function of time. Care should be given to estimations of uncertainties around model predictions —such estimates of uncertainties are permitted by Markov Chain Monte Carlo (MCMC) approaches [42]. Knowing the uncertainties is necessary in assessing projections from competing models in the public policy space. Finally, multivariate approaches, such as Redundancy Analysis (RDA) or Constrained Correspondence Analysis (CCA), will also accept a creative assignment of a host of response and influential variables simultaneously, and can be employed when research is faced with many potentially contributing factors, each of which might explain a portion of the overall variability.We noted a lamentable deficiency in the application of reproducible research practices in many of the publications we reviewed. Clear, precise reporting of data sources and quality, data screening practices, listings of the ancillary data sources used, a detailed account of the data processing and statistical procedures and software used, and the exact reporting of all relevant diagnostic and supporting statistics, tables and figures is essential, particularly in this global emergency, where published data and information are used operationally, and where robust guidance is most likely to emerge from meta-analyses of many studies. Lives, livelihoods, economies, and the public trust in science depend on rigour and reproducibility. It is thus incumbent upon global research organisations and agencies such as the World Health Organisation (WHO) and the World Meteorological Organisation (WMO) to provide leadership and guidance and to define best-practice protocols for the analysis of data and production of information. To this end, the WHO has produced a document entitled “A Coordinated Global Research Roadmap: 2019 Novel Coronavirus” [200]. Its scope is broad, and thus does not specifically address some of the issues raised in our review. The authors are aware [201] that at the time of writing, the WMO has agreed to set up a Task Team which will focus on the environmental aspects of the COVID-19 pandemic.Datasets capturing even the first full seasonal cycle of COVID-19 incidence in one locality, region or globally are not yet available and it is not possible at this stage to conclude that a definitive and unequivocal signal of environmental modulation is apparent from the reviewed literature. However, there is some evidence that environmental drivers played a role in transmission in some regions and at some (early) stages of the pandemic. Under other circumstances, longer and denser datasets would be a minimum requirement to support a thorough statistical treatment to explore evidence of environmental modulation of the COVID-19 pandemic and epidemiological dynamics. Pressure for rapid answers and information has prompted impulsive and dubious forays into signal-finding missions, such as those that dominate the current body of literature that had accumulated to date (15 July 2020). Analyses based on space-for-time substitutions have been inconclusive, primarily due to lack of care taken to account for the effects of strong confounding variables, such as socio-economic influences and effects of NPIs, which exist between jurisdictions. In terms of the outcomes of the published work, most studies are insensitive to the idiosyncratic conditions unique to many Southern Hemisphere countries, rendering it challenging to transfer findings from north to south. Rigorous hypotheses, interrogation of assumptions, and careful selection and development of analytical approaches and statistical models are required to examine environmental signals in complex COVID-19 incidence datasets, especially prior to longer and denser time series data being available. In the interim, there is merit in comparisons of signals among contrasting locations at different scales, and with due consideration paid to the implementation of NPIs and other sources of ‘noise’. This outcome does not discount the role of environmental drivers in modulating the incidence or seasonality of person-to-person transfection mechanisms, or of the morbidity, severity and mortality associated with COVID-19 infections. However, these may become unequivocally discernable only at later stages of the pandemic in 2020 or 2021, and globally coordinated efforts to test this robustly are essential. The following are available online at https://www.mdpi.com/1660-4601/17/16/5634/s1, Figure S1: Discipline backgrounds of authors whose publications were included in this review. Table S1: Studies that have aimed to establish links between SARS-CoV-2 infections and environmental variables, notably temperature and humidity. All studies have in common a finding that the transmission of the virus is enhanced under colder, dryer conditions. Table S2: Studies that have aimed to establish links between SARS-CoV-2 infections and environmental variables, notably temperature and humidity. All studies have in common a finding that the transmission of the virus is enhanced under colder, dryer conditions.The conceptualisation of this study was by A.J.S.; he also performed the data extraction and wrote Section 5 “Critical assessment of studies of COVID-19 climate susceptibility” and Section 6 “Discussion”. Data collection was undertaken by N.A.S., J.M.F., and A.J.S.; J.M.F. contributed Section 2 “Why the southern hemisphere is different”. F.A.E. provided editorial input as required, and contributed his thinking around inverse modelling and some of the climatological considerations. R.J.S. provided the text under Section 3 “Monitoring and modelling the spread of COVID-19”, and also assisted with thorough language editing of the final document. N.S. added Section 1 “Introduction” and Section 4 “Implications for COVID-19 of environmental sensitivity in other viral respiratory diseases”, with G.D. contributing towards the latter section. Funding for the COVID-19 Environmental Reference Group (CERG) endeavour was enabled by N.A.S., and he also provided overall project management for the group’s efforts. 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, but the time of some of the authors of this paper was attributable to funding from the National Research Foundation and our respective home institutions.The authors would like to thank the South African Department of Science and Innovation for encouraging us to proceed with this work, and other members of the COVID-19 Environmental Reference Group (CERG) for providing valuable editorial input and discussion around the development of this review paper.The authors declare no conflict of interest.The number of confirmed COVID-19 cases as of 12 July 2020. Data are shown as the number of cases per 100,000 individuals. COVID-19 case data are from Johns Hopkins University Center for Systems Science and Engineering. The world population data are from the World Bank.Environmental factors that have been suggested to influence a COVID-19-like disease, overlain on the structure of a generic SEIR-type compartment model to show the potential mechanisms of action.A summary of modelling approaches applied to COVID-19.Viability of the human coronavirus, HCoV 229E, as a function of time, temperature and humidity [93].n.d. = not detectable.

Med-MDPI/ijerph_5/ijerph-17-16-05635.txt

@@ -0,0 +1,8 @@
+Humans are living in an uncertain world, with daily risks confronting them from various low to high hazard events, and the COVID-19 pandemic has created its own set of unique risks. Not only has it caused a significant number of fatalities, but in combination with other hazard sources, it may pose a considerably higher multi-risk. In this paper, three hazardous events are studied through the lens of a concurring pandemic. Several low-probability high-risk scenarios are developed by the combination of a pandemic situation with a natural hazard (e.g., earthquakes or floods) or a complex emergency situation (e.g., mass protests or military movements). The hybrid impacts of these multi-hazard situations are then qualitatively studied on the healthcare systems, and their functionality loss. The paper also discusses the impact of pandemic’s (long-term) temporal effects on the type and recovery duration from these adverse events. Finally, the concept of escape from a hazard, evacuation, sheltering and their potential conflict during a pandemic and a natural hazard is briefly reviewed. The findings show the cascading effects of these multi-hazard scenarios, which are unseen nearly in all risk legislation. This paper is an attempt to urge funding agencies to provide additional grants for multi-hazard risk research.Human existence involves exposure to many hazards [1], and various low to high risk scenarios. While understanding a hazard and its associated risks may help prevent or reduce adverse consequences, in many instances, people are unaware of the risks involved, making it difficult to fight against an invisible enemy. Risk aversion is a robust characteristic of human decision making, meaning people are less likely to gamble on something when they are unsure if they will obtain the desired outcome. However, planning for risks becomes even more challenging when considering that we live in a multi-risk world with infinite natural and man-made hazards, many of which we cannot control and can occur at any time.The year 2020 will be remembered in the U.S. for several reasons: (1) The coronavirus began to spread throughout the nation starting in February, with the first confirmed case reported on 21 January, according to CDC (https://www.cdc.gov/mmwr/volumes/69/wr/mm6918e2.htm). (2) There were multiple natural hazards (NH) around the country, including two devastating dam failures in Michigan and over 500 earthquakes in western Nevada (https://www.sfgate.com/earthquakes/article/Nevada-Tonopah-earthquakes-6-5-aftershocks-15283968.php), and an above-normal Atlantic hurricane season expected (https://www.cnn.com/2020/05/11/us/2020-atlantic-hurricane-season-fast-facts/index.html). (3) Complex emergency (CE) situations arose at both national (e.g., Black Lives Matter protests) and international (e.g., U.S.–China and U.S.–WHO tensions) levels. Other notable conditions included the ongoing climate crisis and economic contractions due to nation-wide stay-at-home orders. For example, GDP fell 4.8% in Q1 2020, and unemployment increased by more than 10% between April and May. Figure 1 illustrates the combination of multi-hazard factors that many citizens are facing.This paper tries to briefly review three distinct risk sources that the U.S. is facing, Section 2; followed by the concept of multi-risk analysis in Section 3. A perspective on low- to high-probability risk scenarios for healthcare system impacts, functionality loss, and recovery duration is provided in Section 4. Finally, the concept of NH-induced evacuation and sheltering during pandemic conditions is revisited in Section 5.In December 2019, the novel coronavirus pandemic, known as COVID-19, emerged from Wuhan, China [2]. It is the most recent biological hazard and has resulted in a global outbreak. COVID-19 is a potential zoonotic disease with a low to moderate mortality rate. Person-to-person transmission may occur through droplets or direct contact [3], and therefore isolation of cases and contact tracing are essential to controlling COVID-19 outbreaks, though the probability of successfully controlling an outbreak decreases as the number of initial cases increases [4].As of 4 August 2020, there are about 4.8 million confirmed cases, 2.3 million recovered patients, and about 160,000 deaths related to COVID-19 in the U.S. alone [5]. Multiple researchers have studied the risks associated with a COVID-19 outbreak [6,7]. Although most of these studies are in preliminary stages, different types of forecasting models have been proposed to predict the temporal and spatial distributions of the virus when subjected to various constraints [8,9,10]. These projection models account for factors such as the behaviors of citizens, impacts of social distancing, effectiveness of face coverings, consequences of reopening, capacity of the healthcare system, pre-existing health conditions and age groups [11,12]. It is noteworthy that there is an epistemic uncertainty (lack of current knowledge) [13] in the exact number of infections and their spatial distribution. Therefore, making any decision about the reopening of the states/cities is very difficult and challenging. This already caused extra political problems and legal challenges. According to CNN (https://www.cnn.com/2020/07/16/politics/georgia-kemp-mask-mandate/index.html), Georgia governor, Brian Kemp, announced that he is suing Atlanta Mayor, Keisha Lance Bottoms, over the city’s mask mandate, claiming the measure violates his emergency orders.The risk of exposure to COVID-19 is an important factor in subsequent life loss (LL) estimations. Among other factors, it is highly dependent on location, human concentration, and safety protocols. An approximate COVID-19 risk map is shown in Figure 2a (as of 18 July 2020). The map is based on incidence rate (i.e., the number of confirmed cases per 100,000 people), and since it accounts for population density, it offers a more reliable metric for exposure risk. The map is subjected to temporal changes.Natural hazards are the result of a series of natural processes that have operated throughout earth’s history [14]. Hazard analysis refers to a process of recognizing hazards that may arise from a system or its environment, documenting their unwanted consequences, and analyzing their potential causes [15]. Natural hazards are classified as geophysical (e.g., earthquake, volcanic activity), meteorological (e.g., tornado), hydrological (e.g., flood), climatological (e.g., drought), biological (e.g., epidemic), and extraterrestrial (e.g., impact).Since the start of the COVID-19 pandemic in January 2020, several natural disasters have been reported, including: (1) dozens of tornados in southern states in the U.S. between 12 and 13 April (36 fatalities); (2) A 6.5 magnitude earthquake struck the western area of Nevada on 15 May, damaging the main highway; (3) two dam breaks on 19 May in Michigan, U.S. (with 11,000 evacuees); (4) tropical Storm Cristobal made landfall on 7 June, near the mouth of the Mississippi River and the island of Grand Isle in Louisiana, brought winds of up to 85 km/h, and spawned a tornado in Florida; (5) seven inches of rain caused flash floods in Wisconsin, washed out roads, and declared a state of emergency on 29 June; and (6) as of late July, Hurricane Hanna has roared ashore onto the Texas Gulf Coast as a Category 1 storm.In the U.S., FEMA provides the major disaster declarations every year [https://www.fema.gov/disasters/year/2020?field_dv2_declaration_type_value=All]. Figure 2b is one of many maps showing the approximate locations of various natural hazards across the country. The majority of risk maps for natural disasters are developed based on their economic impact to properties (e.g., loss).While multiple complex emergencies have developed since the start of the pandemic (e.g., U.S.-China tensions, U.S. military movements in the Middle East), only those directly affected by the spreading of COVID-19 inside U.S. borders are discussed in this paper. These can be categorized into two main groups: (1) normal condition actions and (2) emergency condition responses. Both categories are somewhat related to mass gatherings, which pose significant public health challenges to health care professionals and governments [16]. Historically, sporting, religious, music, and other mass gatherings have enabled the global spread of infectious diseases [17]; the situation can become worse when face coverings, social distancing, and other preventative actions are not fully observed by attendees. Authorities in each community must try to flatten the transmission curve to give scientists more time to find a cure; however, mass gatherings move the needle in the opposite direction.Multiple researchers have shown that the perceived risk of COVID-19 is affected by politically-motivated interpretations of the risk. These patterns persist even in the face of state-level mandates to close schools and non-essential businesses [18]. Studies show that political partisanship may play a role in determining perceived risk during a pandemic, with potentially significant changes in public health outcomes. According to Painter and Qiu [19], residents in Republican counties are less likely to comply with stay-at-home orders than those in Democratic counties. Similarly, Democratic-leaning counties responded more to recommendations from Republican governors than from Democratic ones [20]. According to Adolph et al. [21], the results of the state-level database analysis for five social distancing policies across all fifty states revealed that: all else equal, Republican governors and governors from states with more President Trump supporters were slower to adopt social distancing policies. Furthermore, it is reported that U.S. counties with lower per capita income were associated with significantly reduced social distancing mandates [22]. Thus, the geographical location of a pandemic is an important factor in its spread. According to Dincer and Gillanders [23], in communities where corruption is endemic, observing social distancing during sheltering and implementation of mitigation strategies is difficult. We should also highlight the mutual trust between individuals and their communities [24], which yields a successful emergency mission during a pandemic outbreak.The 2020 U.S. presidential election rallies are also a hot topic amongst voters. While many believe the rallies should stop, President Trump did not cancel his June campaign in Tulsa, Oklahoma. According to CNN (https://www.cnn.com/2020/06/25/politics/trump-tulsa-rally-coronavirus/index.html), following this rally (20 June), at least eight staffers who were part of the rally preparation tested positive, and the rest who attended the rally were quarantined. One may note that the incubation period for COVID-19 is about 14 days, but the CDC (https://www.cdc.gov/coronavirus/2019-ncov/hcp/clinical-guidance-management-patients.html) announced the median time of 4–5 days (other resources reported similar data [25]). To date, there is no scientific research showing the direct correlation between rallies and transmission of COVID-19. However, according to Fox News (https://www.foxnews.com/politics/trump-campaign-says-two-more-staffers-who-attended-tulsa-rally-tested-positive-for-coronavirus), at the same time as the rally, Tulsa County was experiencing its own spike in infections, which drew concerns that Trump’s indoor rally could be a “super spreader” event for the virus.One form of mass gathering during the pandemic has been protesting or marching by different civil rights groups, and the U.S. has seen several such protests over the past three months. On 24 April, nearly 1500 people gathered at the Wisconsin State Capitol in Madison to protest. Two weeks later, the Wisconsin Department of Health Services confirmed 1986 cases of COVID-19. Of those, 72 people reported having attended a large gathering, though patients were not asked specifically if they had attended the protest [26]. It was observed that many protesters did not maintain a six-foot distance from others or wear masks. On 15 April, there was a protest in Lansing, Michigan against the state’s governor, Gretchen Whitmer, and her COVID-19 lockdown, Figure 3a. Again, there was no sign that people were taking the COVID-19 related public health advice seriously.Following George Floyd’s death on June 1 by a white police officer in Minnesota, a new wave of Black Lives Matter protests began in the U.S. While many worried about catching COVID-19 during these marches [27], they decided to take the risk anyway, Figure 3b. While some of the protesters tried to follow public health advice (i.e., wearing masks, distancing, using hand sanitizer, and getting tested for COVID-19), but there is no perfectly safe way to demonstrate in large groups during a pandemic. Concerns about racism and discrimination have also arisen during the COVID-19 outbreak [28]. COVID-19 policy responses have disproportionately affected people of color and immigrants-people who are over-represented in lower socioeconomic groups, have limited access to healthcare, and work in precarious jobs [28]. The question for many was: “Which is worse: protesting with an increased short-term risk of fatalities due to COVID-19 or staying at home and enduring sustained systematic racism?”For many black people and their allies, the risks associated with protesting did not outweigh the risks of doing nothing, which some equated with “one in every 1000 black men dying at the hands of police”. According to The Guardian (https://www.theguardian.com/us-news/2020/jun/03/protests-police-covid-19-coronavirus-spread), it is impossible to know how many people at these marches were asymptomatic carriers, and that is really scary. Protests, like those mentioned above, are now taking place nation-wide; Figure 2c maps the locations of protests in the U.S. related to George Floyd’s death and the larger Black Lives Matter movement.Finally, the scenario was worsened by police tactics used to subdue protesters, Figure 3c. According to Wired (https://www.wired.com/story/police-tactics-could-turn-protests-into-covid-19-hot-spots/), some police tactics could turn protests into COVID-19 hot spots. While large crowds already carry a risk of transmission, the situation is exacerbated when police deploy tear gas against protesters, causing them to cough on each other (spraying virus-laden droplets into the environment), or bus them to jails in groups. Tear gas and pepper spray make it nearly impossible to breathe while wearing a mask, and mass arrests or detainments are very risky, not just for the people arrested but also for the jail staff, the court staff, and their families. In any case, the police have to respond to vandalism and theft (in any form) in which people try to damage public properties, steal from stores, and alter the peaceful protest. Also, the police should confront anarchist agitators, and criminal opportunism amid the chaos. It is important to note, therefore, that both protesters and police face significant risks. According to the Military (https://www.military.com/daily-news/2020/06/10/national-guard-covid-19-diagnoses-after-protests-are-disturbing-sign-fauci-says.html), an undisclosed number of the roughly 1200 D.C. Guard members sent to respond to the protests now have COVID-19. Two members of the Nebraska National Guard also tested positive.The readers should note that all the statements (especially the political ones) throughout this paper do not reflect the personal political view point of the author in any way. All the statements are carefully selected from peer-reviewed documents and those reported by the news. None of the statements aim to defend a political party and/or opinion, but to bring to the attention of readers, the complex situation we are living.Finally, all the above-mentioned hazard sources (and their combined effects) should be studied in the context of the healthcare system performance, See Section 4. A better healthcare system may reduce the devastating consequences of large-scale fatalities. According to KHN (https://khn.org/news/as-coronavirus-spreads-widely-millions-of-older-americans-live-in-counties-with-no-icu-beds/), more than half of counties in the U.S. have no hospital ICU beds, which poses a particular danger for more than seven million 60+ years old people facing the spread of COVID-19. They released a map (based on 2018 and 2019 reported data), showing the counties with and without hospitals, and counties that do have ICU beds, Figure 2d. As can be seen, there is considerable heterogeneity in the distribution of the ICU beds (with some having just one bed available for thousands of senior residents). One may normalize the above-discussed multi-hazard sources with respect to the available healthcare system in each county. In this way, the impact of the healthcare system (as a secondary hazard source, in case it is not sufficient) is incorporated in overall risk calculation.Each of the individual risk factors explained in Section 2 can be catastrophic and devastating if the individuals and/or society/community are not prepared already [29]. However, the critical question is “Are we ready for combination of these risks?”. Therefore, we need to talk about the framework of multi-risk analysis, keeping in mind the fundamental differences between hazard and risk:Risk ∼ Hazard × Impact on asset × Consequences of impact; Ref.[30](1)
+		Risk ∼ Hazard × Value at risk × Vulnerability; Ref.[31](2)The concept of multi-risk analysis is well established in natural hazards [30,31]. European Commission [32] defines the multi-hazard assessment as:“to determine the probability of occurrence of different hazards either occurring at the same time or shortly following each other, because they are dependent from one another or because they are caused by the same triggering event or hazard, or merely threatening the same elements at risk without chronological coincidence.”Therefore, a multi-hazard assessment can be studied from two perspectives [33]:Independent hazards threatening a given area: the main concern in this effort is harmonization of the hazard assessment, meaning that all should have a similar basis [31].Hazard interactions, triggering or cascade effects: in this effort, the occurrence of one event could affect the probability of occurrence in others (usually accelerates them). Once this concept is propagated into a chain of events, the Bayesian networks framework can be used [34]. For two events with occurrence of e1 and e2, the probability of e1 occurrence, P[e1], is:
+          (3)P[e1]=P[e1|e2].P[e2]+P[e1|e¯2].P[e¯2]
+          where the bar sign presents the non-occurrence condition. This equation can be generalized for N events.Independent hazards threatening a given area: the main concern in this effort is harmonization of the hazard assessment, meaning that all should have a similar basis [31].Hazard interactions, triggering or cascade effects: in this effort, the occurrence of one event could affect the probability of occurrence in others (usually accelerates them). Once this concept is propagated into a chain of events, the Bayesian networks framework can be used [34]. For two events with occurrence of e1 and e2, the probability of e1 occurrence, P[e1], is:
+          (3)P[e1]=P[e1|e2].P[e2]+P[e1|e¯2].P[e¯2]
+          where the bar sign presents the non-occurrence condition. This equation can be generalized for N events.Liu et al. [34] proposed a simple matrix approach to identifying the interactions between various hazards. According to Figure 4a (upper part), the interaction of any two hazards Ei and Ej can be determined by understanding their impacts on one another. For three hazard sources in this paper, such interactions are assessed in Figure 4a (lower part). While the single-hazard sources take the diagonal cells, their clockwise influence/interaction fill out the off-diagonal cells. For example, this matrix shows that occurrence of a natural hazard may spread the pandemic, while an ongoing pandemic does not change the probability of occurrence of a natural hazard. Each of these major hazard sources also has various sub-categories. For example, natural disasters include earthquakes, floods, and hurricanes, while pandemics and epidemics include outbreaks of COVID-19 or an intensified seasonal influenza.A combination of all single-hazard sub-categories and multi-hazard scenarios can be illustrated on a risk matrix, Figure 4b. A risk matrix is a simple way to present the severity and probability of various events [35], increasing the visibility of risks to assist with decision making. Assuming the evaluation metric is the number of fatalities, a path can be developed connecting all the single natural hazards [36]. The same approach can be followed to add in the effects of a pandemic. Presumably, both the likelihood of fatalities and their impacts are increased (or may stay constant in some instances) when a natural hazard occurs during a pandemic. This can be expanded for any combination of two or three hazard sources (not shown in this figure).Since this paper focuses on life loss (i.e., human fatalities) as a main metric for risk analysis, it is important to distinguish the differences between individual and societal risk measures. The individual risk, RI, is defined as the probability that an average unprotected person, permanently present at a certain location, is killed due to a hazardous event [1]:(4)RI=P[ei].P[LL|ei]
+      where P[ei] is the probability of hazardous event i, and P[LL|ei] is the probability of life loss due to ith hazardous event.On the other hand, the societal risk (the expected loss), RS, can be approximated as:(5)RS= ∫∫ARI(x, y).m(x, y)dxdy
+      where m(x, y) is the population density at location (x, y), and A is the area.Arguably one of the most important tasks in any community is keeping the healthcare system as resilient as possible. Resilience refers to the capacity of a system, community, or society to adapt to potential hazards by resisting or changing in order to reach and maintain an acceptable level of functionality and structure [37,38]. The concept of resilience was first introduced in the field of psychology [39] and has been rapidly adopted by environmental [40] and social sciences [41]. Individual researchers have addressed the resilience of communities against natural hazards (e.g., climate-resilient [42], earthquake-resilient [43], flood-resilient [44], pandemic-resilient [45], and politically-resilient [46]).The concept of resilience has risen in popularity during the COVID-19 pandemic, prompting many researchers from various fields to re-evaluate their protocols, systems, and communities to understand how they could recover from adverse effects of COVID-19. Among hundreds of publications, the most notable have focused on medical resilience [47,48], mental resilience [49,50,51], tourist resilience [52], food system resilience [53], supply chain resilience [54], educational system resilience [55], and socioeconomic resilience [56,57]. Furthermore, several researchers have considered the relationships and interactions between risk, hazard, uncertainty, and resilience in the era of COVID-19 [58,59,60,61,62,63,64].Figure 5 qualitatively illustrates a few potential scenarios (some with rare probability) that could happen within a healthcare system. Again, our primary metric is loss of life, which should be controlled (i.e., reduced) during a pandemic. Similar scenarios can be designed for social and economic aspects, which are ignored in this paper. One may note that combining these three hazard sources with different nature is a challenging task because their spatio-temporal domains are different. While most crises or disasters are constrained within a relatively limited space and time, pandemics persist and reverberate for months or even years [65]. Establishing our three main hazard sources as a NH, pandemic, and CEs, the following five scenarios can be discussed, see Figure 5:Pandemic only: This is a single-hazard scenario and assumes no other concurrent hazard threatens the healthcare system. Based on this figure, the healthcare system is assumed to be initially in either full functionality (i.e., 100%), in a degraded mode (>100%), or in an upgraded mode (<100%). Degraded functionality could be caused by aging facilities or personnel and medical equipment shortages. Alternatively, upgraded functionality could be due to the preparedness of a system with prior knowledge about the possible occurrence and dimensions of such a pandemic [66].The performance, or functionality, of a system is reduced with increasing numbers of positive COVID-19 cases. The system has minimum functionality (more or less) when the pandemic is peaking. By reducing the number of infections and designating additional monetary and logistical recourses to the issue, the system recovers from this adverse effect. The transitioning from response to recovery, including consideration of assessment, management, and communication of risk and uncertainty over time was discussed by Menoni and Schwarze [67].Pandemic + Natural Hazard: This is a double-hazard situation that assumes a natural hazard (e.g., earthquake, flood) hits the community during a pandemic. Examples are provided in Section 2.2. The NH-induced functionality loss in this scenario is fairly rapid, compared to the slow reduction of functionality in the pandemic-only case. A natural disaster may impose extra pressure on the healthcare system by occupying a considerable amount of overall hospital capacity. It can also cause a large evacuation, which in turn increases the risk of viral infections among displaced people.Following the sudden functionality loss due to NH, the final compound loss of functionality in this scenario is more than in the pandemic-only one, assuming that the natural hazard can turn into a disaster. Recovery in this scenario is also longer because the natural disaster may cause some physical damage to the healthcare system, which would not occur in the pandemic-only scenario.Pandemic + CE: This is also a double-hazard situation in which multiple CEs (e.g., political conflict, protests) occur during a pandemic. Each of these events, depending on their severity, may or may not reduce the functionality of the healthcare system, including reductions in financial resources, global collaborations, and/or data sharing. Compared to the pandemic-only scenario, the recovery time is higher.On the other hand, the occurrence of such CEs may impact the original pandemic transmission curve by intensifying its peak and elongating its endurance time, see Figure 5 (transition from light gray to darker one).Pandemic + NH + CE: This is a very low-probability, high-consequence situation in which all three hazard sources occur in a relatively short timeframe, though not necessarily at the same time. Such a scenario might cause the largest functionality loss and longest recovery time. One may recognize some states within the U.S. exposed to such multi-risk by overlapping the three maps in Figure 2.The final scenario is an intense version of any of the previous four scenarios. The healthcare system in each county has a limited capacity (e.g., ICU rooms, ventilator machines), and may fail if the imposed demand becomes higher than the “ultimate capacity” of the system [68]. A potential solution is to flatten the transmission curve by imposing stronger stay-at-home orders.Pandemic only: This is a single-hazard scenario and assumes no other concurrent hazard threatens the healthcare system. Based on this figure, the healthcare system is assumed to be initially in either full functionality (i.e., 100%), in a degraded mode (>100%), or in an upgraded mode (<100%). Degraded functionality could be caused by aging facilities or personnel and medical equipment shortages. Alternatively, upgraded functionality could be due to the preparedness of a system with prior knowledge about the possible occurrence and dimensions of such a pandemic [66].The performance, or functionality, of a system is reduced with increasing numbers of positive COVID-19 cases. The system has minimum functionality (more or less) when the pandemic is peaking. By reducing the number of infections and designating additional monetary and logistical recourses to the issue, the system recovers from this adverse effect. The transitioning from response to recovery, including consideration of assessment, management, and communication of risk and uncertainty over time was discussed by Menoni and Schwarze [67].Pandemic + Natural Hazard: This is a double-hazard situation that assumes a natural hazard (e.g., earthquake, flood) hits the community during a pandemic. Examples are provided in Section 2.2. The NH-induced functionality loss in this scenario is fairly rapid, compared to the slow reduction of functionality in the pandemic-only case. A natural disaster may impose extra pressure on the healthcare system by occupying a considerable amount of overall hospital capacity. It can also cause a large evacuation, which in turn increases the risk of viral infections among displaced people.Following the sudden functionality loss due to NH, the final compound loss of functionality in this scenario is more than in the pandemic-only one, assuming that the natural hazard can turn into a disaster. Recovery in this scenario is also longer because the natural disaster may cause some physical damage to the healthcare system, which would not occur in the pandemic-only scenario.Pandemic + CE: This is also a double-hazard situation in which multiple CEs (e.g., political conflict, protests) occur during a pandemic. Each of these events, depending on their severity, may or may not reduce the functionality of the healthcare system, including reductions in financial resources, global collaborations, and/or data sharing. Compared to the pandemic-only scenario, the recovery time is higher.On the other hand, the occurrence of such CEs may impact the original pandemic transmission curve by intensifying its peak and elongating its endurance time, see Figure 5 (transition from light gray to darker one).Pandemic + NH + CE: This is a very low-probability, high-consequence situation in which all three hazard sources occur in a relatively short timeframe, though not necessarily at the same time. Such a scenario might cause the largest functionality loss and longest recovery time. One may recognize some states within the U.S. exposed to such multi-risk by overlapping the three maps in Figure 2.The final scenario is an intense version of any of the previous four scenarios. The healthcare system in each county has a limited capacity (e.g., ICU rooms, ventilator machines), and may fail if the imposed demand becomes higher than the “ultimate capacity” of the system [68]. A potential solution is to flatten the transmission curve by imposing stronger stay-at-home orders.While each of the above-mentioned single hazards may lead to direct fatalities (the main metric discussed in this paper), they may also cause some indirect effects. More specifically, the combination of a pandemic with either NHs or CEs will cause higher infection rates and potentially more fatalities. While mass gatherings due to CEs can be controlled or prevented to some extent, see Section 2.3, evacuations forced by NHs are usually inevitable.Evacuating a large number of people during a pandemic is challenging, given the public health advice to slow the spread of new infections. As mentioned in Section 2.2, during the 2020 Michigan dam failures, a total of 11,000 people were evacuated. The concept of crowd simulation was already studied in different forms [69,70,71]. While there are multiple models to simulate the evacuation of people during hazards, such as wildfires [72,73], earthquakes [74], and tsunamis [75], very little research can be found that directly addresses this issue during a pandemic or epidemic [76]. Therefore, developing such multi-hazard evacuation models is a missing link towards overall community resilience.Figure 6 qualitatively presents a general framework to simulate the evacuations during a concurring pandemic and natural hazard. Any new model should include microscopic and macroscopic crowd models. While the microscopic models take into account spatial-temporal information at the individual-level, city-level evacuations, sheltering, and effective social distancing are governed by macroscopic models. A natural hazard usually hits the entire city (like during an earthquake) or just part of it (like in a flood), forcing people to evacuate with little time. Having an emergency action plan (EAP) is a key factor in responding quickly to hazards, as these plans guide people to the nearest, safest shelters. One may isolate only a small portion of the city (as shown in Figure 6; top right) and develop the model in three parts:Confined space crowd models investigate the occupants’ (or agents’) exposure inside a building during a pandemic and right before or after a natural hazard. In this model, various factors, such as the distance between individuals, the type of transmission contact (e.g., airborne, droplets), and time of exposure, should be considered. The uncertainty associated with the spread of disease can be addressed as one of four potential cases shown in Figure 6:Direct physical contact (e.g., touching).Within the social distance: the exposure might happen if the agentj falls within the social distance (about 2.0 m) of agenti.Being face-to-face within the social distance: the transmission of COVID-19 is higher when the individuals are facing each other or their faces are at a certain angle of each other. This is an important factor especially in commercial or service centers.Being in the same confined area.One may add the following further details to each of four above-mentioned cases:All cases are time-dependent and should be analyzed in the transient mode.All the interactions should be modeled between any two combination of agenti and agentj.Various constraints should be applied to the simulations including but not limited to: using face covering, contagious with or without symptoms, etc.Evacuation models are divided into two parts: evacuating a building and heading towards a shelter. For the latter, factors such as duration, length of travel, difficulty of paths, speed of each individual, potential touching of common surfaces/objects, blocked paths by a group of individuals, and violations of social distancing should be considered.Lastly, sheltering is another major concern during a pandemic, and the capacity of shelters should be recalculated to account for safe distancing between individuals, as well as the length of time evacuees will remain there. Among other factors, the functionality of ventilation systems should be managed to avoid potential damage by a natural hazard.During all three models, a portion of evacuees might become injured, which should be accounted for in evacuation models and added to the resiliency of the healthcare system, Figure 5.This paper highlights the importance and impacts of natural hazards and various complex emergencies in a pandemic era and explains the concept of multi-hazard risk in three hazard scenarios. Two major ideas are qualitatively proposed for future detailed research: (1) the need for resilience models that explore the healthcare system under multi-hazard risk, potential forms of functionality loss, and the recovery duration; and (2) the need for pandemic-specific evacuation and sheltering models that also cover the risks posed by NHs.While the skeleton of the paper was formed based on the data, hazards, and events that have been reported in the U.S., the idea can be expanded to any other country without loss of generality. While all countries in the world are fighting COVID-19 outbreak, natural hazards (from different types) are also inevitable. For example, since January 2020, several major natural disasters have been reported worldwide, including: (1) 5.3 magnitude earthquake on 22 March in Zagreb, Croatia [77] (one fatality, 27 injured); (2) 5.1 magnitude earthquake on 7 May in Tehran, Iran (two fatalities, 38 injured); (3) earth-fill dam break on 1 May in Uzbekistan (four fatalities, 50 injured, and 70,000 evacuees) [78]; (4) Tropical Storm Amanda, formed on 31 May, along the coast of Guatemala (at least 17 fatalities); (5) Cyclone Harold [79] in the Solomon Islands, Vanuatu, Fiji, and Tonga between 1 and 6 April, which destroyed many homes; (6) wildfire outbreaks in the west, southwest, and south of Iran that burned more than 500 hectares of forest between 27 May and 3 June; (7) multiple floods impacted large tracts of Southern China in June and July due to heavy rains, which affected more than 37 million people, and left about 140 death/missing.Also, add the ongoing (or new) international complex emergencies to this multi-hazard scenario. The examples of ongoing challenges are: Hong Kong protests, the war/conflicts in the Middle East (such as Syria, Yemen, Afghanistan), etc.Some other types of complex emergencies are predictable; however, blocking them temporarily may even cause extra future consequences. For example, the authorities in Iran have to face the dilemma of canceling (due to COVID-19) nation-wide university entrance exam early August with about 1.5 million participants (which practically cripple the entire educational system for the upcoming academic year), or risking their lives by a half-a-day exam in the indoor classrooms/environments.One major conclusion out of this paper is that a multi-hazard situation combining any three hazard sources of pandemic, natural hazard, and complex emergency might have a cascading effect. Since various dimensions of this problem is still unknown (i.e., we do not have a quantitative metric to evaluate the risk, and we clearly are not prepared to face it), the authors implore governments to allocate additional financial resources to multi-hazard risk research, paving the way for a safer, less uncertain future. While the COVID-19 pandemic and all its consequences were unfortunate for the society, some researchers note that it might yield positive impacts for future resilience design, plans, and politics within built environments [65,80]. Last but not least, the author believes that anyone in any position should contribute (to the extent possible) to improve the knowledge related to the COVID-19 outbreak, and as Haas [81] truly said: “Risk analysts and risk analysis researchers should not be shy about contributing our skills to important policy developments during this crisis.”This paper proposed a multi-risk assessment framework in the following general form, g, in which the risk increases in the presence of a pandemic, natural hazard, complex emergencies, and in the lack of a sufficient healthcare system:(6)Multi Risk∼g (Pandemic, Natural Hazard, Complex EmergencyHealthcare System)While this paper proposed a general framework, we did not present a quantitative example (case study). At the time of publication of this paper, the world was in the middle of a COVID-19 pandemic, with no definite database on full interaction of different hazard sources. As a future work, the idea presented in this paper can be applied to the database collected at the national or international levels.This research received no external funding.The author would like to pay his tribute and appreciation to all front-line workers who are fighting the COVID-19 pandemic.The author declares no conflict of interest.A multi-risk condition.Spatial distribution of risk of exposure (pandemic), potential natural hazard, and mass gathering (of protests) across the U.S.; all maps are approximate and for illustration purposes only.Complex emergencies and protesting during COVID-19.Matrix presentation of multi-hazard and multi-risk.Response and recovery of healthcare system under multi-hazard scenarios during pandemic; Five color lines present the response/recovery of the healthcare system; color (red, blue and yellow) circles show the occurrence of three hazard sources; the black circle shows the start and end point of the resilience curves; the white circle presents the minimum resiliency (or peak pandemic); the light and dark gray bell shapes are the pandemic progress over time in the original form and altered by CE, respectively; and finally, the colored transparent rectangle on the top left side of the figure presents the individual/cumulative functionality loss.Evacuation and sheltering in pandemic era after a natural hazard; in city level figure (top left), the circles and squares present different types of buildings; the timeline axis presents the pandemic era before natural hazard occurrence, t0, during staying inside a confined place, t1, during evacuation, t2, and during sheltering, t3.

Med-MDPI/ijerph_5/ijerph-17-16-05636.txt

@@ -0,0 +1 @@
+Family physicians act as gatekeepers of the healthcare system and have an indispensable role in providing holistic care in the primary care system. While previous studies had focused on the geographic maldistribution of family physicians, the current study investigated the distribution of job opportunities for family physicians by analyzing recruitment advertisements posted in medical association journals, as an indirect way to observe the marketplace demand for physicians. We collected all the recruitment advertisements for family physicians in the twelve issues of the Taiwan Medical Journal, the official organ of the Taiwan Medical Association, in 2018. In contrast to 124 new trainees annually, 739 advertisements for family physicians were posted within the entire year. After eliminating repeated advertisements, there were 302 distinct advertisements, of which hospitals accounted for 18.9% (n = 57). The job opportunities at hospitals were offered mainly by regional hospitals (n = 26) and community hospitals (n = 29), but only two by medical centers. Family physicians in Taiwan were in great demand not only by primary care clinics but also by hospitals. The role of family physicians in hospitals is worth further study.Family medicine is a specialty with the most versatility among all the medical specialties, providing comprehensive medical care to patients of all ages and genders. Family physicians are just as diverse as their patients and are known to have much flexibility in the choice of practicing. Compared with other specialties that often need advanced equipment which can only be offered in hospitals, family physicians are more likely to practice in diverse locations, including the rural or countryside areas. However, previous studies have discussed the phenomenon of geographic maldistribution of primary care physicians in the United States and in Japan, showing the dearth of family physicians in rural areas [1,2,3]. A 2018 statistics report in Taiwan showed that about 70% of family physicians work in metropolises, which account for 30% of Taiwan’s land area in 2018 [4]. Aside from collecting data by questionnaires to evaluate the subjective factors for career choice, some studies also analyze the need for family physicians by observing recruitment advertisements posted in medical journals. The latter method is more objective in describing the social recognition of family physicians [5]. The changing number of recruitment advertisements provides an idea of the change of marketplace demand for physicians where the ongoing analysis of advertisements provides timely information about the demand for physicians in a rapidly changing health care system. Previous studies had discussed about the recruitment of different specialties including radiation oncology [6], radiology [7], dermatology [8], dentistry [9], and general medicine [10]. In the study of the radiation oncology job market, job availability differed among regions, and more open positions were in rural locations [6]. A help-wanted index was used in the detecting of an interventional radiologist job market, revealing a dramatic job shift [7]. With regards to the field of dermatology, a rising shortage of dermatologists and an increase in demand for dermatologic services were noted [8], so as the finding in the dental faculty [9]. A similar predicament was revealed in general physicians [1]. Due to the general physician shortage, most family physician-related studies focused on strategies to recruit and retain primary care doctors [10,11,12].Within the healthcare system and National Health Insurance program in Taiwan, patients can visit specialists of any type at any level of facilities directly without referrals. Inevitably, the role of family physicians as gatekeepers is weakened and hospitals take responsibility for primary health care. Therefore, many family physicians choose to remain to work in hospitals (e.g., medical center, regional hospital, and community hospital) rather than in rural clinics [12]. The Taiwan government had been promoting a Family Physician Integrated Care Project since 2003 to highlight community medical group-based practice [13].The current study aimed to analyze the nationwide recruitment of family physicians in Taiwan in 2018, stratified by area and type of medical facilities. Furthermore, the focus would be on situations in hospitals, especially the workforce distribution and need for family physicians at different levels of hospitals under the current health system and policy.Taiwan is a country with 23 million people, located at the eastern part of Asia. The Taiwan Medical Association was established in 1930, and there were over 49 thousand members in 2018, practicing in 22 medical specialties [14]. Family medicine, which topped the list of the specialties announced by the Ministry of Health and Welfare, was introduced to provide holistic care and to act as a gatekeeper of the health care system [15]. The Association of Family Medicine was formally established in 1986 in Taiwan to promote family medicine research and development, to develop a family medicine specialist system in Taiwan, to strengthen contact and exchange with international family medicine organizations, and to raise primary medical care and family medicine standards [16]. In 2018, there were 3655 family medicine practitioners working in different levels of medical facilities, including residents who were training under the family medicine specialty [17].After six years of studying in medical schools and five years working as residents, there are approximately 120 newly trained family medicine practitioners each year in Taiwan, and most of them work as general practitioners after obtaining a specialist license, either in hospitals or in local clinics [18]. We collected all the recruitment advertisements in the twelve issues of the Taiwan Medical Journal in 2018, which were published by the Taiwan Medical Association every month since 1957, targeted for all practitioners in Taiwan, with the content of updating medical information of various specialties. We recorded the advertisements on family physician recruitment and marked the advertisements that were repeatedly posted. We also recorded the advertisements of each medical care facility with their location and accreditation level. The numbers of medical care facilities and practitioners presented were from the statistics of the Taiwan Medical Association. The data of land area and population were from the statistics of Taiwan’s Ministry of the Interior.We collected the information of all the advertisements, including the name of the facility, location, specialty, number of physicians required, and special requirements. There were some advertisements that did not show the exact number of physicians needed, and we counted them as one recruitment each. Some facilities have different branches in different cities, and we counted them as independent facilities separately. There were repeatedly posted advertisements. We marked them and compared the number between newly posted and repeatedly posted advertisements in each month. The data collected in the twelve issues of the Taiwan Medical Journal were first grouped into two groups—hospital and clinic—and the data in the hospital group were then further divided into three groups—medical center, regional hospital, and community hospital. To observe the difference between different locations, the data were divided into 7 groups according to their location: northern Taiwan consists of Taipei City, New Taipei City, and Keelung City; northwest Taiwan consists of Taoyuan City, Hsinchu County, Hsinchu City, and Miaoli County; central Taiwan consists of Taichung City, Changhua County, and Nantou County; southwest Taiwan consists of Yunlin County, Chiayi County, Chiayi City, and Tainan City; southern Taiwan consists of Kaohsiung City and Pingtung City; eastern Taiwan consists of Yilan County, Hualien County, and Taitung County; and offshore islands consist of Lienchiang County, Kinmen County, and Penghu County. We also surveyed the population density, dependency ratio, number of family physicians, number of family medicine residents, and number of family physician training facilities of each location. We considered that population and age structure might contribute to the demand for family physicians under the presumption that each family physician can serve the same number of patients.We surveyed the number of practitioners, training residents, and recruitments. The number of practitioners was calculated as the number of family medicine members minus the number of residents, with data collected from the Taiwan Medical Association and the Taiwan Association of Family Medicine, respectively.The data collection and analysis were performed with Microsoft Excel 2016 (Redmond, Washington, DC, USA) and presented in descriptive statistics. The categorical variables were presented in numbers and percentages.There was a total of 12 Taiwan Medical Journal issues published in 2018, one for each month, and an overall number of 739 recruitment advertisements for family physicians were posted in these issues. There were 312 advertisements posted by hospitals and 427 advertisements by clinics, and the number varied in each month (Figure 1). The advertisements were further grouped into newly posted and repeatedly posted ones. For the entire year of 2018, hospitals posted 57 new and 255 repeat advertisements, while clinics posted 245 new and 182 repeat advertisements.There were more advertisements posted by clinics from September to December 2018 wherein the combined number of newly posted and repeatedly posted advertisements reached more than 40 per month. In contrast, the advertisements posted by hospitals remained mostly the same, with the combined number of advertisements around 20 to 30 per month. It was notable that for every month in 2018, clinics consistently posted more newly posted recruitment advertisements than hospitals (Figure 1).General information about each area is shown in Table 1 including population density, dependency ratio, number of family physicians, number of family medicine residents, number of family physician training facilities, and number of recruitment advertisements. With regards to the geographic distribution of family physicians and recruitments in Taiwan, central Taiwan had the largest number of recruitment advertisements (276 out of 739), as well as distinct recruitments (94 out of 302) and ratio of recruitment over physician (42.9%). Northwest Taiwan (141, 32.1%) and southwest Taiwan (108, 21.2%) were the second and third most in number and ratio of recruitment over physician. The smallest numbers of recruitments were from eastern Taiwan (3) and offshore islands (0) (Table 1).With regard to the distribution of family physicians and recruitments by accreditation level of the medical care facility, there were 2143 (67.9%) physicians practicing in clinics and 1011 (32.1%) physicians practicing in hospitals, with 461 (14.6%) in regional hospitals, 333 (10.6%) in community hospitals, and 217 (6.9%) in medical centers. There were 245 (81.1%) distinct recruitments posted by clinics and 57 (18.9%) recruitments posted by hospitals, with 29 (9.6%) by community hospitals, 26 (8.6%) by regional hospitals, and 2 (0.7%) by medical centers. The ratio of recruitment over physician was 11.4% in clinics, 8.7% in community hospitals, 5.6% in regional hospitals, and 0.9% in medical centers. Most residents were trained in medical centers (296, 59.1%), followed by regional hospitals (186, 37.1%) and community hospitals (19, 3.8%) (Table 2).Our study investigated the factors that influence family physician recruitment conditions in Taiwan, specifically through recruitment advertisements. We found that the number of advertisements posted by clinics was more than those posted by hospitals, and the numbers differed by month, location, and accreditation level of the medical facilities. The number of advertisements posted by clinics surpassed those by hospitals, especially in September to December 2018. This finding may be correlated with the schedule of the family medicine specialist examination where the exam is held in November and December every year, starting with a written exam and followed by an oral exam two weeks later [25]. During this period, most residents have finished the required training course in hospitals and would search for opportunities in various medical facilities. This would influence the recruitment condition and turnover of the job market. However, this phenomenon was not reflected in the recruitment advertisements posted by hospitals. Previous studies showed that family physicians working in hospitals were significantly more likely than their non-hospitalist peers to work longer hours, have better pay, and be more satisfied with their work [26,27]. These attributes may lead to a low turnover rate in hospitals.Northern Taiwan is an area in Taiwan with the largest and densest population in the nation, and central Taiwan has about two-thirds of the population of northern Taiwan. A larger population would presumably present a larger demand for physicians. However, findings revealed that the number of recruitment advertisements in central Taiwan was twice that in northern Taiwan. This indicated that the turnover rate in central Taiwan was higher regardless of its small population. One reason may be that the number of medical centers differed in these two areas. There are 12 medical centers in northern Taiwan, but only 6 in central Taiwan, since most hospitals in that area were either regional or community hospitals. The high turnover rate in central Taiwan may indicate that physicians tend to search for new jobs after residency training instead of staying in the hospitals where they were initially trained in [28,29].In Taiwan, only 67.9% of family physicians work in local clinics. However, most recruitment advertisements (81%) were posted by local clinics. This showed that the supply was less than the demand. Another reason that may be attributed to this finding is that because all residents were trained in hospitals, there would be a certain number of physicians who chose to stay in the same hospitals they were trained in without the help of recruitment advertisements [30]. In this study, we observed that the majority of residents were trained in medical centers (59.1%) as opposed to regional or community hospitals. This may be the reason why the number of advertisements posted by medical centers (2, 0.7%) was less than that posted by regional hospitals (26, 8.6%) and community hospitals (29, 9.6%). The supply was more than the demand in medical centers. Therefore, advertisements were not the main source of recruitment. On the other hand, there were about 124 new trainees annually, and the demand shown by recruitment advertisements by hospitals accounted for nearly half of the numbers despite there being other sources of recruitment. The roles of family physicians in hospitals are worth further research [31].Overall, this study presented the annual family physician demand for different months, different localities, and different accreditation levels of the medical care facility. This corresponds to the fact that family physicians in Taiwan not only provide frontline medical service in primary care facilities but also play an indispensable role in hospitals. Most residents were trained in hospitals, and they tend to stay in the same hospital after finishing training. Nevertheless, hospitals still need to post recruitment advertisements because the lack of family physicians in hospitals still exists. This might be related to the additional workload of family physicians aside from outpatient service, where they would need to promote medical screening for at-risk groups and vaccination programs, among others. The work and role of family physicians in hospitals that is related to public health policy development calls for a further study.The source of recruitment advertisements for our study was limited to the Taiwan Medical Journal. There were other recruitment platforms such as the internet, newspapers, and internal recruitment that were not included. Even with the only journal, our data might be underestimated because some advertisements without specifying the number of physicians needed were operationally deemed as one physician needed. In this study, we counted only the amount of recruitments and did not analyze the working conditions, salaries, and benefits, which were usually not provided in the advertisements. The advertisements may not be limited only to residents, but targeted to family physicians in general. The number of advertisements in an area may also correlate with a high turnover rate where, for example, an area with a higher number of advertisements may not be in a greater shortage of family physicians than other areas. Furthermore, our study described only job offers, but the actual applications and the results remained unknown, which warrants further research for clarification.According to our analysis of recruitment advertisements in the Taiwan Medical Journal, family physicians in Taiwan were in great demand not only by primary care clinics, but also by hospitals, especially by regional and community hospitals. Beyond the fact that family physicians provide frontline medical services in the primary care system, there is a need to conduct further studies discussing the roles of family physicians in hospitals and the association between the supply and demand of the family physician workforce.Y.-S.C. and T.-J.C. conceived and designed the study. Y.-S.C., A.C.T., Y.-C.H. and T.-J.C. conducted the investigations and interpreted the data. Y.-S.C. wrote the manuscript. A.C.T. and T.-J.C. revised the manuscript. All authors have read and agreed to the published version of the manuscript.This study was supported by grant (V109E-002-1) from Taipei Veterans General Hospital.The authors declare no conflict of interest.Family physician recruitment advertisements in the 2018 Taiwan Medical Journal.Distribution of family physicians and recruitments by area in Taiwan.* Dependency ratio: a measure of the number of dependents aged 0 to 14 and over the age of 65, compared with the total population aged 15 to 64, calculated as: (numbers of aged 0 to 14 and over the age of 65)/(numbers of aged 15 to 64).Distribution of family physicians and recruitments by medical care facility accreditation level in Taiwan.

Med-MDPI/ijerph_5/ijerph-17-16-05637.txt

@@ -0,0 +1,2 @@
+Background: Endocrine mechanisms can be a determining factor in the neuromuscular performance of young athletes. Objective: The objective of the present study was to relate maturational and hormonal markers to neuromuscular performance, as well as to verify whether young athletes with different testosterone levels show differences in muscle strength. Methods: The sample consisted of 37 young male Brazilian athletes (11.3 ± 0.94 years) who were members of a sports initiation project. Hormonal markers were analyzed biochemically by blood samples, and maturation markers by mathematical models based on anthropometry. Body composition was verified by tetrapolar bioimpedance. The performance of upper and lower limb strength and body speed were analyzed. Results: Hormonal and maturational markers were related to neuromuscular performance (p < 0.05). Young people with higher testosterone levels showed higher muscle strength (p < 0.05). Artificial neural networks showed that testosterone predicted the performance of upper limbs by 49%, and maturation by 60%. Maturation foreshadowed the performance of lower limbs by 30.3%. Conclusion: Biological maturation and hormonal levels can be related to neuromuscular performance, and young people with higher testosterone levels show superior muscle strength in relation to the others.The quality of neuromuscular performance, especially of the upper and lower limbs, has been identified as a determinant for physical health and success in several sports skills [1,2]. However, several biological and environmental factors can interfere with the neuromuscular quality of skills that require the use of motor coordination and muscle strength [3,4,5]. At the biological level, we can highlight the puberty process, which is the triggering of improvements in neurological, muscular, skeletal, and endocrine systems [6]. Among the environmental factors, we can highlight the level of physical activity, nutrition and quality of sleep, which are extremely important for the quality of the biological maturation process [7,8].The interaction between environmental and biological determinants can influence the pace of puberty process, that is, individuals born on the same chronological date may have different biological ages due to these interactions [9]. In this sense, puberty can be classified as early, synchronized or late; these different levels are determined according to the timing of the chronological and biological ages [10]. Each stage of puberty has different characteristics in relation to biological maturation; the endocrine system receives different stimuli, and hormonal levels stand out as one of the main differences between the different levels of puberty [11,12]. Subjects with precocious puberty generally have hormonal levels higher than those of late and synchronized puberty [13,14].The maturation process resulting from adolescence is strongly governed by reactions from a range of hormones, which affect the maturation rhythm of young people in favor of the increase in sex hormones [15]. Sex hormones, such as estradiol, testosterone, and growth hormone, have extremely important functions in human biological development, which can mainly affect morphological structure, skeletal growth, and muscle strength levels [9,10,11,12,13,14,15,16,17]. In this context, among neuromuscular parameters, muscle strength and body speed are two of the most essential for body functionality in relation to physical health, and for the development of specific skills related to sports performance [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19].In female subjects, as puberty advances, the levels of steroid hormones tend to increase, pointing to a significant relationship with the levels of muscle strength in the upper and lower limbs [14]. In addition, the secretion of steroid hormones during puberty generates changes in the physical profiles and secondary sexual characteristics of subjects of both sexes [20,21]. At the endocrine level, the stages of advanced puberty are characterized by a predominance of sex hormones in the body [14,15]. However, young people from the same maturation stage may have different levels of hormones in the body [22]. Maturation cannot be the only determinant of neuromuscular characteristics in the pediatric population; analyses of hormone thresholds are necessary to obtain a parameter with a central tendency that serves as a reference point to group the subjects [22,23,24].The stages of puberty relate to physical abilities. This aspect is widely addressed in sports with the aim of obtaining information about the process of detecting and guiding young talents [1]. In addition, hormonal markers also point to relationships with physical abilities [14]. In this context, both themes should be observed together so that more accurate results can be obtained in relation to factors related to neuromuscular performance in young athletes, especially when observing that motor performance can provide crucial information in the talent selection process in sport.In this context, in scientific research linear analyzes are interesting to discriminate the importance of variables in relation to performance in sport [14]. Moreover, non-linear patterns can be significant in relation to the characteristics analyzed [14]. In this sense, linear (i.e., regression) and non-linear (i.e., artificial neural networks) analyses have been shown to be effective in providing specific data, in relation to the muscular strength of young sports initiation athletes [14].Given the assumptions, the present study hypothesized that: (1) young males with higher concentrations of testosterone and with more advanced pubertal stage may be superior in neuromuscular performance when compared with young people with lower concentrations of testosterone and with later pubertal stage; (2) biological maturation and hormonal markers may be related to the neuromuscular performance of young males. Therefore, the aim of the present study was to relate maturational and hormonal markers to neuromuscular performance, as well as to verify whether young athletes with different testosterone levels show differences in muscle strength.The study was observational with a cross-sectional design. The sample consisted of 37 male adolescent athletes (age of 11.3 ± 0.94 years) who were members of a sports initiation project in the city of Natal, Brazil. The sample size for this research was established a priori, based on a previous study [13], and through an effect size of 0.66 and an α < 0.05 with a β = 0.80. As inclusion criteria, volunteers were between 10 and 12 years old, male and had no clinically diagnosed hormonal dysfunction. All participants who took food or hormonal supplements or who performed vigorous activities in the last 24 h before the exams were excluded.This research was analyzed and approved by the Ethics and Research Committee-CEP of the Federal University of Rio Grande do Norte (Opinion: 1249937), according to resolution 466/12 of the National Health Council on 12/12/2012, while strictly respecting the ethical principles contained in the Declaration of Helsinki. In addition, the present study complies with all items on the STROBE checklist for observational studies (i.e., checklist to strengthen the reporting of observational studies in epidemiology) [25].The sample participants were accompanied by a team of nursing and physical education professionals previously trained and qualified by the Federal University of Rio Grande do Norte to carry out the data collection. Thus, the sample came to the laboratory on three occasions. The first visit was used to inform the volunteers and their respective guardians about the research objectives and the methodology adopted in the study. On the second visit, which took place 24 h after the first, after the volunteers and their respective guardians signed the terms of free and informed consent, blood samples were taken from the fasting individuals for biochemical tests referring to hormonal markers. Next, the body composition test was applied using a tetrapolarbioimpedance scale, and volunteers were offered a snack after an hour-long break, anthropometric measurements were taken. On the third visit, again after 24 h, arm and leg strength and body speed tests were performed, in that order. During the second and third visits, the volunteers were instructed to suspend vigorous activities in the 24 h prior to data collection (Figure 1). It should be noted that during the evaluations the sample and its guardians did not have access to any of the results during the tests performed.The body composition examination was performed using a tetrapolar bioimpedance equipment model BIA1010 with high precision ((Resistance: Range: 0–1000 Ohms; FS: 1000 Ohms; Resolution: 0.1 Ohm; Accuracy = 0.5% FS); (Reactance: Range: 0–1000 Ohms; FS: 1000 Ohms; Resolution: 0.1 Ohm; Accuracy = 1% FS))(Sanny®, São Paulo, Brazil), where the evaluated subject was placed in the supine position on a stretcher isolated from electric conductors, and remained at rest for 10 min without wearing any metallic object (i.e., earrings, bracelets, piercings, etc.). Subsequently, emitting electrodes were placed on the surface of the hand and the right foot, close to the joints of the metacarpal and metatarsal phalanges, respectively. The receiving electrodes were placed at the midpoint between the distal prominences of the radius and the ulna of the right wrist, and between the medial and lateral malleolus of the right ankle. Subsequently, pediatric algorithms were selected in the equipment and after the procedures, the evaluator activated the bioimpedance, and the equipment recorded the body composition values through a portable computer attached to the device.The anthropometric evaluation was carried out through body mass, height, perimeters, bone diameters and skin folds. All assessments were based on the ISAK protocol (International Society of the Advancement of Kinanthropometry) [26]. Body mass was measured using a digital scale (Filizola® São Paulo, Brazil; with a capacity of up to 150 kg and a variation of 0.10 kg); height was measured with a stadiometer (Sanny®, São Paulo, Brazil; accurate to 0.1 mm); skinfolds were measured using a scientific adipometer (Harpenden®, Londres, Inglaterra; John Bull Indicators Ltd.; accurate to 0.1 mm); perimetry was measured using an anthropometric tape (Sanny®, São Paulo, Brazil); an anthropometric ruler (Sanny®, São Paulo, Brazil) measured the length of the tibia and femur; and a caliper (Sanny®, São Paulo, Brazil) was used for bone diameters of the humerus and femur.Biological maturation was analyzed using three distinct parameters: skeletal maturation, which refers to the growth of the human skeleton and the closure of the main bone epiphyzes; sexual maturation, which is associated with secondary sexual characteristics (i.e., pubic hair and genital growth); and somatic maturation, which is linked to the morphology of biological tissues—epithelial, connective, muscle and nervous tissues). Skeletal maturation was analyzed through bone age, which was verified by the equation proposed by Cabral et al. [27]. The equation is highly reliable compared with the gold standard hand and wrist X-ray method (r = 0.868; r² = 0.754; p < 0.05). The mathematical model consists of the formula:Bone age = −11.620 + 7.004 × (height (cm)) + 1.226 × (Dsexo) + 0.749 × (age)−0.068 × (triceps skinfold (mm)) + 0.214 × (corrected arm circumference (cm))−0.588 × (humerus diameter (cm)) + 0.388 × (femoral diameter (cm)).(1)For the male Dsexo = 0. For the female Dsexo = 1.To determine maturation stage, the result of skeletal maturation must be compared with the chronological age and when the subject presents values below the chronological age the individual is classified as late, when the values are similar they are classified as synchronous, and when bone age value exceeds the value of chronological age the individual is classified as accelerated [1].Sexual maturation was verified by the mathematical model proposed by Medeiros et al. [28]. The equation has a strong validation coefficient with the sexual maturation exam performed using a medical exam in a specific office (Kappa validation coefficient = 0.840; p < 0.05). The formula is as follows:Male sexual maturation = (0.49436 × age) + (10.74526 × trunk height (cm)) + 0.11583 × acromion length (cm)) − (0.1394 × tibial length (cm)) − (0.2808 × femur length (cm)) + (0.5963 × forearm circumference (cm)) + (0.22397 × neck circumference (cm)) − (0.5155 × waist circumference (cm)) − 19.69139.(2)Based on the score resulting from the equation, the stages of sexual maturity are estimated according to the following discriminant score (DS) values: prepubertal stage 1 (DS ≤ −1.67491), pubertal stage 2 (−1.67491 ≤ DS ≤ −0.79575), stage 3 pubertal (−0.79575 ≤ DS ≤ 0.27226), stage 4 post-pubertal (0.27226 ≤ DS ≤ 2.21446) and stage 5 post-pubertal (DS ≥ 2.21446).Somatic maturation was verified by the equation proposed by Mirwald et al. [29]. The equation has a strong validation with the longitudinal monitoring of maturation through bone mineral accumulation (r = 0.959; r² = 0.920; p < 0.05) and is as follows:Peak of speed growth (PSG) in male sex = −9.236 + [0.0002708 × (leg length (cm) × trunk height (cm))] + [−0.001663 × (age × leg length (cm))] + [0.007216 × (age × trunk height (cm))] + [0.02292 × (weight (kg)/ stature (cm)) × 100](3)Based on the final values of the equation results, in relation to chronological age, subjects can be classified into three stages of somatic maturation: (1) Pre-PSG (PSG < −1); (2) During PSG (PSG ≥ −1 or PSG ≤ + 1); and (3) Post-PSG (PSG > +1).In this study, the upper limb power (ULP) was analyzed using the medicine ball test, previously validated in Brazilian children [30]. Participants were asked to sit with their backs against a wall and their knees extended. At the evaluator’s signal, a 2-kg medicine ball (Ax Sports®, Tangará, Brazil) was positioned at the height of the sternum and they were asked to throw it horizontally using both hands. The use of trunk movement was not allowed. The test was performed consecutively three times, interspersed with a passive recovery period of 3 min. The best attempt measurement was considered for the analysis.Lower limb power was assessed by a squat jump (SJ) using a jump mat (Cefise®, Brazil) connected to the Jump Test Pro 2.10 software [31]. It is noteworthy that the test used is effective and feasible in Brazilian children and adolescents [13,14]. To reduce errors while executing the protocol, participants were familiarized with the test. Starting from an orthostatic position, held for 3 s, with the knees flexed at approximately 90º and the hands fixed on the waist, the participants were instructed to perform the jump as high as possible. Three attempts were made, interspersed with 40 s of passive recovery, and the best attempt measurement was included for data analysis.To assess upper limb speed (ULS), participants performed the tapping test, previously validated in children and adolescents, in which the participant uses the dominant hand in the shortest possible time [32]. The test is performed with the participant in front of a table of adjustable height to the hip level so that the non-dominant hand is positioned in a central rectangle drawn on the table. The participants then touch the disks drawn on the sides of the table with their dominant hand. In this study, each participant performed 25 cycles between one side disk and another in the shortest possible time.The body speed test with change of direction (BScD) was conducted according to the recommendations of Nimphius et al. [33]. Two vertical lines were drawn on the ground with a distance of 10 m between them and the central point was positioned in the middle of the 5-m distance marked on the ground with the drawing of a circle. This helped the participants run as fast as they could from the circle, touch the left line, change direction, touch the right line, and return to the central circle, immediately starting the same procedure again three times in a row. This protocol was performed in two attempts, interspersed with a 3-min passive recovery time, and the best attempt measurement was considered for data analysis.Peripheral blood samples (10 mL) from the antecubital vein were obtained from participants to analyze hormone levels. The blood sample tubes were centrifuged with a clot activator at 6000 rpm for 10 min, to obtain 0.5 mL of serum. The serum samples were kept on ice at −20 °C and monitored directly (the transport lasted 5 min) to analyze the serum dosage of the hormones (growth hormone–GH, testosterone–TRT, and estrogen-type estrogen–EST). Subsequently, measurements of growth hormone, testosterone, and estradiol were made in nanogram per deciliter (ng/dL). The levels of these hormones were also measured using the direct chemiluminescence method (i.e., reduction of light as a result of a chemical reaction in a blood sample) with the ADVIA Centaur® XP–SIEMENS (i.e., photomultiplier). The process transforms the light emitted by the chemiluminescence method into electrical impulses and thus the impulses are read in “count” of light per second (i.e., this unit is proportional to the unit of measurement of the hormone levels present in the sample).The normality of the data was verified using the Kolmogorov–Smirnov and Z-score tests for asymmetry and kurtosis (−1.96–1.96). The estradiol variable had the assumption of normality denied and it was transformed from non-parametric to parametric by the log at the base of 10. Correlations were made using Pearson’s test. In the partial correlations, the effect of the maturational and hormonal variables was statistically controlled. The Schober et al. [34] scale was used for the correlations: Insignificant: r < 0.10; Weak: r = 0.10–0.39; Moderate: r = 0.40–0.69; Strong: r = 0.70–0.89; Very strong: r = 0.90–1.00. Through the median split of the testosterone variable, the sample was categorized into a testosterone group < 100 ng/dL and a testosterone group > 100 ng/dL. Bonferroni correction was applied, and later comparisons were performed using Student’s independent t-test. The effect size and the respective 95% confidence intervals were obtained using the Cohen test (d). The magnitude of the effect size followed the classification recommended by Espírito Santo and Daniel [35]: insignificant < 0.19; small 0.20–0.49; mean 0.50–0.79; large 0.80–1.29; very large < 1.30). Linear regressions were performed and the models had the homogeneity tested by the Breush–Pegan test and the assumptions of normality, variance and independence of the data were not denied. Non-linear artificial neural networks of the preceptron type with multiple hidden layers, containing Gausian distributions were programmed. The objective was to determine the prediction of the probability of correct predictions of the maturational and hormonal variables in relation to the neuromuscular variables were used to perform synaptic weight adjustments. Thus, the network activation functions followed the following binary interpretation: (0) False prediction: U < 0, so U is equal to zero; (1) True prediction: U ≤ 0, so U is equal to 1. In addition, for the validation of neural networks, 89.1% of the dataset was used to train the neural networks and 10.9% was used to test the applicability of the programming. During training, the networks were programmed to stop the executions of the seasons when finding the lowest possible error rate for the dataset under analysis, thus an average of 10.000 training seasons was performed. Subsequently, to carry out cross-validation, the database was subdivided into four groups of datasets, 3 with 10 subjects and 1 with 7 subjects, then there was the rotation between training the preceptron network and testing the network, until all groups had gone through both conditions. At the end, the forecasts obtained by the four different groups were added and the average was considered as the final result. For the technical error of anthropometric measurements, the following magnitude was used: Acceptable for skin folds ≤ 5.0%; Acceptable for other anthropometric measurements ≤ 1.0% [36]. All analyses were performed using the R statistical software (version 4.0.1; R Foundation for Statistical Computing®, Vienna, Austria), and the significance level of p < 0.05 was considered.The characterization of the sample shows that the subjects had delayed skeletal maturation, somatic maturation at the prepeak of growth speed and prepubertal sexual maturation (Table 1). The sample showed superiority of lean mass in relation to fat mass in kilograms, and mean levels of growth hormone of 2.27 (ng/dL) and steroid hormones between 20.7 and 40.0 (ng/dL) (estradiol and testosterone, respectively). Regarding anthropometric measurements, technical errors of intra-rater measurements below 3% for skin folds and below 1% for other anthropometric measurements were pointed out. The margin of error pointed out for the sample was 4.83%, being below 5% and the calculated sample power was 0.82. There was no sample loss.Table 2 shows that testosterone showed a significant relationship with lean mass (kg), estradiol with the performance of SJ and BScD’s and growth hormone showed a relationship with lean mass and the percentage of fat. It was not possible to observe, when statistically controlling for the effect of biological maturation by three different parameters (maturation: skeletal, sexual, and somatic), the correlations of testosterone and growth hormone with lean mass. However, the correlations between estradiol and the performance of SJ and BScD’s remained significant. The same occurred for the correlation of growth hormone with the percentage of body fat.Table 3 shows that sexual, skeletal, and somatic maturation have significant correlations with lean mass and the performance of ULP and SJ. In addition, it appears that skeletal maturation also showed a significant correlation with ULS.The linear regression models contained in Table 4 report that biological maturation presented statistical significance as predictors for lean mass, ULP and SJ (W/kg). In addition, skeletal maturation was shown to be significant in helping to predict BScD. Regarding hormone levels, testosterone was significant for the prediction of lean mass, estradiol for the performance of SJ and BScD. Regarding growth hormone, it was only found statistically significant for the prediction of body fat percentage.The analyses of multilayer neural networks of the preceptron type in Table 5 show that testosterone (ng/dL) has a 49.5% probability of estimating lean mass (kg) and a 49% chance of predicting ULP. Sexual maturation showed 32.5% of correct answers to predict lean mass and 51% of chances in predicting ULP. Skeletal maturation indicated a 35.1% chance of predicting lean mass, 55.4% of predicting ULP and a 34% chance of predicting SJ. While somatic maturation indicated 83.1% of capacity to foresee lean mass, 60% in ULP and 30.3% in SJ. The hormone estradiol (ng/dL) and the growth hormone did not present probabilities of success above 30% for the predetermined predictions.Regarding sexual maturation, 16 subjects were classified as prepubertal (sexual maturation: −2.62 ± 0.54) and 21 as pubertal (sexual maturation: −0.97 ± 0.43). In Figure 2, comparisons between individuals according to maturation stage show that pubertal subjects had higher testosterone levels (effect size: 1.08; 95% CI: [0.36–1.80]; p = 0.04); higher concentration of lean mass (effect size: 1.74; 95% CI: [0.95–2.53]; p < 0.0001); lower percentage of fat (effect size: 0.72; 95% CI: [0.02–1.41]; p = 0.003); and superior performance in ULP (effect size: 1.21; 95% CI: [0.48–1.95]; p = 0.0007) and SJ (effect size: 1.29; 95% CI: [0.55–2.03]; p = 0.0004), when compared with prepubertal individuals.When categorizing the sample according to testosterone levels, the two classes discriminated were: testosterone below 100 ng/dL, and testosterone above 100 ng/dL. Thus, the group with testosterone <100 ng/dL had 13 subjects and the testosterone group >100 ng/dL obtained 24 members. The comparisons contained in Table 6 show that the subjects with testosterone >100 ng/dL were statistically higher in relation to ULP and SJ and had a lower body fat percentage compared with individuals with testosterone <100 ng/dL.The aim of the present study was to relate maturation and hormonal markers to neuromuscular performance, as well as to verify whether young athletes with different levels of testosterone show differences in muscle strength. The main results of the present study were: (1) The hormonal markers showed significant relationships with neuromuscular performance, especially with lower limbs and body morphology. (2) The biological maturation markers were significantly related to body morphology and the performance of ULP, SJ and ULS. (3) The hormonal and biological maturation markers were significant in foreseeing the performance of ULP, SJ, BScD and the levels of lean mass. (4) Subjects categorized as pubescent showed higher concentrations of testosterone and lean mass and better performance of ULP and SJ. (5) Young athletes with TRT > 100 (ng/dL) showed superior performance of ULP and SJ and had a lower percentage of body fat.The present study found that in young males, hormonal markers point to significant relationships with neuromuscular performance and with levels of lean body mass. Crewther et al. [37] corroborate the findings of the present research when they identified that the levels of steroid hormones in young male weightlifting athletes are related to body mass and show a significant relationship with neuromuscular performance (p < 0.05). In a similar perspective, Chin et al. [38] showed in a previous study that the testosterone levels of male subjects indicate significant relationships with the higher levels of upper limb strength and lean body mass. In this sense, Aslam [39] points out that during puberty there is a peak of hormonal levels such as testosterone, estradiol, and growth hormone, which can significantly influence the morphological structure and the acquisition of muscle strength in young people of both genders.Thus, the findings of the present study also showed significant relationships between biological maturation markers and levels of lean body mass and with the neuromuscular performance of upper and lower limbs. However, the maturation markers did not show significant relationships with the hormonal markers analyzed. Our results differ from the data obtained in the study by Pinto et al. [13], where the authors found significant relationships between skeletal maturation and testosterone in young males (p = 0.04). These differences can be justified due to the characteristics of the samples from the two studies (i.e., exact biological age, nutritional factors) and due to the different statistical treatment in relation to the control of laboratory tests. On the other hand, as in the present study, these authors found significant relationships between maturation and neuromuscular performance of the upper and lower limbs of the young people analyzed (p < 0.001). In addition, Dantas et al. [40] identified significant relationships between biological maturation and neuromuscular performance of the upper and lower limbs in young Brazilian rowers of both sexes (p < 0.05).The findings of the present study show that the levels of lean mass and the performance of ULP and SJ can be foreshadowed by hormonal and maturation markers. Our group has already demonstrated that through linear regression analyses and artificial neural networks, the biological maturation markers of female subjects show about a 50% probability of predicting the neuromotor performance of upper and lower limbs [14], while hormonal markers discriminated a potential 95% chance of predicting upper limb neuromotor performance [14]. The approach highlights that although environmental factors may significantly interfere with hormonal levels (i.e., food, sleep quality, stress levels), there is a significant association between the stages of puberty and hormonal levels in children and teenagers [41,42,43].In this context, the results of the present study show that young people with testosterone levels above 100 (ng/dL) are superior in the neuromuscular performance of upper and lower limbs and point to lower body adiposity when compared with their peers with testosterone levels less than 100 (ng/dL). In this perspective, it is emphasized that the findings of the present study indicate that testosterone also increases with growth and development. Therefore, the cut-off value of “100 ng / dL” used in this study is experimental and needs more research to be analyzed in depth. Wirth [44] states that testosterone is related to the acquisition of muscle strength during different periods of human life, including the puberty phase. In this sense, anabolism of skeletal muscle tissue occurs through mechanisms of protein synthesis triggered by testosterone concentrations in the body [45,46].Moreover, it was observed in the present study that young pubertal boys show higher concentrations of testosterone, greater lean mass and superior neuromuscular performance of upper and lower limbs than the prepubertals. Our results corroborate the study by Distefano et al. [47], where the authors identified that young athletes with advanced puberty had greater knee extension strength when compared with delayed maturation youth. In addition, Pinto et al. [13] found that young people with advanced biological maturation have superior upper and lower limb strength than young people with delayed maturation.It should be noted that the present study also demonstrated that chronological age is not a safe marker of biological age, when observing that the sample used was in the age group between 10 and 12 years old (age: 11.3 ± 0.96) and the maturation stage typical for this age group is the prepubertal [10,22,41]. However, it is known that environmental factors such as nutrition, quality of sleep, physical and cognitive stimuli interact with the puberty process. Therefore, the stimuli provided can interact with the biological rhythm which can cause the delay or the advancement of puberty in children and adolescents of both sexes [1,10,22,41].However, despite the relevance of the results, the present study has some limitations: (i) its design is observational, which makes it impossible to establish a cause and effect relationship. (ii) The methodological procedures did not control for the nutritional history of the analyzed subjects, which may have interfered with hormonal concentrations. (iii) Biological maturation markers were measured using predictive formulas based on anthropometry. This fact can lead to a divergence of results in comparison to the analysis of biological maturation using the gold standard (hand and wrist X-ray). (iv) The tests performed may have been influenced by anatomical factors (i.e., length of the upper and lower limbs), biological factors (i.e., level of cerebral excitation and the composition of types of muscle fibers) and an extra level of physical activity (it was not controlled if subjects practiced additional sports).Biological maturation and hormone levels may be related to lean mass and neuromuscular performance. Young athletes with higher testosterone levels display less fatty tissue and superior muscle strength in the upper and lower limbs compared with the others. Thus, the present study shows the influence that biological maturation and endocrine markers can have on muscle strength, providing important information for coaches to select sports talents as well as facilitating the application in daily training routines.Conceptualization, P.F.d.A.-N., V.C.M.P., A.B.-C., F.J.A. and B.G.d.A.T.C.; Data curation, P.F.d.A.-N. and V.C.M.P.; Formal analysis, P.F.d.A.-N., D.G.d.M., T.d.M.C., F.J.A. and B.G.d.A.T.C.; Investigation, P.F.d.A.-N., V.C.M.P. and T.d.M.C.; Methodology, P.F.d.A.-N., D.G.d.M., V.C.M.P., L.F.d.S., F.J.A. and B.G.d.A.T.C.; Project administration, T.d.M.C., A.B.-C. and B.G.d.A.T.C.; Supervision, P.M.S.D., F.J.A. and B.G.d.A.T.C.; Validation, B.G.d.A.T.C.; Visualization, L.F.d.S. and B.G.d.A.T.C.; Writing—original draft, P.F.d.A.-N., P.M.S.D., L.F.d.S. and B.G.d.A.T.C.; Writing—review & editing, D.G.d.M., A.B.-C. and B.G.d.A.T.C. All authors have read and agreed to the published version of the manuscript.This research received no external funding.For your support and encouragement for the development of this academic article, we thank the Federal University of Rio Grande do Norte (UFRN), the Physical Activity and Health (AFISA) research base, the Child and Adolescent Maturation Research Group (GEPMAC). The National Council for Scientific Development (CNPQ) and the Higher Education Personnel Improvement Coordination (CAPES).The authors declare no conflict of interest. Study design. a = Study information for volunteers and their respective guardians. b = Biochemical tests. c = Body composition test. d = Collecting anthropometric measurements. e = Upper limb power test. f = Upper limb speed test. g = Lower limb power test. h = Body speed with change of direction test.Comparison according to the stage of sexual maturation. * Statistically significant. TRT = Testosterone. EST = Estradiol. GH = Growth Hormone. ULP = Upper limbs power. SJ = Squat Jump. ULS = Upper limb speed. BScD = Body speed with change of direction.Sample characterization.BScD = Body speed with change of directionCorrelations of hormonal levels with the study variables.ULP = Upper limbs power. SJ = Squat Jump. ULS = Upper limb speed. BScD’s = Body speed with change of direction. TRT = Testosterone. EST = Estradiol. GH = Growth Hormone. * Statistically significant.Correlations of biological maturation with the study variables.ULP = Upper limbs power. SJ = Squat Jump. ULS = Upper limb speed. BScD = Body speed with change of direction. TRT = Testosterone. EST = Estradiol. GH = Growth Hormone. * Statistically significant.Linear regression of hormonal levels and biological maturation with the study variables.ULP = Upper limbs power. SJ = Squat Jump. ULS = Upper limb speed. BScD = Body speed with change of direction. TRT = Testosterone. EST = Estradiol. GH = Growth Hormone. * Statistically significant.Analysis of perceptron neural networks for forecasting possibilities.ULP = Upper limbs power. SJ = Squat Jump. ULS = Upper limb speed. BScD = Body speed with change of direction. TRT = Testosterone. EST = Estradiol. GH = Growth Hormone. 
+			U = Domain of the Gaussian function of binary activation of the neural network. % E = Percentage of total neural network learning error. | P% = Percentage of the probability of the forecast is correct. * = True forecast.Comparisons between subjects with testosterone levels < 100 (ng/dL) and > 100 (ng/dL).ULP = Upper limbs power. SJ = Squat Jump. ULS = Upper limb speed. BScD = Body speed with change of direction. TRT = Testosterone. EST = Estradiol. GH = Growth Hormone. * Statistically significant. ES—effect size

Med-MDPI/ijerph_5/ijerph-17-16-05638.txt

@@ -0,0 +1,4 @@
+Investigating initial behavioral changes caused by irradiation of animals might provide important information to aid understanding of early health effects of radiation exposure and clinical features of radiation injury. Although previous studies in rodents suggested that radiation exposure leads to reduced activity, detailed properties of the effects were unrevealed due to a lack of proper statistical analysis, which is needed to better elucidate details of changes in locomotor activity. Ten-week-old male Wistar rats were subjected to single point external whole-body irradiation with 60Co gamma rays at 0, 2.0, 3.5, and 5.0 Gy (four rats per group). Infrared sensors were used to continuously record the locomotor activity of each rat. The cumulative number of movements during the night was defined as “activity” for each day. A non-linear mixed effects model accounting for individual differences and daily fluctuation of activity was applied to analyze the rats’ longitudinal locomotor data. Our statistical method revealed characteristics of the changes in locomotor activity after radiation exposure, showing that (1) reduction in activity occurred immediately—and in a dose-dependent manner—after irradiation and (2) recovery to pre-irradiation levels required almost one week, with the same recovery rate in each dose group.In humans, one of the earliest effects of radiation exposure to the whole body or to a large portion of the whole body is a prodromal period of nonspecific signs and symptoms such as nausea, emesis, fatigue, fever, and anorexia [1,2]. The prodromal syndrome is generally mild or absent at total body doses of 1 Gy or less and occurs from minutes to days following exposure [3,4,5]. However, it is unclear to what extent these symptoms are psychogenic versus radiation-induced. Therefore, the relationship between initial symptoms and radiation dose is not well understood. Early effects of irradiation have been studied in regard to radiation therapy. In a detailed study of the incidence and severity of side effects during the course of radiation therapy, fatigue was the most prevalent and the most severe symptom reported by patients [6]. With fractionated doses of radiation for cancer treatment, radiation-induced fatigue sets in within a few days after start of treatment and decreases after treatment completion [7]. Although the underlying mechanisms of fatigue have been studied under several disease conditions, an understanding of the etiology, mechanisms, and risk factors of radiation-induced fatigue remains elusive, and this symptom remains poorly managed [8,9,10]. Investigating initial radiation-related behavioral changes by using animals might provide important information to aid understanding of the health effects of radiation exposure and clinical features of radiation injury.In animals, there have been many studies of radiation-induced behavioral effects, and performance decrement after irradiation has been noted in several reports. A sub-lethal dose of gamma radiation suppressed aggressive behavior in male mice [11], a lethal dose of gamma radiation suppressed locomotor activity in mice [12], and a sub-lethal dose of X-irradiation suppressed volitional activity in rats [13]. Landauer (2002) provided a review of expected performance decrement after radiation exposure [14]. These reports showed that ionizing radiation temporarily suppresses animals’ behavior, but that the effect does not continue for a long period. York et al. reported that, 6 h after gamma irradiation with 50 or 200 cGy, spontaneous locomotor activity in mice was 35% or 36% lower, respectively, than in sham irradiated controls, and that their activity recovered to sham irradiated level 12 h after irradiation [15].Although many animal behavioral experiments have a time-dependent data structure with variation among individuals, analyses have typically been performed only at individual time points with no parameterization of the trend in activity over time. Therefore, quantitative analyses have not been made directly on the chronological features. To obtain more detailed and accurate information from data obtained in animal behavior experiments with time-dependent structure and individual variability, application of statistical theory would suggest that analysis based on a mixed effects model [16,17] is both appropriate and effective.The reason for the experiments was to investigate the early health effects of radiation exposure. Then, statistical models were used to examine in detail the changes over time in locomotor activity of rats immediately after external irradiation with 60Co gamma rays. Specifically, we aimed to assess the time when reduction of locomotor activity begins, the time when locomotor activity recovers to pre-irradiation level, the dose dependency of the degree of reduction in locomotor activity, and the dose dependency of the rate of recovery. There are individual differences in animal behavior that cannot be ignored, even if the animal type, gender, and weight are uniform. In addition, when animals are observed over a long period of time, it is expected that common changes in behavior will occur due to indoor conditions such as temperature, humidity, and noise, which can change daily, and it is necessary to adjust for these sources of variation.The experiment was approved by the Animal Experiment Committee of Semey Medical University, Republic of Kazakhstan, and was conducted in accordance with the Institutional Guide for Animal Care and Use. Ten one-week-old male Wistar rats were purchased from the Kazakh Scientific Center of Quarantine and Zoonotic Diseases, Almaty, Kazakhstan and allowed free access to a basal diet and tap water. Animal rooms were maintained at 19–22 °C with relative humidity 30–70% and a 12 h light cycle. Body weights were measured twice a week during the experiment. At 11 weeks of age, the rats were randomly divided into four groups: control (4 rats) and three irradiated groups (4 rats/group). Each irradiated group received 2, 3.5, or 5.0 Gy of whole body gamma irradiation. Controls were handled with all conditions the same as with the other groups, except that they were not irradiated (dose 0 Gy). The LD50(30) for this strain of Wistar rats is 7 Gy with cobalt-60 radiation [18]. Irradiation was performed with a Teragam K-2 unit (UJP Praha, Praha-Zbraslav, Czech Republic) at the Regional Oncology Dispensary of Semey. Rats were irradiated at 1 m distance from the 60Co source at a dose rate of 2.6 Gy/min. Half of the radiation dose was administered from the top and the other half was administered from the bottom. A radiophotoluminescence glass dosimeter, GD-302M [Chiyoda Technol Co., Tokyo, Japan], was used for measuring the doses. Locomotor activities (hereafter abbreviated as “activities”) of the rats were measured with infra-red sensors (Model NS-AS01; Neuroscience, Inc., Tokyo, Japan) placed 16 cm above the open-top cages (26.5 × 43 × 14.5 cm). Numbers of movements were counted on the basis of change in the strength of infra-red rays emitted from the animals. The rats were placed in separate cages, each outfitted with a sensor, and movements were continuously counted by a computerized analysis system (16 channel Multi-digital Counter System [MDC] and DAS System software, Neuroscience, Inc. Tokyo, Japan). Measurements were started 3 days before irradiation and continued for 20 days after irradiation. All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. The animal experiment was approved by the Animal Experiment Committee of Semey Medical University, Republic of Kazakhstan (Protocol No 5 dated 16.04.2014), and conducted in accordance with the Institutional Guide for Animal Care and Use.Because rats are nocturnal animals [19], the cumulative number of movements was recorded during the period between 18:00 and 06:00; the number of movements so recorded was defined as activity of a rat in one day. As shown in Figure 1, rates of increase in cumulative movements (slopes) were steeper during nighttime (18:00–05:59) than during daytime (06:00–17:59); i.e., the rats were more active at night, as expected. This suggests that the activity defined in this study represents the nocturnal characteristic of rats and it shows that the measure has relevance as an indicator of a rat’s activity.Logarithmic values of daily activity of each rat as a function of elapsed time relative to day of irradiation are shown for each group in Figure 2. An acute decrease in activity after irradiation followed by quick recovery to the pre-irradiation level can be seen in every exposed group, whereas no such change or trend was observed in the control group. There was also large inter-animal variation with daily fluctuation in activity. Therefore, we assumed a non-linear mixed effects model (NLMM) [16,17] that takes into account the dose dependency of the decrease in activity, the dose dependency of the recovery rate, individual differences among animals, and daily fluctuations within individual animals. For comparison, we fit a simple non-linear regression model (NLRM) in which individual differences and daily fluctuations were not taken into account.Let yit be the log transformed observed activity of rat i at time t in days since irradiation with dose Di  (t=−3,…,20 ;  i=1,…,16), where “t=0” indicates day of irradiation. We assume the model:yit=f(t|Di,θ)+δi+ηt+εi t
+          f(t|Di, θ)=ξ0+ξ1t+ξ2t2−(β1Di+β2Di2)⋅exp[{−ω1⋅e−ω2(Di−D0)}t]⋅h(t)
+          (1)δi∼N(0, ψ2), ηt∼N(0, φ2), εit∼N(0, σ2),t=−3,−2,…,20, i=1,…,  16,        
+          where θ=(ξ0, ξ1 , ξ2, β1, β2, ω1,ω2) denotes unknown parameters for fixed effects to be estimated. The term ξ0+ξ1t+ξ2t2 expresses the time dependency of activities without radiation exposure. The term β1Di+β2Di2 expresses whether the dose effect in the initial decrease is linear (β2=0) or quadratic (β2≠0), and the term −ω1⋅e−ω2(Di−D0) denotes whether the recovery rate depends on dose (��2≠0) or not (ω2=0). D0 denotes a fixed pre-assigned dose value for covariate centering (in this study 2.75 Gy is adopted),  Δ=(ψ2,   φ2,σ2) are unknown dispersion parameters to be estimated, and the terms δi, ηt, and εit represent independent random effects due to individual variability, daily fluctuation, and measurement error, respectively. The function  h(t):h(t)=0 (t<0),  h(t)=1 (t≥0) denotes the Heaviside function of t to indicate pre- and post-irradiation dichotomy. Let y=(y1′,…,y16′)′,  yi=(yi,−3,…,yi,20)′, i=1,…,16. It follows from Model (1) that y has a multivariate normal distribution with mean μ(θ)=(μ1(θ)′,…,μ16(θ)′)′, μi(θ)=f(t|Di,θ), t=(−3,−2,…,20)′, i=1,…,16, and variance-covariance matrix Ω(Δ)=I16⊗(ρ2J41+σ2I41)+J16⊗ψ2I41, where Im denotes an m-dimensional unit matrix, and Jm=1m⊗1m′. Then the likelihood function of (θ,Δ) can be expressed as L(θ,Δ)=1(2π)8|Ω(Δ)|exp(−12{y−μ(θ)}′Ω(Δ)−1{y−μ(θ)}). Therefore, the maximum likelihood estimates of (θ,Δ), denoted by (θ^,Δ^), are obtained by minimizing the quantity Q(θ,Δ)=log(|Ω(Δ)|)+{y−μ(θ)}′Ω(Δ)−1{y−μ(θ)}+16×log(2π). When ψ2=φ2=0, Model (1) reduces to an ordinary non-linear regression model (NLRM). The unknown parameters were estimated by using an algorithm for optimization with the limited-memory version of the Broyden–Fletcher–Goldfarb–Shanno method [20] to maximize the likelihood derived from the model (1), and the AIC (Akaike Information Criterion) [21] and BIC (Bayesian information criterion) [22,23] were calculated. The function ‘optim’ in the R software ver. 3.5.1 was used for carrying out numerical analyses. Maximum likelihood (ML) or restricted maximum likelihood (REML) [24] estimates of the parameters in the linear mixed-effects models can be computed with the “lmer” function in the “lme4” package for R [25]. In this study, the ML method was used to compare the goodness-of-fit of models with the AIC criterion. Estimation results were almost the same with both methods. Regression analysis was first performed with all parameters of the NLMM (full NLMM), then model selection was applied by choosing the smallest AIC to determine the optimal NLMM (optimal NLMM). The full NLRM and optimal NLRM were defined in the same way. Estimates of fixed-effect parameters and their 95% confidence intervals under the full and optimal NLMM are shown in Table 1a and Table 1b respectively; those under the full and optimal NLRM are shown in Table 2a and Table 2b, respectively.In the optimal NLMM, variances of the random effects due to individual differences, daily variation, and measurement error were 0.0018, 0.0019, and 0.0015, which account for 35%, 36%, and 29% of the total variance, respectively. Predictions of individual differences (δ^1,δ^2,⋯,δ^16) and those of daily fluctuation (η^−3, η^−2,⋯,η^20) were obtained by calculating posterior means. The predictions δ^i in each of the four groups (control group and three irradiated groups) and the predictions η^t by day are shown in panels (a) and (b) of Figure 3, respectively. Residuals in the optimal NLMM and in the optimal NLRM are given by  yit−f^(t|Di,θ)−δ^i−η^t and yit−f^(t|Di,θ) , respectively. The standard deviations of residual errors in the optimal NLMM and optimal NLRM were 0.038 and 0.071, respectively. The distributions of residuals in the NLMM and NLRM are shown in Figure 4. There is a large difference between the AICs of the optimal NLMM and the optimal NLRM, which were −1271.80 and −930.25, respectively (See Table 1b and Table 2b). The measurement error variances of the NLMM and NLRM were 0.0015 and 0.0058 (See Table 1b and Table 2b). Therefore the fit of the NLMM was preferable to that of the NLRM in terms of prediction and accuracy. The estimated time dependency of activity in each group under the optimal NLMM is shown in Figure 5.In each of the irradiated groups, activity decreased immediately after irradiation but recovered to the pre-irradiation level within a few days with a common recovery rate irrespective of dose. One of the advantages of using the more complex NLMM structure, as demonstrated in this paper, is that a second-order dose dependency could be detected in the initial decrease, which was not found with the NLRM (which estimated a linear dependency). Estimated magnitudes of initial decreases at t=0 by dose group and their 95% confidence intervals in the optimal NLMM and those in the optimal NLRM are shown in Figure 6. The plots of predictions of individual differences δ^i by dose group (Figure 3a) show that the assumption of homoscedasticity for distributions of individual difference between the four dose groups seems to be satisfied. This means that the random assignment of rats to the four groups was effective in terms of individual differences. The plots of predictions of time-dependent daily fluctuation η^t (Figure 3b) show that the assumption of independency of each of the random variables ηt seems to be satisfied. The Durbin Watson statistic [26] for η^t was 2.33 (p-value 0.902), indicating that no strong autocorrelation is observed in daily fluctuation. Because acute changes were the focus in this experiment, longer observation was not performed, but it is necessary to investigate late effects. The irradiation was a single and sub-lethal dose, so it is considered that damage was acute, disappearing in a short period of time, and resilience to allow recovery from the damage was not affected by irradiation. The effects of chronic low dose exposure remain as future issues to be addressed. As one important example of the need for assessing effects of chronic exposure, a giant earthquake of magnitude M9 struck East Japan on 11 March 2011. Subsequently a ‘tsunami’ engulfed the Fukushima Daiichi Nuclear Power Plant (FDNPP). As a result, FDNPP reactors 1−3 suffered meltdown and significant amounts of radioactive materials were released into the environment [27]. The dose to the public is estimated to be low [28], but many Japanese people are worried about the resulting health effects of chronic low dose exposure. In the present study, the effects of irradiation on the behavior of rats were investigated efficiently, despite a small number of animals with large individual differences. This was achieved by using a statistical method that accounts for inter-animal differences and daily fluctuation in activity—a non-linear mixed model fit to repeated measurements. With such an efficient approach, we were able to demonstrate a temporary, but dose-dependent, decrease in activity following irradiation and a dose-independent common recovery rate. The statistical framework for analyzing longitudinal locomotor data in this study should be generally applicable to other repeated measurement data with a similar structure. Conceptualization, M.H., K.O., M.O., and N.F.; methodology, K.O., M.O., and N.F.; formal analysis, K.O., and M.O.; investigation, A.S., N.C., and H.S.; data curation K.O.; writing—original draft preparation, K.O., M.O., and N.F.; writing—review and editing, K.O., M.O., N.F., and M.H.; supervision, N.C. and M.H.; project administration, T.R., N.K., and M.H.; funding acquisition, M.H. All authors have read and agreed to the published version of the manuscript. This work was supported by JSPS KAKENHI Grant Numbers 26257501 (April 2014–March 2018) and 19H01149 (April 2019–March 2023).The authors would like to thank the staffs of Semey Medical University for technical assistance with the experiments. The authors declare no conflict of interest.Cumulative number of movements of each of the 16 rats over a 36-h period.Daily activity of each of the four rats belonging to the four groups. The vertical axis shows the logarithm of daily activity (number of nocturnal movements) and the horizontal axis shows elapsed time in days relative to the day of irradiation (indicated by arrows): (a) the control group, (b) 2.0 Gy group, (c) 3.5 Gy group, and (d) 5.0 Gy group.Predictions of random values. Predictions of random values by individual δ^i by group are shown in panel (a) and predictions of random values by day η^t are shown in panel (b).Parallel boxplots of residual errors in the non-linear mixed model (NLMM) and ordinary non-linear regression model (NLRM).Estimated mean trends of daily locomotor activity in rats by dose group under the optimal NLMM.Fitted dose-response curves from the optimal NLMM and the optimal NLRM. The estimated magnitudes of decrease at t=0 by dose group and their 95% confidence intervals and fitted dose-response curves with dotted line from the NLMM and the NLRM are shown in panels (a) and (b), respectively. Cross marks show observed data of individual rats. The fitted dose-response curve from the optimal NLMM was a downward convex quadratic curve.Estimated fixed effects parameters in the full non-linear mixed effects model (NLMM) (a) and those in the optimal NLMM (b).(a): **: p<0.01, *: 0.01≤p<0.05. Estimated random effect parameters: (ψ2^, φ2^, σ2^)=(0.0018, 0.0019, 0.0015). Log-likelihood: 643.47, AIC: −1266.94, BIC: −1227.44.(b): **: p<0.01, *: 0.01≤p<0.05. Estimated random effect parameters: (ψ2^, φ2^, σ2^)=(0.0018, 0.0019, 0.0015). Log-likelihood: 642.90, AIC: −1271.80, BIC: −1244.15.Estimated fixed effect parameters in the full non-linear regression model (NLRM) (a) and those in the optimal NLRM (b).(a): **: p<0.01, *: 0.01≤p<0.05. Estimated residual variance: σ2^=0.00502. Log-likelihood: 744.091, AIC: −928.17, BIC: −883.56.(b): **: p<0.01, *: 0.01≤p<0.05. Estimated residual variance: σ2^=0.0058. Log-likelihood: 742.12, AIC: −930.25, BIC: −899.02.

Med-MDPI/ijerph_5/ijerph-17-16-05639.txt

@@ -0,0 +1 @@
+Microplastics (MPs) have generated worldwide attention due to their global distribution in the environment, and their potential harmful effects on human and animal health. To analyze MPs-related scientific publications from a global point of view, we created a bibliometric profile, by searching the Web of Science Core Collection database for the topic “microplastic* or (micro near/1 plastic*)”, in publications dated from 2004 to 2019. The results revealed an increasing trend in publication output, and identified contributions of different countries and their collaborations, as well as influential authors and productive journals in the field of MPs research. Using co-citation network analysis in VOSviewer, we mined cited references for knowledge bases about analytical methods, potential sources and spatial distributions of MPs, the impacts of MPs on organisms, and the interaction of MPs with contaminants, as well as microorganisms. We also identified four global hotspots for MPs related research, using author keywords co-occurrence network analysis of all extracted publications, as well as Essential Science Indicators highly cited papers from Clarivate Analytics. Results of this study provide a valuable reference for ongoing MPs-related research, which may be of intrigue and awesome noteworthiness for relevant researchers.Plastics used in our daily life and in a wide range of manufacturing processes provide numerous societal benefits, due to their lightweight, durable, and economic nature [1]. However, plastics are resistant to aging, and their refractory degradation makes plastic waste a serious environmental issue [2,3,4]. Microplastics (MPs), which are smaller items of plastic litter, are of increasing concern due to their ubiquitous global distribution in aquatic environments [5,6,7], and their close interactions with biota [8]. Although no universal definition of MP size exists, a diameter smaller than 5 mm is commonly accepted [9]. Examples of MPs include resin pellets, microbeads used for cosmetics or associated with industrial spillages (primary source) [10,11,12], or pieces broken off of larger plastic litter by ultraviolet radiation, oxidation, or mechanical abrasion (secondary source) [13]. Release of synthetic fibers by textile washing is another potential source of MPs [14].As there has been increasing concern about MPs research, scholars have reviewed literatures in this domain covering different aspects. Initially, reviews of MPs research focused mainly on the marine environment. For example, Cole et al. discussed the sources and transfer of MPs into the marine environment, and assessed the spatial and temporal distribution of MPs in the worldwide marine environment [3], and they concluded that the fate of these MPs was still elusive. Wright et al. investigated the impacts of MPs on marine invertebrates [8]. Later, Horton et al. critically reviewed the presence, behavior, and fate of MPs in terrestrial environments, by evaluating studies of the extent of MPs pollution in freshwater, treated water sources, and even agriculture soil [15]. Recently, the biological effects of MPs have emerged as areas of interest. Researchers have summarized the potential health effects of MPs present in the food chain [16], and emphasized the interaction between MPs and microorganisms [17]. The authors of these reviews amassed, summarized, and extended the MPs-related research based on their long-term research experiences. To date, studies of the evolution of MPs-related scientific research from a global point of view over time were still insufficient. Bibliometric analysis, which takes advantage of bibliometric theory using mathematical and statistical approaches, is a method that can be used to address this knowledge gap. It has been applied to analyze pertinent literatures in various research fields [18,19,20], including environment-related fields [21]. With regard to MPs research, Ivar do Sul et al. summarized the common denominator between MPs and microbiology, using the bibliometric approach [22], and Barboza et al. evaluated research trends and future perspectives on MPs in the marine environment for the period 2004–2014, using the cross-disciplinary quantitative analysis method [23]. As MPs research has increased substantially since 2011, Zhang et al. conducted an in-depth statistical analysis of global MPs research, using the number of publications as a primary metric for productivity of countries, institution, authors and journals [24]. An up-to-date comprehensive review of the scientific literature, which interprets the influence and importance of different countries, authors and journals, as well as co-occurrence keywords analysis initiated in both extracted literatures from the database and Essential Science Indicators (ESI) highly cited papers from Clarivate Analytics; this is still needed to trace global research hotspots in MPs research.In this study, we conducted an integrated bibliometric analysis of the literatures on MPs research published from 2004 to 2019. The initial time was set as 2004, because that year, Thompson et al. [25] coined the term “microplastics (MPs)” to define the smaller plastic litter. We used the analysis to identify influential countries, international collaborations, contributing authors, preferred journals, a knowledge base of MPs studies, and research hotspots. The results of our analysis provide a valuable picture of the status of current global MP research, and help illuminate the next steps for future studies.The Web of Science Core Collection (WoSCC), which generates standardized and high-quality academic publication information, is used extensively for the bibliometric examination of the evolution of scientific issues [18,26,27]. On March 12, 2020, all original data were extracted from the online version of the WoSCC database (indexes: Science Citation Index-Expanded and Conference Proceeding Citation Index), using the TOPIC “microplastic* or (micro near/1 plastic*)” for the years 2004 to 2019. ESI highly cited individuals along with the number of their MPs-related publications and ESI highly cited papers were collected by the same TOPIC from Clarivate Analytics on the same day.Initially, 3246 publications (after removal of duplications) were extracted using our data searching strategy, including some articles related to material science studies. The latter publications could not be removed simply by excluding some keywords in the data search (e.g., by using “not ‘micro-plastic deformation behavior’” or “not ‘micro-plasticity’”), because some studies of biodegradable polymers relate to both material science (composites modification) and our study objective (safe for environment), such as [28]. Thus, we conducted content analyses of titles and abstracts of all 3246 publications, and sometimes the full manuscripts were evaluated to exclude irrelevant publications. Ultimately, 2637 publications written in English or with an English abstract remained after the manual screening of four types of documents (articles, reviews, proceeding papers and book chapters). Because these types of documents contained novel concepts, none of them were excluded from our analysis, thus, these 2637 publications were all included in the bibliometric analysis. Moreover, a total of 395 ESI highly cited MPs-related papers were extracted from Clarivate Analytics.The basic bibliometric analysis method used a range of indicators to identify distributed characteristics and structural patterns of the general bibliographic data for MPs ongoing work. For example, the year-wise distribution of research output demonstrated the developing trend of increasing work in the MPs discipline. The contribution of an individual country/academic researcher in the MPs scientific research field was ranked by how many times their publication was cited by others (non-self-citation, NSC), and other data recorded included their total number of publications (TNP), sum of times cited (STC), non-self-citation ratio (NSCR), number of publications cited by more than 100 and 50 times, and number of ESI highly cited publications. Preferred journals were identified as those that delivered academic articles and contributed to the development of the research field [29]. Both the journal impact factor (IF) and quartiles in relevant categories were derived from Journal Citation Report (JCR) 2018, and used to explore the publishing journal’s influence in the MPs field and their interdisciplinary research areas. All bibliographic data were analyzed using Microsoft Excel 2016, and figures were created using GraphPad Prism (version 7.04, GraphPad Software Inc., San Diego, CA, USA) and VOSviewer software (version 1.6.9, Centre for Science and Technology Studies, Leiden University, Leiden, The Netherlands). We used the results of quantitative analysis of the evolution of literature, as well as the bibliometric indicators, to present a general informative overview of MPs research during the study time period.VOSviewer is a free software tool based on the Java environment, that is suitable for constructing complex networks using large-scale data. Therefore, we used VOSviewer software (version 1.6.9) to conduct an in-depth network analysis to visualize the connections between various MPs-related items, and to explain their network structure.Countries co-authorship network: We conducted the co-authorship analysis to identify collaboration networks among different countries in the MPs research field. The nodes represented countries contributing to MPs research, and the links between items implied cooperative relationships. The size of the node increased as the number of articles published by an individual country increased. The value of the links indicated the number of times a given country shared co-authorship with others. The strength of the link increased as the number of co-authorships increased.Cited reference co-citation network: Analyzing a knowledge base in a certain research field can be conducted by co-citation network analysis for the cited references [18]. In our co-citation analysis, the nodes represented scientific references. The node size represented the number of times a reference was cited. The distance between two references indicated the correlation of the articles according to co-citation links, based on the assumption that more frequently co-cited references exhibited greater co-citation strength.Author keywords co-occurrence network: The keywords that authors provided for their articles about MPs research represented their academic viewpoints. Thus, our author keywords co-occurrence analysis identified important terms in the MPs academic, as well as the research hotspots in the MPs discipline. The nodes represented high-frequency author keywords, and the size of an individual node represented how many times that keyword occurred. The link strength between two nodes indicated the number of articles in which two keywords occurred together.The year-wise distribution of publication output revealed the progress of MPs research over time (Figure 1a). The number of publications related to MPs fluctuated slightly from 2004 to 2008, and the annual publications were all less than 100 until 2014. Obvious growth of MPs research began in 2014, when the first United Nations Environment Assembly of the United Nations Environment Programme (UNEP) issued the resolution UNEP/EA.1/L.8, which emphasized critical activities to address marine plastic debris and MPs challenges. Governments worldwide began a shared commitment to addressing MPs problems and to conducting the systematic research of many aspects of MPs [30,31,32,33]. The cumulative number of annual publications since 2004 follows an exponential model (Figure 1b), and the simulation results suggest that publications about MPs issues might increase to 1703 in 2020. On May 8, 2020, we collected data using the same TOPIC from the WoSCC database, for the time period spanning January 1, 2020 to April 30, 2020, and identified 442 relevant publications. This value was less than the expected value for one-third of the year 2020 (567), which may be a result of the significant disruption that is being caused by the COVID-19 pandemic.According to the author address information in the 2637 publications, 104 countries made contributions to MPs research during the study time period. Table 1 lists the top 10 countries by their NSC times of publications in MPs research field, as well as TNP (with ranking), STC (with ranking), NSCR, and the number of publications cited more than 100 and 50 times. Developed countries, including six European countries, two North American countries, and one Oceania country, along with a developing country (China) occupied leading positions in MPs research. England led the NSC index and was second for STC, which indicated that it produced high quality MPs research. The USA also performed well in the areas of research depth and influence, as it ranked first for STC and had 62 publications that were cited more than 100 times, and 107 publications that were cited more than 50 times. China had the most publications (459), but ranked seventh for NSC. Canada, the other five Europe countries, and Australia also performed well. The 10 countries listed in Table 1 contributed 92.8% of all publications, and the high NSCR values (89.38% on average) indicated that these countries had great external influence not only in numbers of studies but also in the quality of MPs-related publications.According to the statistics, 8191 authors (without debugging repetitions of authors’ names) have contributed to the increasing scientific knowledge about MPs. Table 2 lists the top 10 influential authors, ranked by the criteria of NSC on MPs issues, as well as their institution (the latest one), country, TNP, STC, and the number of ESI highly cited papers (NEHC). These data were manually debugged to improve the quality of analysis, as a single author may have different forms of abbreviations but with separately counted articles. Half of these influential authors are from European countries, which is in agreement with the known active participation of European countries in MPs-related research. Thompson, R.C. from England was the most productive and influential author, as his 48 publications were cited 11,617 times, and half of his publications were listed as ESI highly cited papers. Galloway, T.S. and Cole, M., from the University of Exeter, England, ranked second and fourth, respectively, with the ratio NEHC/TNP > 55%. Shi, H.H., who ranked ninth, was the most active MPs researcher in China, and 50% of his publications were included as ESI highly cited papers. Galgani, F.; Koelmans, A.A.; Thiel, M.; Rochman, C.M.; Shim, W.J. and Costa, M.F. also performed well with their MPs-related research, and contributed information about the MPs distribution in their regional marine environments, MPs analytical methods, hazardous chemical sorption of MPs, and release of MPs. These scholars were all listed as ESI highly cited researchers in the Ecology/Environment field, which indicated that their articles had significant influence on subsequent research.The 2637 publications were retrieved from 399 journals. Among them, most journals (384, 96.2%) published fewer than 20 articles about MPs. Table 3 shows the top 15 most productive journals in which more than 65% of publications related to MPs were published during the period 2004-2019. These journals were classified in six categories, and all placed in higher quartiles in category (Q1/Q2) according the 2018 JCR report. Ten of the journals were grouped in the Environmental Sciences category and three were grouped in the Marine and Freshwater Biology category, which indicated that MPs in the aquatic environment was the research hotspot. Mar Pollut Bull published the most articles (536) and had the highest STC and NSC, but its NSCR was lower than that of the other 14 journals. Environ Pollut, with 327 articles, ranked second. Water Res ranked eighth for TNP, but had the highest IF (7.913) among the ten Environmental Sciences journals. Sci Rep-UK and Plos One, were grouped in the Multidisciplinary Sciences category, and Trac-Trend Anal Chem and Analmethods-UK were grouped in the Chemistry, Analytical category. Articles in Environ Toxicol Chem and Ecotox Environ Safe, which were classified in the Toxicology category, focused on the ecotoxicological effects of MPs.Figure 2 illustrates the collaboration network of countries conducting MPs research from 2004–2019. The number of publications threshold was set at 30, and of the 104 countries considered, 29 met this threshold. The whole network consisted of 29 nodes (referred to as countries) and 306 links (total link strength = 1815). England and the USA were the most affiliated countries; their close international cooperation was indicated by 28 links and a total link strength of 353 and 369, respectively. They were followed by Germany (links = 27, total link strength = 274), France (links = 26, total link strength = 218), and the Netherlands (links = 25, total link strength = 239). Academic collaboration between China and the USA was far more frequent than that of any other two countries (link strength = 52), which may be attributed to the high number of Chinese postgraduates/visiting scholars studying or working on MPs research in the USA. Other countries had fewer academic exchanges, such as Turkey (links = 9, total link strength = 12), possibly due to the consequence of language and finance barriers.Of the 57,834 cited references from MPs articles published between 2004 and 2019, 713 references that were cited at least 30 times were used to create the co-citation network diagram (five clusters with different colors, Figure 3). Each cluster contained some core literatures with high citation rates and academic relationships, which revealed a knowledge base in the MPs research field.In cluster red, the references with the largest nodes were the articles by Browne et al. [34] and Hidalgo-Ruz et al. [35], published in Environ Sci Technol, both with 712 co-citations and total link strengths of 25,800 and 24,418, respectively. Browne et al. [34] was the first study to explore the global distribution of MPs, which formed the knowledge base for MPs spatial distribution research. Hidalgo-Ruz et al. [35] reviewed 68 studies, to compare the methodologies used for MPs identification and quantification from seawater and sediment samples, and they called for standardized sampling programs to develop a more comprehensive understanding of MPs distribution. This study undoubtedly formed the knowledge base for MPs analytical methods. In cluster green, the documents with the largest nodes were authored by Andray [36] (published in Mar. Pollut. Bull.) and Thompson et al. [25] (published in Science). These articles were co-cited 712 times and had total link strengths of 29,817 and 26,743, respectively, indicating that they played a crucial role in the MPs co-citation network structure. Thompson et al. [25] clearly defined the term “MPs” and initiated global research on them. Andray [36] discussed the mechanism by which MPs are derived from marine debris, forming the knowledge base for MPs sources. In cluster blue, the document with the largest node (712 co-citations, 23,574 total link strength) was the article authored by Wright et al. [8] and published in Environ Pollut. Additionally, the laboratory experiments conducted by Setälä et al. [37] and Mattsson et al. [38] confirmed that MPs could transfer through food chains, and that lower trophic organisms could be the vector. In cluster yellow, the document with the largest node was written by Teuten et al. [39], who examined the uptake and subsequent release of hydrophobic organic contaminants present on plastic debris. This study formed a knowledge base about the interaction of MPs with contaminants. In cluster purple, Zettler et al. [40] first described a microbial community as a “plasticphere”, and called for research on the interaction between MPs and microorganisms.(1) In Publications Extracted from WoSCCThere were 4957 unique author keywords recorded in extracted publications from WoSCC. Among them, 3785 words (76.4%) were only used once, 566 (11.4%) were used twice, and 178 (3.6%) were used three times. These author keywords emphasized the breadth of MPs-related research, but also indicated a lack of continuity in research focuses. Some author keywords had different forms, but the same meaning (e.g., “FT-IR Spectroscopy” and “FT-IR” manually standardized as “FT-IR”), so we manually standardized 299 author keywords (with a minimum of 5 occurrences) to 230 keywords, and used them for co-occurrence network analysis (Figure 4). “Microplastic” (the biggest dot, Occurrence = 1146) was the most frequently used author keyword (and was used as our search term). The keywords “marine environment pollution” (occurrence = 366), “marine debris” (occurrence = 296), “ingestion” (occurrence = 105), “nanoplastic” (occurrence = 89), “sediments” (occurrence = 88), “polystyrene” (occurrence = 69), “FT-IR” (occurrence = 66), “freshwater” (occurrence = 60), and “polyethylene” (occurrence = 57) ranked second to tenth in the author keywords analysis, during the period from 2004 to 2019. These keywords were used in a large number of articles dealing with the distribution of MPs in different environments (e.g., marine environment, sediments, and freshwater), the ingestion of MPs by organisms, analytical techniques and quantification of these particles, and the biological effects of exposure to polystyrene or polyethylene nanoplastic. Figure 4 shows that the author keywords provided for MPs articles formed different clusters (by color), which represented global hotspots in MPs-related field (see Section 3.3).(2) In ESI highly Cited PapersThere were 859 unique author keywords recorded in 395 ESI highly cited papers of MPs research, among which a total of 26 author keywords (manually standardized) appeared at least five times. The co-occurrence network analysis of these 26 author keywords were shown in Figure 5. “Microplastic” (the biggest dot, Occurrence = 198) was undoubtedly the most frequently used author keyword in ESI highly cited papers of MPs research, followed by “marine environment pollution” (occurrence = 72), “marine debris” (occurrence = 59), “sediments” (occurrence = 40), “ingestion” (occurrence = 28), “freshwater” (occurrence = 24), “nanoplastic” (occurrence = 23), “fish” (occurrence = 21), “accumulation” (occurrence = 13), “mussels” (occurrence = 12), ranked second to tenth. Most of these keywords were the same as those in publications extracted from WoSCC, and they also contributed to exploring the global hotspots of MPs-related research in the next section.Contributed keywords can be found in cluster red and green in Figure 4 and Figure 5, including marine plastic pollution, marine debris, beach, surface water, Mediterranean, accumulation, etc.To address the key MPs problems, the international community has already developed policy responses, such as the European Marine Strategy Framework Directive, which is an ambitious program aimed at preventing MPs pollution. The scientific literature related to sources, pathways, and distribution of MPs is already substantial and constantly growing. MPs have been observed in almost every habitat of the aquatic environment, including ocean surface water [41], the water column [42], beaches [43], subtidal and deep-sea sediments [44,45], and freshwater lakes [46]. However, how to integrate MPs monitoring into existing environmental monitoring programs with reliable quantification routines requires further work.Contributed keywords can be found in cluster yellow in Figure 4 and cluster purple in Figure 5, including identification, density separation, elutriation, sediment, spectroscopy, FT-IR, Raman, etc.MPs analytical methods were developed to meet different purposes in MPs surveys, and can be divided into three steps: (1) sampling; (2) pretreatment for MPs extraction; and (3) qualitative and quantitative analysis. For sediment samples, high density solutions were commonly used to extract MPs, based on density separation [47]. In addition, elutriation columns proved to be useful tools with high extraction efficiency [48]. For seawater or water samples, selective sampling [49,50] and bulk sampling [51] methods were used in different studies, as were different pretreatment methods, such as enzyme digestion with subsequent filtration [49], and the sieve method [50]. For biota samples, the sampling methods always depended on the organisms being studied. For example, plankton trawls and nets were used to study the accumulation of MPs in plankton [52], whereas the dissection of different organs [53] was used for other target species. Digestion pretreatment methods for MPs extraction from biota samples have received a lot of attention, and many studies compared different digestion pretreatment methods, in pursuit of higher extraction efficiency [54]. For qualitative and quantitative analysis, MPs were usually measured and classified by shape, size, color, and chemical components. Most commonly, the extracted MPs were visually sorted under a microscope, and in some cases, the chemical components of MPs were determined by Raman spectroscopy [49], FT-IR micro-spectroscopy [51], and FT-IR [53]. It is notable that MPs analytical methods are now still debatable and have yet to be standardized.Contributed keywords can be found in cluster purple in Figure 4 and cluster green in Figure 5, including sorption, heavy metals, polychlorinated biphenyls (PCBs), polycyclic aromatic hydrocarbons (PAHs), phthalates, persistent organic pollutants (POPs), etc.Partitioning of contaminations to MPs has also been well documented. It has been evidenced that POPs [55], such as PCBs and PAHs, as well as heavy metals, can be adsorbed onto the surface of MP particles [56], and the sorption capacity is not only influenced by external factors, such as salinity, environment temperature, and weathering, but also by polymer type [57]. Additionally, contaminants adsorbed to MPs as well as MPs additives (e.g., phthalates, alkylphenols) can be desorbed from the surface of MPs during their transit through the digestive tract in organisms [58], which may have negative impacts on them [59]. The challenges of understanding MPs sorption/desorption behavior in different environments and their combined toxic effects on organisms require further research.Contributed keywords can be found in cluster blue in Figure 4 and cluster blue and yellow in Figure 5, including ingestion, biomarkers, mussel, fish, seafood, trophic transfer, food web, toxicity, human health, etc.MPs have been detected in many wild aquatic organisms, such as beaked whale [53], lobster [60], crab [61], fish [62,63], bivalves [64,65,66], and zooplankton [52,67,68]. Laboratory experiments have shown that MPs can cause some physical harm to a diverse array of organisms upon ingestion [8]. Additional threats of MPs are their capacity to leach toxic additives such as monomers and plasticizers, and to be potential vectors for hydrophobic POPs, which may cause further health problems, such as endocrine disruption and even carcinogenesis to organisms upon ingestion [69]. To quantify the exposure risk from ingested MPs and to evaluate theirpotential eco-toxicological risk, scientists have studied the survival rate [70,71], growth [72], reproductive status [73,74] and gene expression [75] of target species. As biomarkers are useful indicators of exposure, they should be used to identify ecologically significant effects of MPs on sentinel species.In recent years, the global presence of MPs and nanoplastics in foodstuff, drinking water, and air samples has been well documented. Consequently, human exposure to MPs via ingestion and inhalationis inevitable. Initial concern about human exposure to MPs focused on ingestion of MPs contaminated aquatic organisms. Evidence has shown that the consumption of entirely consumed organisms such as oysters and mussels may pose a higher risk of MPs exposure than consumption of eviscerated ones [76]. Other sources of human exposure to MPs include commercial salt and bottled drinks, as well as airborne MPs that can be inhaled [77]. Although MPs have been detected in human stool samples [78], nanoplastics reportedly might decrease the viability of human Caco-2 cells [79], and they can induce pro-inflammatory responses [80]—adverse effects of MPs on human health have not been reported to date. Thus, studies of the effects of MPs on human health are still urgently needed.MPs, as an environmental pollutant, have become a global problem and they may pose a risk to human health. To summarize research progress and identify future research topics based on current hotspots, we conducted a bibliometric profile of MPs relevant research, using data from the WoSCC database for the period 2004–2019. We found that the scientific output of MPs-related research experienced rapid growth during the past 16 years, and that this booming research area has expanded into many related fields. Developed countries were important contributors to MPs research, as England, the USA, and Germany occupied the top three positions, based on the criterion of NSC. China was the only developing country in the top 10 national contributors. These influential countries foster close academic collaborations, as shown by the co-authorship network analysis of countries. However, more exchanges and cooperation between these countries and others are needed. We also found that Thompson RC, who defined the term “MPs”, was the most productive author, as well as the most influential one. All of the top 10 authors identified based on the criterion of NSC were ESI highly cited researchers in the Ecology/Environment field, which highlights their significant influence on MPs-related research. Our results show that the issue of MPs is a multidisciplinary research field, because journals classified in six different categories contained MPs-related articles. The results of co-citation network analysis of cited references indicated that in-depth research laid a solid foundation for the MPs scientific field. The internal composition relationships of MPs studies were visualized by co-occurrence networks of author keywords, both in extracted publications and ESI highly cited papers, which identified the following research hotspots: potential sources and spatial distributions of MPs, analytical methods, the interaction of MPs with contaminants, and the impacts of MPs on organisms as well as human beings. Future MPs studies should focus on the following five aspects: (1) integration of MPs monitoring into existing environmental monitoring programs; (2) unified technical standards and reliable quantification routines; (3) sorption/desorption behavior of contaminants on MPs in different environments; (4) biological effects on sentinel species and molecular toxicology mechanisms; and (5) the effects of MPs exposure on human health.Conceptualization, F.Q., J.D. and H.W.; methodology and software, F.Q., J.G. and G.L.; formal analysis, J.D., J.G., G.L. and Y.S.; investigation, A.Y., Y.D. and Q.W.; resources, J.G., G.L. and H.W.; writing—original draft preparation, review and editing, F.Q., J.D. and H.W.; visualization, F.Q., Y.S. and A.Y.; funding acquisition, J.D. and H.W. All authors have read and agreed to the published version of the manuscript.This research was funded by The Science and Technology Innovation Fund of Dalian, P.R. China, grant number 2019J13SN119; Key Laboratory of Huanghuai Water Environment and Pollution Control, Ministry of Education, P.R. China, grant number KFJJ-2017-11; Department of Ocean and Fishery of Liaoning Province, P.R. China, grant number 201808.We would like to thank XianWen Wang of Dalian University of Technology for his technical support on VOSviewer software.The authors declare no conflict of interest.(a) Annual number of publications on microplastics (MPs) research from 2004 to 2019, retrieved from WOSCC; (b) The cumulative annual number of publications since 2004 follows an exponential model.Co-authorship network diagram showing cooperation between countries (with a threshold of 30).Co-citation network diagram of cited references from MPs articles cited a minimum of 30 times.Co-occurrence network diagram of author keywords, appearing in a minimum of five publications between 2004 and 2019.Co-occurrence network diagram of author keywords appearing in a minimum of five ESI highly cited papers, between 2010 and 2019.The top 10 countries for MPs research ranked by NSC; values for other criteria are given as well.1 NSC: Non-self-citation; 2 TNP (R): Total number of publications (ranking); 3 STC(R): Sum of times cited (ranking).4 NSCR: Non-self-citation ratio; 5 ≥100: the number of publications cited more than 100 times; 6 ≥50: the number of publications cited more than 50 times.The top 10 authors for MPs research ranked by NSC; values for other criteria are given as well.1 NEHC: Number of ESI highly cited papers; 2 Ifremer: Institut Français de Recherche pour I’Exploitation de la Mer; 3 KIOST: Korea Institute of Ocean Science Technology.The top 15 productive journals that published articles about the MPs issue.

Med-MDPI/ijerph_5/ijerph-17-16-05640.txt

@@ -0,0 +1 @@
+Background: this study aimed to estimate the economic impact and health-related quality of life (HRQOL) of patients with spinal muscular atrophy (SMA) in three European countries. It was used a cross-sectional study carried out in France, Germany, and the United Kingdom. Data were collected from July 2015 to November 2015. Healthcare costs (hospitalizations, emergencies, medical tests, drugs used, visits to general practitioners (GPs) and specialists, medical material and healthcare transport), and non-healthcare costs (social services and informal care) were identified and valued. EuroQol instruments, the Zarit interview, and the Barthel Index were also used to reflect the burden and the social impact of the disease beyond the cost of healthcare. Results: we included 86 children with SMA, 26.7% of them had Type I, and 73.3% Type II or III. The annual average cost associated with SMA reaches €54,295 in the UK, €32,042 in France and €51,983 in Germany. The direct non-healthcare costs ranged between 79–86% of the total cost and the informal care costs were the main component of these costs. Additionally, people suffering from this disease have a very low health-related quality of life, and there are large differences between countries. Conclusions: SMA has a high socioeconomic impact in terms of healthcare and social costs. It was also observed that the HRQOL of affected children was extremely reduced. The figures shown in this study may help to design more efficient and equitable policies, with special emphasis on the support provided to the families or on non-healthcare aid.Studies of the economic impact of a disease, commonly known as cost-of-illness studies, are a type of analysis that is well known and disseminated in the scientific literature in the field of health economics. The interest of such studies lies in revealing an insufficiently known dimension of a disease—its economic burden—and in incorporating this information into the body of knowledge about it. Thus, in the field of health economics, these studies are the equivalent of epidemiological studies in the field of public health. Moreover, although they do not allow us to identify the most effective or efficient interventions in a specific disease, they do help to raise awareness of its social impact [1,2,3,4,5,6,7].Although, in the field of high-prevalence diseases, the presence of this type of studies is frequent and growing [7,8], this is not the case in the field of rare diseases, due to the inherent difficulty of obtaining information about the people who suffer from them. Even though, in recent years, efforts have been made to find more information about the economic burden posed by rare diseases [9], there is still a serious lack of information about many of them. First, due to their low prevalence, the correct diagnosis of rare diseases is complex and subject to significant delays. Moreover, most rare diseases have no cure, and, for many, either there is no effective treatment available or, if treatments do exist, there is no guarantee of improvement in life expectancy or quality of life.Several factors could explain why this disease has such a strong social impact on sufferers. They are its severity, uncertainty in the diagnosis, and the lack of effective treatments. They suffer from degenerative and life threatening, not only for the sufferers, but also for their families. It has been shown in the literature that approximately 50% have onset in childhood, and over one-third of deaths of children under one year old are due to rare diseases [9,10]. The health-related quality of life (HRQOL) of people who suffer from these diseases is also seriously threatened. The results of recent studies are coincident in concluding that people suffering from these diseases show results well below those of the general population, as several dimensions of the HRQOL being affected simultaneously [11].One very common rare disease is spinal muscular atrophy (SMA). SMA is an autosomal recessive neuromuscular disorder caused by the degeneration of alpha motor neurons in the anterior horns of the spinal cord. It is caused by homozygous absence or pathogenic variants in the survival motor neuron gene 1 (SMN1), and the phenotype is mainly influenced by the number of copies of a highly homologous gene, survival motor neuron gene 2 (SMN2), present in all patients. Weakness is the most important manifestation, with several complications such as respiratory insufficiency, scoliosis, contractures, and nutritional problems. SMA is classified in three main types according to age of onset and motor milestones achieved. Type I starts in the first weeks or months of life, the patients never sit, and death occurs in the first two years of life. Type II manifests after 6 months, patients never walk and are wheelchair-bound for life. And Type III appears after 18 months and patients may walk for several years but may lose this ability later [12].In fact, SMA has an incidence of 1/5000 to 1/10,000 births and a carrier frequency of 1/35 to 1/50 [13], being one of the most severe hereditary diseases among children. Furthermore, the disability caused by this disease increases the difficulty of carrying out the activities of daily living (ADL) and the burden borne by the families [11]. Therefore, in order to quantify the real economic burden of this disease, it is therefore necessary to take a broader view, and to consider the cost of formal care and unpaid care, as well as other household costs.To our knowledge, evidence of the total economic impact of SMA in Europe is scarce [11,13]. Thus, the main aim of this study is to fill two gaps in the information about SMA. First, by estimating the costs related to SMA from a societal perspective in the three European countries with the largest populations: Germany, France, and the UK. Secondly, by studying the HRQOL of SMA patients and their caregivers.This was a cross-sectional study of patients diagnosed with SMA who received outpatient care at the time of the study in three different European countries: France, Germany, and the United Kingdom. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were followed in the study [14].Data were collected from July 2015 to November 2015. Children/adolescents diagnosed with SMA were eligible. Thus, 86 children and their caregivers were included in the study. The caregivers completed the administered questionnaires to supply information about the use of public health and non-health resources. More detailed information about the design and procedure of the study is available elsewhere [11]. We defined informal caregiving as a heterogeneous personal service, composed of various specific tasks provided to cover the basic or instrumental needs of a person with limited autonomy. More precisely, it is a non-professional activity, in the sense that the people who provide this care do not enjoy recognized employment rights, including weekly schedules and rest periods. Likewise, although they may receive family assistance or a public subsidy, it is not usually a paid activity. Particularly, we considered informal caregivers to be those who reported that they provide at least one hour of care per day.The survey was totally anonymous, a patient organization/registry contacted the patients, and their responses were not associated with any identifying data (name, ID, address, e-mail). However, the study was submitted to an ethical committee in Germany to have a justification that the project and the collection of the data was made according to the ethic conditions.The health-related quality of life of both patients and their caregivers was analysed. For this purpose, we used the proxy version of the EuroQol 5-dimentions and 3-levels (EQ-5D-3L) for patients and EuroQol 5-dimentions and 5-levels (EQ-5D-5L) for caregivers. The main reason for using different versions of questionnaires was that the EQ-5D-5L is only validated for adult responders. These questionnaires are generic instruments that are used for assessing the quality of life by considering five different dimensions: mobility, self-care, everyday activities, pain/discomfort, and anxiety/depression [15]. The standardized reference values are 0 (death or equivalent to death) and 1 (perfect health), although negative values (health status worse than death) also are possible. In addition, the EQ-5D instruments include a Visual Analogue Scale (VAS) that represents values from 0 (the worst health status) to 100 (the best) when asking participants to rate their overall health on the day of the interview.Two other instruments were used to analyse the degree of patients’ dependence and the burden of care for caregivers. First, the Barthel Index, which measures the (dis)ability to perform the activities of daily living, assessing the degree of dependence [16,17,18], going from 0 points (totally dependent) to 100 points (totally independent). Secondly, the Zarit burden interview (22-item version) was used to measure the burden borne by caregivers due to the tasks provided. In this case, caregivers are supposed to answer questions about how they regard the care. The total score ranges from 0 to 88, with scores under 21 corresponding to little or no burden and scores above 61 to severe burden [19].We used a prevalence approach from a societal perspective. In order to estimate the economic burden of SMA, several questionnaires were used to collect information about the utilisation of health and non-health resources in the six-month period prior to the study (except for hospital admissions, for which the period used was the 12 months prior to the study). Costs were extrapolated to show annual costs. Information about hospital admissions, emergencies, medical tests, drug consumption, visits to general practitioners (GPs) and specialists, medical material and healthcare transport was collected in each country, and national reference prices were applied. More precisely, for the UK, the National Tariff Payment System was used to evaluate performance by the English National Health Service. For Germany, the information was obtained from Rosenfluh Publikationen AG, which provides official information about the German health system. In the case of France, the information was obtained from the websites “la sécurité sociale française” and “eureka Santé par vidal”, which provide official information about the health system in France.Regarding non-healthcare costs, information was obtained about the social services used as well as about the informal (non-professional) care. Thus, the number of caregiving hours was obtained from the questions about the time spent providing children with help to carry out the Activities of Daily Living (ADL). The number of caregiving hours was then assessed by the proxy good method [20,21,22]. This technique assesses the value of the care provided by considering how much it would cost if informal caregivers had to be replaced in the employment market by a close substitute. Professional care wages per hour were €23.88 in the UK, €12.02 in France and €17.40 in Germany, according to the public rates provided by each country. Finally, for social services utilization, the information was collected from the questionnaires about the use of programmed home care, day centres, supportive social work, occupational centres, respiratory physiotherapy, physiotherapy, occupational physiotherapy, information/advice/assessment, psychosocial care for families, residential centres, hydrotherapy, and respite in temporary stays. The unitary cost for each service was obtained from local official sources. All prices are based on the year 2014.A total of 34, 27, and 25 children with SMA, and their caregivers, completed the questionnaires in the United Kingdom, France and Germany respectively (Table 1). Most of the children with SMA were classified as type II (58% and 48% for the UK, France, and Germany respectively), and had average ages of 5.5, 6.1 and 9.5 respectively for the three countries considered. The majority of children went to ordinary schools, although there were quite a few of them who went to nursery schools, especially in Germany (20%).We identified 56 caregivers, 75% of whom reported a positive number of caregiving hours. Most of them were females, mainly in France (94% vs. 78% and 64% in Germany and the UK, respectively) (Table 1). They were also, on average, older in Germany (42 years old compared to 41 and 36 years in the UK and France respectively). In relation to the daily number of caregiving hours, it was observed that in the UK the intensity was higher in comparison with France and Germany. Informal caregivers from the UK provided an average of 12.50 h per day, while in France and Germany people cared for 10.65 and 9.31 h per day respectively. Finally, even though in France the intensity of caregiving was lower (in comparison with the UK and Germany), the burden borne due to the care was higher. More precisely, French caregivers produced a Zarit score of 40.37 versus 26.63 and 21.33 in the UK and Germany, respectively.Table 2 shows the health-related quality of life for both patients and their caregivers. Regarding children, it was observed that French children had a lower quality of life, with a utility score of 0.12, a figure quite similar to that of children in the UK, with 0.17. By contrast, German children had a significantly better quality of life with a time trade-off (TTO) of 0.53. The dimension with the worst results was that of self-care, in which 41.18% of children in the UK said that they were unable to wash or dress themselves (44.44% in the case of France and 32% in Germany). These results correspond with those obtained in caregivers, as the French had an average utility score of 0.39 (the lowest), compared with TTO scores of 0.85 and 0.80 in the UK and Germany respectively.The annual average cost associated with SMA reached €54,295 in the UK, €32,042 in France, and €51,983 in Germany (Table 3). In the three countries analysed, direct non-healthcare costs ranged between 79–86% of the total cost associated with SMA. More precisely, the item with the highest weight among the total costs was the care provided by relatives (informal care). Nevertheless, although the weights of healthcare and non-healthcare costs as a proportion of the total cost associated with SMA were quite similar, the amounts of such costs differ among the countries considered in this study. What is more, the percentage healthcare and non-healthcare costs above the total cost in each country are quite similar while the absolute figures differ. For instance, main informal caregivers cost entails around 61–63% of the non-healthcare cost of each country, but their absolute figures differ significantly, ranging from 17,500€ (in France) to 27,500€ (in Germany). (Figure 1).In the United Kingdom, the total cost was the highest one, at €54,295, and the direct non-healthcare costs were also the highest, at €43,214 (that is, 79% of the total cost) and with direct healthcare costs amounting to €11,081 (20.4%). Within direct healthcare costs, the cost of specialist visits was €4569 (8.4% of the total cost), of health material, €1958 (about 3.6% of the total cost), of hospitalization, €2219 (4.1% of the total cost), of medical tests, €874, of healthcare transport, €58€ and of GPs and Emergencies, €842. Finally, drugs were valued at €560. Regarding direct non-healthcare costs, the cost of informal care was estimated at €40,526 (74.6% of the total cost) per year, while the cost of social services was €2187 (almost 4% of the total cost).France had the lowest total cost associated with SMA (€32,042). €4672 (14.6%) corresponded to direct healthcare costs, while direct non-healthcare costs amounted to €27,370 (representing 85.4%). Within the cost of direct healthcare, the cost of medical visits was the highest, at €1870 (5.8% of the total cost), followed by hospitalization, the cost of which reached €1229 (3.8%). Regarding the cost of direct non-healthcare—that of informal care was estimated at €25,619 (80.0% of the total cost) per year, and the cost of social service was €1029 (3.2%).Germany and the UK had the highest estimated costs (€51,983 and €54,295 respectively). 86% of the total cost corresponded to direct non-healthcare costs (€44,670), while direct healthcare represented 14% (€7313). Within the cost of direct healthcare, that of hospitalization was the most relevant, at €3170€ (6%). Regarding direct non-healthcare cost, the care provided by main caregivers was valued at €27,436, and the estimated value of the care provided by other carers was €12,490. The cost of social services was €4380 (8% of the total cost) (Table 3).In brief, the results showed that the economic impact of SMA involves 1.40 times the GDP per capita in Germany, 1.02 times the GDP per capita in France and 1.70 times the GDP per capita in UK. Then, the figures display that the weight of the economic impact that disease has is quite similar across the countries included.A table with costs in PPP is included in the Appendix A to a better comparison of these countries.This study represents the first complete and realistic costing study to date of the burden of SMA patients in Europe. Particularly, our results show that SMA is a disease that has a great economic impact from the perspective of society. In the countries considered, the total costs range from €32,000 to €54,000 per person per year, depending on the country. These figures include both high expenditure on health (between €4,700 and €11,000 on average per person, depending on the country) and even higher non-health costs (between €27,000 and €45,000 per person, depending on the country). Even though spending on health is very relevant, the importance of the resources invested in social services must also be highlighted (Germany stands out at €4,400 per person per year). However, the cost of informal care stands out as the main cost item, oscillating between 75% and 80% of the total economic impact. In addition to the impact on health, therefore, most of the economic impact falls on families in the form of time spent on care.Broadly, our estimates do not greatly differ from those made previously in Germany [13] and in Spain [12], but some figures need to be clarified. For instance, the total costs of SMA were €54,721 for German patients and €33,721 for Spanish patients (slightly different from ours). The main differences may be due to differences in the method of accounting for direct healthcare resources and in the economic assessment of informal care. First, the previous study carried out in Germany included, in the direct healthcare costs, the more expensive healthcare resources (such as artificial nutrition systems, rehabilitation services and respiratory management) that we did not include in either this study or the one carried out in Spain. Secondly, the method used to assess the cost of informal caregiving time was not the same as the method used in the previous German study. Klug et al. only included the economic assessment of informal care costs for non-working parents in order to avoid double accounting because they also estimated the loss of productivity of working parents (which we did not) [13].Regarding the HRQOL, this study is the first that uses the EQ-5D instrument to estimate the utility index score associated with the state of health of SMA patients and their caregivers. The previous study carried out by Klug et al. used a different tool for estimates connected with this condition, and the results cannot be compared. However, the instrument used in our paper was the same as the one used previously in the Spanish study. In this sense, the utility score of Spanish SMA patients was a lot lower (0.16 vs. 0.53) than for German SMA patients. However, the VAS score results included in the EQ-5D did not show such huge differences (54 vs. 69). Meanwhile, the HRQOL of informal caregivers was also different in the utility index score (0.49 vs. 0.81) but very similar in the VAS results (69 vs. 71).Another point that should be highlighted is the large difference, identified in our study, between the HRQOL of the patients in France and the UK compared with that of patients in Germany. It seems that neither the age of the patients nor the degree of disease progression explains the differences observed. Therefore, we can only point out this fact and leave open, as a line of future research, the analysis of HRQOL in several countries, with as many samples as possible and with questions specially designed to give understanding of this variability. The same can be said about the important differences identified between the HRQOL of French caregivers and that of their counterparts in the UK and Germany.Among the main limitations of the analysis, we can mention the limited sample size. Obviously, since it is a disease of low prevalence, it is expected that large sample sizes will not be available. However, it must be recognized that, given the distribution of SMA types over the total samples, it was not possible to perform an additional analysis for each phase of disease progression. Thus, another possible limitation could be the fact that the collection of information on patients was though patients’ organizations, and this might entail a selection bias. However, this method for collecting data has been also performed in other studies focused on rare diseases [23,24,25,26]. Likewise, the design of our study was cross-sectional, and the questions related to the health and non-health resources were retrospective. Ideally, the study would be a prospective one, using a longitudinal cohort of people with SMA, thus avoiding, for instance, the recall bias. However, we used an ad-hoc questionnaire aimed at avoiding recall bias, and in which the participants could answer the questions with no time limit. On the other hand, we should take into account the fact that, in studies where the subjects are children, it is difficult to identify how much time of care was provided due the illness and how much time was provided due to the development of the child. In this sense, we included questions aimed at estimating the informal care time, with a specific statement, which highlighted that the time provided should be related only to the illness.Among rare diseases, SMA is one of those that receive attention from the health authorities and from society as a whole. First, because of its social consequences worldwide, and secondly, because of its incidence, prevalence and consequences in terms of loss of quality of life, mortality and morbidity. In fact, the estimated cost of SMA proves to be higher than the social costs (i.e., informal care costs) of other rare diseases such as ataxia (€18,776, base year 2004) [23] and similar to those of the fragile X syndrome (€31,008, base year 2012) [24], amyotrophic lateral sclerosis (€36,194) [25], and Duchenne muscular dystrophy (€36,970, base year 2012) [26].In brief, the economic assessment of the informal caregiving time provided due to SMA disease reached figures higher than 70% of the total costs. This means that the majority of resources needed by SMA patients seem to come from outside the healthcare system. Even though some authors have assessed how the activities performed by informal caregivers affect their wellbeing [27,28,29] or their promotion in the workplace [30], there is still a lack of evidence in the field of SMA. Consequently, further research, focusing on identifying the effects of care on the health, employment status and socio-familiar dimension of informal caregivers, is needed. This information would help decision-makers to understand the vast effect of this disease in society, beyond its consequences for patients and the healthcare system. For instance, the inclusion of public health strategies focused on respite services may mitigate the aftermath of this illness beyond the patients [31,32], even in other caregiving populations [33]. Furthermore, financial aid for low-income households would also help the families that have—in addition to the health-related problems, or work-related problems—suffered by the main informal caregiver, financial problems due to the expensive medical material that they require. In fact, the cost of the families’ medical material and out-of-pocket expenses (such as those for house adaptation) may increase the cost of this illness beyond the healthcare resources needed [34]. Therapeutic agents for the treatment of SMA have therefore been approved for clinical use or are now in ongoing clinical trials. An antisense oligonucleotide that affects splicing of the pre-mRNA from the SMN2 gene (nusinersen-Spinraza®) was approved by the U.S. Food and Drug Administration (FDA) in December 2016 and by the European Medicines Agency (EMA) in June 2017 [35]. A self-complementary adeno-associated virus serotype 9 (AAV9) SMN1 gene therapy (Onasemnogene Abeparvovec, ZolgenSMA®) was approved recently in May 2019. New disease trajectories and evolving phenotypes are observed with these treatments [36]. The high cost of these treatments raises concerns about access and equity that should also be considered in the total burden of the disease and in the evaluation of these treatments.SMA produces considerable societal costs in France, Germany, and the UK due to its relevant economic impact and the deterioration in the HRQOL, not only of the patients, but also of their caregivers. For this reason, when designing and evaluating any strategy or intervention for this population, the economic impact should be considered, as well as the economic evaluation of new treatments in this field.Conceptualization, J.L.-B. and J.O.-M. methodology, L.M.P.-L.; curation, I.A.-R. and L.M.P.-L; writing—original draft preparation: L.M.P.-L., I.A.-R., J.D.-M., S.L., I.D.-Z., E.T., and J.L.-B. All authors have read and agreed to the published version of the manuscript.This analysis has been supported by BIOGEN.The authors declare no conflict of interest.The datasets generated and/or analysed during the current study are not publicly available due to [it is own by third party] but are available from the corresponding author on reasonable request.Average annual costs in PPP by country (€) (2014).Note: Bold/italic/shade are the main results of each category. a This includes costs associated with non-healthcare transport.Weight of non-health care resources in the total costs by country. Source: Own preparation. Note: figures are percentages over the total costs.Demographic characteristics of participants and their caregivers by country.Source: own preparation. a Number of daily hours allocated to informal caregiving was higher than 0. b People with a Zarit score equal to or higher than 55 points. SD: standard deviations. NA: no answer.Health-related quality of life (HRQOL) of patients and caregivers by country.a (SD). Source: own preparation. SD: standard deviations. TTO: Time trade-off. VAS: visual analogue scale.Average annual costs by country (€) (2014).a This includes costs associated with non-healthcare transport and housing and vehicle adaptation. Source: own preparation SD: standard deviations. GP: general practitioner. Note: Bold/italic/shade data are the main results of each category

Med-MDPI/ijerph_5/ijerph-17-16-05641.txt

@@ -0,0 +1 @@
+School-based child nutrition programs provide students with meals and snacks that align with guidelines for a healthy eating pattern. However, participation is not universal, and research on the determinants of food selection is needed to improve school nutrition practices and policies. The purpose of this study was to examine the relationships between grade level (i.e., grade school, middle school, or high school) as well as meal participation category (i.e., only breakfast, only lunch, or both) and food trying and liking in a large urban school district. Outcomes were measured using an online survey completed by students from 2nd through 12th grade (n = 21,540). Breakfast and lunch item liking scores were higher among the grade school and middle school students than among the high school students. Breakfast and lunch liking scores were also higher among those who participated in both breakfast and lunch as opposed to those who only participated in one meal. Food item liking scores were positively correlated with the percentage of students who had tried the particular foods (r = 0.52, p < 0.001), and the number of foods tried was dependent on both grade level and meal participation category (F(4, 21,531) = 10.994, p < 0.001). In this survey of students, both grade level and meal participation category were found to be related to the liking of foods, while foods that were tried more often tended to be liked more. Future studies should consider grade level and meal participation when exploring student preferences. School nutrition programs should also consider these factors when assessing satisfaction.Optimal growth and development is reliant on sound nutrition [1,2,3]. Yet, the diets of many children and adolescents in the United States (U.S.) do not meet nutrient recommendations [4] and fall short of current guidelines for a healthy eating pattern [5]. Behaviors that contribute to dietary shortfall among children (e.g., consumption of sugar-sweetened beverages, limited fruit and vegetable intake) are associated with an array of individual and environmental factors including infant/child feeding practices [6,7,8,9], food neophobia [6,8], parental eating patterns [6,9,10], food insecurity [11], television watching [9,12], the food environment [9,10], and public policies [13]. Schools provide numerous opportunities to influence long-term food habits, thereby making them a prime environment for improving the eating patterns of students [14]. The U.S. Department of Agriculture (USDA) school-based child nutrition programs are important public programs developed to ensure that children receive meals and snacks that align with current guidelines for a healthy eating pattern [15]. The National School Lunch Program (NSLP), which serves nearly 30 billion meals per day, is the second largest food and nutrition assistance program in the U.S. [15]. Other school-based child nutrition programs include the School Breakfast Program (SBP), the Fresh Fruit and Vegetable Program, and the Special Milk Program [15]. While these USDA child nutrition programs provide many benefits, participation is not universal [16]. Furthermore, among students participating in school meal programs, food acceptance and food waste is an ongoing issue [17,18,19,20,21,22]. In addition, satisfaction, in particular taste, is a common reason that students cite for throwing away food [17]. Needless to say, research on the determinants of food selection and food satisfaction has important implications for school nutrition practices and policies. The role of initiatives such as choice architecture [23], motivational incentives [22,23], and the Farm to School Program [19,24] has been explored in a variety of manners. The relationships between socioeconomic and demographic characteristics and school nutrition program outcomes have been studied as well [25,26,27,28]. However, in all but a few exceptions, the underlying effects of grade level (i.e., grade school, middle school, or high school) and meal participation category (i.e., only breakfast, only lunch or both) have been overlooked.There are important socioeconomic differences associated with school nutrition program participation. NSLP participation rates vary among students based on free, reduced-price, and paid eligibility category [29]. In addition, the NSLP and the SBP differ with regard to both the number of meals served and the proportion of participants who qualify for free or reduced-price meals. The SBP serves fewer meals, but of the meals served, a higher proportion is provided to students who qualify for free and reduced-price meals [30]. Compared to students who pay full price, students receiving free or reduced-price meals are less likely to select a vegetable and more likely to select milk [25]. In addition, students paying full price are more likely to bring lunch from home and more likely to select school entrées higher in protein [27]. In turn, these differences may result in food preferences that are associated with meal participation category. In studies designed to examine food consumption and plate waste, grade-level differences in whole fruit [18], vegetable [21], grain [21], and protein food [21] consumption have been noted. A study of school children’s food preferences by Cooke and Wardle [31] found age-related differences in the number of foods tried and the number of foods liked. A survey conducted within an Ohio school district revealed food preference differences among elementary, middle and high school students [32]. However, this survey of students in Ohio was conducted prior to the implementation of the Healthy, Hunger-Free Kids Act of 2010 (HHFKA) [33], and many of the most-liked foods were high in fat and calories. Since implementation of the HHFKA, school meal programs have had to adopt a number of changes, which have resulted in the reduction of saturated fat [34] and sodium [34,35] and improved overall diet quality [35]. However, these changes have also resulted in modifications to, or the disappearance of, various familiar foods [36]. The purpose of this study was to examine the relationship between grade level as well as meal participation category (i.e., only breakfast, only lunch or both) and food trying and liking in a large urban school district following the implementation of the HHFKA.Responses were analyzed from a volunteer sample (n = 21,540) of school children from one urban school district in the U.S. The children, students from 2nd through 12th grade, all participated in the meal service (i.e., breakfast, lunch, or breakfast and lunch). The survey was developed in cooperation with the school district’s nutrition department and the food service management firm responsible for providing meals to the district’s schools. The survey was designed to collect data on a variety of aspects including dining experience satisfaction, food selection and liking, least and most liked condiments, favorite foods (e.g., Mexican food, Asian food, soul food/traditional southern food) and flavoring profiles, willingness to try different items, and preference for various diets or eating patterns (i.e., vegan, Kosher, Halal). This study was limited to a set of items regarding food selection and liking. The food liking items included 48 core school foods which were comprised of 23 breakfast items and 25 lunch items. Before being asked about food item liking, students were first asked to indicate which of the core school foods they had previously eaten. Liking responses were only collected for foods that were eaten. In addition to the core items, all students rated their liking of fruit as a broad category, cold vegetables as a broad category, and hot vegetables as a broad category. Individual breakfast and lunch core items included a pictorial prompt of the food items, while the fruit, cold vegetable and hot vegetable categories included a graphic illustrating a variety of items. As a method to establish a benchmark for liking beyond school food items, students who indicated that they brought food from home were asked to rate their liking of food from home. Students indicated their liking rating for each of the individual food items and broad food categories using a 7-point emoji facial scale [37]. This emoji scale was developed as an alternative to the Peryam and Kroll (super good/super bad) 9-point hedonic scale for children [38]. Findings from a comparison study [37] suggest the emoji scale is a suitable (and more modern) option. Prior to being utilized in the current study, the 7-point emoji facial scale was incorporated into preliminary versions of the survey that were pilot tested once in 2017 and again in 2018. In both cases, the proportion of surveys left incomplete was not deemed excessive. For analysis purposes, the response options were coded 1 through 7, with the most negative emoji being coded as a 1 and the most positive emoji being coded as a 7. Liking for a food was defined as a score of 6 or 7, and disliking as a score of 1 or 2. For the Cronbach’s alpha analysis, “never tried” was scored a 0.It should be noted that the demographic information solicited from student participants was limited to grade level. While gender, ethnicity or other socio-demographic information are commonly solicited in surveys, the school district considered these questions sensitive in nature, so they were specifically excluded from the survey. All schools in the district were eligible to participate. The school food service team worked with school administrators to distribute an anonymous survey link to their student population (n = 361,314). The survey was completed either as a classroom activity or during the students’ personal time. Based on a review of time stamps, a majority of the surveys were completed during the school day and seemed to cluster time-wise within a building. Of the 24,767 surveys that were completed, 21,540 surveys from 449 school buildings contained the completed responses necessary for inclusion in this study. Information regarding whether the survey was read aloud for those children who had difficulty reading was not obtained. The survey was noninvasive and voluntary. Students and their parents were not required to complete consent forms. The Kansas State University Institutional Review Board approved the research protocol (IRB proposal #5930). The survey was a self-administered, online survey using Compusense Cloud (Compusense, Inc., Guelph, ON, Canada) available from 13 January to 22 March 2019.Statistical analysis was performed using the IBM SPSS Statistics for Windows, Version 25 (IBM SPSS Statistics for Windows, IBM Corporation, Armonk, NY, USA). For all analyses, the significance level was identified at p < 0.05. Correlation analysis examined the relationship between the percentage of children indicating they had eaten each core food and the mean liking score of those who had eaten it. Participation in school meals was categorized as one of three levels: breakfast only, lunch only, or breakfast and lunch. Three grade level groupings were created: elementary school, 2nd through 5th grades; middle school, 6th through 8th grades; and high school, 9th through 12th grades. Univariate ANOVA was used to examine the effects of grade level, meal participation category and grade level-by-meal participation category interaction on the number of foods tried, while ANCOVA was conducted to determine the impact of the same main effects and interaction on the number of foods liked or disliked, controlling for the number of foods tried. The core food items were divided into two groupings, breakfast (23 items) and lunch (25 items), to be assessed by Cronbach’s alpha. Further analysis investigated the effect of grade level and meal participation level on mean liking scores for each category of foods.Survey responses were obtained from elementary (n = 9731), middle (n = 8668), and high school (n = 3141) students. Respondents had tried, on average, fourteen of the 48 core school food items. Across the sample, the five most highly rated breakfast items were tator tots (M = 6.12), strawberry and yogurt parfait (M = 5.90), French toast sticks (M = 5.76), waffles (M = 5.72) and cereal (M = 5.69). Of these five items, tator tots, strawberry and yogurt parfait, and French toast sticks were among the top five most highly rated items for each grade level. Overall, the five most highly rated menu items for lunch were yogurt and cheese kit (M = 5.83), spicy popcorn chicken (M = 5.64), nachos (M = 5.62), chicken nuggets (M = 5.46) and cheese pizza sticks with marinara sauce (M = 5.38). All but chicken nuggets and cheese pizza sticks with marinara sauce were among the most highly rated for each grade level. Among all students, the five breakfast items with the lowest ratings were egg and cheese on English muffin (M = 4.68), pear and yogurt parfait (M = 4.67), strawberry kiwi bar (M = 4.60) oatmeal raisin bars (M = 4.58), and egg and cheese quesadilla (M = 4.28). All but egg and cheese on English muffin and oatmeal raisin bars were consistently among the bottom five menu items for each grade level. Overall, the lunch items with the lowest ratings were cheesy nacho bake (M = 4.53), cheese ravioli (M = 4.42), peanut and butter jelly sandwiches (M = 4.42), burritos (M = 4.38) and tuna melt (M = 3.91). Of these lunch items, cheesy nacho bake and cheese ravioli were among the lunch items with the lowest rated items for each grade group.Across the sample, liking scores for each of the core items were positively correlated with the percentage of students who had tried the particular foods (r = 0.52, p < 0.001). ANOVA revealed a statistically significant interaction between the effects of grade level and meal participation category on the number of foods tried (F(4, 21,531) = 10.994, p < 0.001). The mean number of items tried with 95% confidence intervals (CI) are presented in Figure 1. The number of foods tried was greatest among students participating in both breakfast and lunch meal service. The number of foods tried did not vary with grade level among students who only participated in the breakfast meal service. However, among students who participated in lunch only, as well as among those who participated in both breakfast and lunch, middle school students tried more food items than either the elementary or the high school students.The ANCOVA controlling for the number of foods tried revealed a significant two-way interaction between grade level and meal participation (F(4, 21,530) = 84.651, p < 0.001). The mean number of foods liked with 95% CI is presented in Figure 2. Within each meal participation category, elementary students liked more food items than any other grade level. Among students who only ate breakfast as well as among those who only ate lunch, middle school and high school students liked a similar number of items, while high school students participating in both breakfast and lunch liked fewer items than the middle school students in the same participation category. As with the number of food items liked, ANCOVA revealed that the number of food items disliked was dependent on both grade level and participation category (F(4, 21,530) = 7.759, p < 0.001) (Figure 3). The number of items disliked did not vary significantly among grade level groups who participated in breakfast only or lunch only. However, among students who participated in both breakfast and lunch meal service, high school and middle school students disliked more items than elementary students. To examine the pattern of food preferences, core menu item scores were grouped by meal. Cronbach’s alphas were 0.85 and 0.86 for breakfast and lunch, respectively. Category-based satisfaction scales were produced for both breakfast items and lunch items by calculating the mean of the liking scores of the foods in each category. Item means for both the breakfast and lunch scales along with the mean liking for food from home, fruit, hot vegetables and cold vegetables (by grade group and participation level) are presented in Table 1. Across all school groups, food from home was most well-liked, followed by breakfast items, fruit, lunch items and lastly vegetables (cold and hot). Age-related differences in liking were observed for both meal item scales (i.e., breakfast and lunch) and individual items (i.e., food from home, fruit, cold vegetables, and hot vegetables). Liking was highest among the youngest students for both breakfast and lunch scales and all individual food items, except food brought from home, where liking was highest among middle school students. Liking was lowest among the oldest students for both scales and all individual items with two exceptions, cold and hot vegetables. For these two items, liking was lowest among middle school students.Lunch, fruit and vegetables (hot and cold) were most well-liked by students who participated in both the breakfast and lunch meal service, while liking was highest for food from home and breakfast among students who only participated in the breakfast meal service. Liking was lowest for food from home, lunch items and fruit among students that participated in lunch meal service exclusively. Liking for vegetables (cold and hot) was lowest among the students who only participated in the breakfast meal service. Finally, liking for breakfast items was lowest among students who ate both breakfast and lunch provided meals.In this survey of students from 2nd through 12th grade, grade level and meal participation category, were found to be related to both the trying and the liking of foods. In addition, we observed that foods that were tried more often tended to be liked more. In all, these findings add to the overall understanding of meal service outcomes and in turn have important implications for school nutrition practices and policies.Aside from the group of students who only ate breakfast, the number of food items tried increased with grade level for those in elementary through middle school. This finding is similar to that obtained by Cooke and Wardle [31], who found that the number of foods tried increased with age in their study of school children. However, contrary to what might be expected, the number of foods tried decreased with grade level for those in middle through high school. This observation may be due in part to the timing of the menu changes that accompanied the implementation of the HHFKA. The high school students in this study were in elementary school at the onset of the implementation of the HHFKA. Those that were in middle school were either not yet in school or were in the very early elementary grades (e.g., kindergarten, 1st grade) when implementation of the HHFKA began. As such, the high school students had greater exposure to the pre-HHFKA school foods, and therefore may have been reluctant to try the new healthier menu items. The middle school students, some of whom started school after implementation of the HHFKA, had less exposure to the pre-HHFKA school foods. Having known little else, these middle school students may have been more willing to try the menu items. The association between grade level and food liking uncovered here is in line with previous research in which ratings of many food types differed by age or grade level [31,32]. However, whereas school food liking was the highest among the elementary school students in the current study, the liking of most entrée food groupings was highest among high school students in an earlier study [32].The most well-liked item was food from home, and students have previously indicated a preference for food from home as a reason for not eating school lunch [39]. These observations are concerning, as Farris et al. [40] noted that the majority of lunches from home include a dessert item, many include sugar-sweetened beverages, and a large number are missing both fruits and vegetables. Highly rated lunch items in this study were similar to those found to be well liked in an earlier study (i.e., pizza, chicken nuggets, yogurt, and cheese) [32]. Several of the lunch items with the lowest ratings were also similar to those with lower ratings in the same earlier study (e.g., peanut butter and jelly sandwich, tuna salad sandwich, burrito) [32]. Students have cited appearance, quality, and taste [39] as factors impacting their eating decisions, and efforts should be made to determine what factors are behind the low ratings of the less liked items in this study. Vegetables were the least liked menu items, with hot vegetables receiving lower ratings than cold vegetables. Aside from cooked potato items, vegetables were not found to be a favorite among school children in an earlier study [32]. Vegetables as a whole have received low ratings [31] and tend to be the most wasted category of school foods [17,20]. The liking of vegetables in this current study was lowest among middle school students. Conversely, fruit liking ratings were highest among the elementary school students, and this finding is supported by an earlier observation in which fruit was more positively rated by elementary students as opposed to the older middle school and high school students [32]. These findings regarding fruit are not surprising, as children 4- to 8 years of age consume a larger portion of their total intake as fruit compared to those 9- to 13 years of age as well as those 14- to 18 years of age [5]. On average, each student had tried less than one-third of the core food items listed on the survey. This relatively low rate of trying is concerning and may be a result of the overall participation rate. As was reported previously by Cook and Wardle [31], a positive correlation was observed between food item liking and the proportion of students who tried a food in this current study as well. Providing students with an opportunity to try menu items by way of taste testing events or product sampling can increase exposure and may, in turn, increase the acceptance of menu items. Data from an earlier study of school children revealed that some students participate in school meals regularly, while others participate rarely [27]. Offering opportunities to taste menu items may be a way in which to increase exposure, especially among students who do not participate regularly. Given that trying and liking are associated, opportunities to try school menu items may also be an approach to increase overall participation. Likewise, the local procurement of produce, which has been shown to positively affect vegetable consumption, may increase the trying and liking of foods [24]. Lastly, non-food motivational incentives such as stickers and stamps may also encourage positive eating behaviors [22,23]. However, the effect of such awards appears to vary by country [22], and as Decosta et al. [23] pointed out, incentives should be used with caution, as additional “research is needed to examine how instrumental feeding might affect children’s intrinsic drive to explore novel food”. Strengths of this study include the large sample size and the post-HHFKA timing of the survey, which was seven years after the initial implementation of the HHFKA. Willingness to try a given food [41] and food liking [42,43] are known to increase with repeated exposure. Preference ratings such as ours, which were collected after the initial HHFKA-adjustment period, provide insight beyond that which was obtained during the early phases of implementation. The uncovering of a relationship between meal participation and food preferences is novel and adds to the body of knowledge. While significant, many of the differences in liking scores may be considered small. However, one will note that the range for mean liking scores was relatively narrow at 1.84 for breakfast item scores and 1.92 for lunch item scores. In addition, of the 25 lunch items, the lowest scoring “top five” item was cheese pizza sticks with marinara sauce (M = 5.38), while the highest scoring “bottom five” item was cheesy nacho bake (M = 4.53). The difference between the mean values for these lunch items was 0.85. Similarly, for the 23 core breakfast items, the lowest scoring “top five” item was cereal (M = 5.69), while the highest scoring “bottom five” item was egg and cheese on English muffin (M = 4.68) with the difference between these values at 1.01. Understandably, larger differences in liking have greater “real-world” meaning for school nutrition programs. Nonetheless, establishing a meaningful size cut-point for these differences is difficult, and this is particularly so when ordinal scales (e.g., Likert scales) are utilized. As such, the number of students impacted and the decisions at stake should all be taken into consideration when establishing cut-points. Furthermore, because this was an observational study, causal relationships cannot be inferred. For school nutrition programs, interpretations of our results should be made with this limitation and the broader body of knowledge in mind. Although beyond our control, the inability to collect demographic information beyond grade level and meal participation category is a limitation of this study. Reliance on a volunteer sample from one school district is an additional limitation which constrains generalizability to other schools. In addition, students completed the survey in a variety of environments (i.e., at home, at school during class or at school on their own time), and measures were not in place to prevent students from taking the survey more than once. Although, given that the average time to complete the survey was approximately 12 min, the occurrence of multiple survey submissions was unlikely. School nutrition programs offer many benefits, but program success is dependent on a variety of factors. The timing and the length of the lunch period, school policies, and parental perceptions, are all associated with school nutrition program outcomes. In this survey of students, both grade level and meal participation category were found to be related to the trying and the liking of foods. While additional research is needed, the results of this study build upon that which is already known about the relationship between socioeconomic and demographic characteristics and child nutrition program outcomes. Future studies should consider the inclusion of both grade level and meal participation category when exploring student preferences. School nutrition programs should also consider these factors when assessing student satisfaction. Conceptualization, J.H. and J.E.; methodology, M.S.-S. and J.E.; software, J.E.; formal analysis, J.E.; investigation, M.S.-S. and J.E.; resources, M.S.-S.; data curation, J.E.; writing—original draft preparation, J.H., J.E.; writing—review and editing, J.H., J.E. and M.S.-S.; visualization, J.E.; supervision, M.S.-S.; project administration, M.S.-S.; All authors have read and agreed to the published version of the manuscript.This research received no external funding.We thank Grace Duebler for her assistance with technical support in administering the survey.The authors declare no conflict of interest.Number of foods tried by different grade groups within meal service participation level. Values are means with 95% CI indicated by vertical bars.Number of foods liked, adjusted for number of foods tried, by different age groups within meal service participation. Values are means with 95% CI indicated by vertical bars.Number of foods disliked, adjusted for number of foods tried, by different age groups within meal service participation. Values are means with 95% CI indicated by vertical bars.Mean liking 1 of items in each food category 2 (by grade group and meal service participation).1. Response scale for each item is 1–7, 1 (), 2 (), 3 (), 4 (), 5 (), 6 (), 7 (). 2. Mean values within a row with unlike letters were significantly different (p < 0.05).

Med-MDPI/ijerph_5/ijerph-17-16-05642.txt

@@ -0,0 +1 @@
+Contamination of the water and sediment with per- and polyfluoroalkyl substances (PFAS) was studied for the lake impacted by the release of PFAS-containing aqueous film forming foam (AFFF). PFAS concentrations were analyzed in lake water and sediment core samples. ΣPFAS concentrations were in the range of 95–100 ng L−1 in the lake water and 3.0–61 µg kg−1 dry weight (dw) in sediment core samples, both dominated by perfluorohexane sulfonate, perfluorooctane sulfonate; 6:2 fluortelomer sulfonate was inconsistently present in water and sediment core samples. The sediment–water partitioning coefficients (log Kd) were estimated and ranged 0.6–2.3 L kg−1 for individual perfluoroalkyl carboxylates (PFCAs) and 0.9–5.6 L kg−1 for individual perfluoroalkane sulfonates (PFSAs). The influence of the sediment inorganic content and organic matter on PFAS distribution was investigated. In studied sediments, the mineral content (corresponding to <5% of the bulk media mass) was mainly represented by sulfur, iron and calcium. The PFAS distribution was found strongly connected to the sediment mineral content (i.e., Fe, Pb, Rb and As), whereas the sediment organic carbon content did not to have a direct influence on the PFAS distribution. The aim of this study was to improve our understanding of the PFAS distribution in the natural heterogeneous media.Per- and polyfluoroalkyl substances (PFAS) have frequently been in focus due to their persistent, bio-accumulative and potentially toxic characteristics [1,2]. The distribution of PFAS in the environment has mainly been associated with mobility and chemical stability of the substances in aquatic environment [3,4]. Important emission sources were often connected to use of the PFAS-containing aqueous film forming foam (AFFF) [4,5]. The AFFF application in extinguishing of hydrocarbon-fuel fires during training and emergency events can lead to severe contamination of the site surrounding environment [5,6].The ubiquitous presence of PFAS in soil, water and biota represents a potential hazard to nature and public health [7,8,9]. Furthermore, the source water contamination can lead to continues human exposure via drinking water [10,11]. The population exposure and related health effects are yet subject to further investigation [12,13,14], not least concerning AFFF. In Sweden, approximately one-third of the airfields were connected to the source water contamination due to historical AFFF emission [15,16]. In present study, the investigated area represents the case of severe drinking water contamination, with over 16,000 individuals exposed among the population between the 1980s and 2013 [17,18]. The health-related risks have been under extensive investigation, including the assessment of exposure conditions and associated PFAS levels in human serum [13,19]. A better understanding of the PFAS distribution in the area was needed.The retrospective estimates on contamination levels in drinking water are important for the human exposure investigation, in particular for the dose–response relationships [14,19]. PFAS mass balance and transport have been studied for surface water [20,21], groundwater [14,22] and water treatment plants [23,24]. However, there is a certain lack of understanding in interpretation of the retardation factors; for the historical AFFF emission scenarios, the lack of data on both emission rates and periods causes a significant uncertainty in estimates. Field-derived distribution predictors for PFAS are often estimated for different environmental and emission conditions, and show a large variation between sites [25,26,27].From the environmental perspective, the surfactant nature of PFAS and low concentrations in the carrier medium make it difficult to predict the PFAS transport [27,28]. Laboratory studies have shown that PFAS distribution in aqueous–solid interfaces can be strongly affected by the ionic composition of the aqueous phase [29,30]. The intermolecular interaction between PFAS and solids has been reported as electrostatic and dependent on the solid surface characteristics [29,31]. Furthermore, the molecular chain-length and functional groups have been shown to have an effect on interactions with media [30,32,33]. In the natural heterogeneous media, surfaces of the solids as well as suspended solids, represent a complicated interaction system, including the ionic interactions and impact of the zeta-potential of the system [34,35]. The PFAS interaction with media organic content has been subjected to the hydrophobic interactions and often suggested as a primary driving factor for sorption [29,33]. However, it was also shown that organic matter composition can have a significant impact on PFAS distribution (in connection to PFAS chain-length and functional group) [30].The scientific data on PFAS in the natural heterogeneous media is still limited and further investigation is required. Understanding of the PFAS behavior in the natural aqueous–solid interfaces is important for the reliability of transport and fate estimates. The detailed investigation is necessary for the verification of the spatial and temporal distribution factors. This is of high importance for far-flied transport prediction and risk assessments.The objective of the present study was to investigate the vertical PFAS distribution in the sediment cores extracted from the contaminated lake. The specific objectives were: (i) to assess the water and sediment contamination levels; (ii) to evaluate the sediment–water distribution coefficients; and (iii) to investigate the role and impact of the sediment related characteristics on PFAS distribution in the sediment column.The studied Lake Sänksjön is located approximately 100 m northeast from the F17 airfield (Blekinge Air Force Wing) near Ronneby in southern Sweden (Figure 1a). Sänksjön has suggestively been contaminated due to release of the PFAS-containing AFFF at the airfield territory. AFFF was used by the airfield fire brigade for fire training exercises between the 1980s and 2000s (estimated). The fire training activities were mainly conducted for simulation of the aircraft crash and rescue missions (according to former personal). This involved extinguishing of real scale hydrocarbon fuel fires.There were two main emission sources suggested by Ronneby municipality: fire training facility 900 m south and fire station 1200 m southwest of the Sänksjön (Figure 1b). Due to complex topographic and hydrogeological features of the area, it was difficult to conclude an exact PFAS emission source for the lake. According to the Swedish Geological Survey, the lake body confines an apparent connection to the underlying groundwater (Figure 1b). However, despite PFAS contamination, the interaction with groundwater is unlikely due to groundwater flow direction and lake depth (Figure 1b,c). Therefore, the PFAS transport to Sänksjön was suggestively associated with surface run-off.Lake water depth was measured using GPS receiver coupled with acoustic sonar on 16 June, 2017; measurements were taken at 18 locations with average water depth of 1.7 ± 0.34 m (Figure 1c).Duplicate water samples (bulk water) were collected at Locations F and G (corresponding to north and south of the lake, respectively) at 1.6 m (F, n = 2) and 2.2 m (G, n = 2) water depth using 1 L polypropylene bottles and manual grab sampler on 20–21 June, 2016 (Figure 1c).Sediment core samples were collected at Locations E–G (corresponding to center, north and south of the lake, respectively) at 1.6 m (E), 1.6 m (F) and 2.2 m (G) depth on 20–21 June 2016. Sediment cores were extracted from the lakebed (in acrylic tube) using manual core sampler. Each sediment core was gently ejected (on site) from the tube (using vertical stand with threaded mechanism), sliced (using acrylic slicer) and transferred into polypropylene jars. Sediment Cores E–G (with 0.34, 0.42 and 0.39 m of depth, respectively), were distributed in segments of 2 cm (n = 17, Core E) and 3 cm (n = 14, Core F; n = 13, Core G).All samples were stored at 3 °C (water) and −20 °C (sediment) prior to extraction and analysis. All sampling containers (polypropylene bottles, jars, acrylic slicer and core tubes) were pre-rinsed with methanol (×5). Sediment sampler components (slicer and core tubes) were rinsed on site with Milli-Q water (×5) and methanol (×15) prior to each core sampling.In this study, 26 PFAS were analyzed including four perfluoroalkane sulfonates (C4,6,8,10 PFSAs) (PFBS, PFHxS, PFOS and PFDS), 13 perfluoroalkyl carboxylates (C3–13,15,17 PFCAs) (PFBA, PFPeA, PFHxA, PFHpA, PFOA, PFNA, PFDA, PFUnDA, PFDoDA, PFTriDA, PFTeDA, PFHxDA and PFOcDA), three perfluorooctane sulfonamides (FOSAs) (FOSA, MeFOSA and EtFOSA), two perfluorooctane sulfonamidoethanols (FOSEs) (MeFOSE and EtFOSE), three perfluorooctane sulfonamidoacetic acids (FOSAAs) (FOSAA, MeFOSAA and EtFOSAA) and one fluorotelomer carboxylate (6:2 FTSA). In addition, 16 internal standards (i.e., 13C8-FOSA, d3-MeFOSAA, d5-EtFOSAA, d3-MeFOSA, d5-EtFOSA, d7-MeFOSE, d9-EtFOSE, 13C4-PFBA, 13C2-PFHxA, 13C4-PFOA, 13C5-PFNA, 13C2-PFDA, 13C2-PFUnDA, 13C2-PFDoDA, 18O2-PFHxS and 13C4-PFOS), and one injection standard (13C8-PFOA) were used.The analytical procedures of the sample extraction for water and sediment and PFAS analysis were performed as described elsewhere [36,37,38]. Extracted samples were analyzed using high-performance liquid chromatography coupled to tandem mass-spectrometry (6460 Triple Quadrupole LC/MS System, Agilent Technologies, Santa Clara, CA, USA). Betasil C18 LC column (50 × 2.1 mm, 5 µm particle size, Thermo Fisher Scientific, Waltham, MA, USA) and Hypersil Gold pre-column (10 × 2.1 mm, 5 µm particle size, Thermo Fisher Scientific, Waltham, MA, USA) were used as analytical and guard columns, respectively. The branched isomer concentration of PFHxS and PFOS (i.e., B-PFHxS and B-PFOS) was estimated using the response factors of the respective linear isomers (i.e., L-PFHxS and L-PFOS, respectively).In total, 4 water and 44 sediment samples were analyzed. Procedural blanks were applied in duplicate for each sediment core batch (n = 6) and once for each duplicate water sample (n = 4). The method detection limits (MDLs) were determined at an S/N of 3, and ranged 0.04–0.05 ng L−1 for water and 0.03–0.4 µg kg−1 dry weight (dw) for sediment.The field-derived sediment–water partitioning coefficients (Kd) and carbon normalized sediment–water partitioning coefficients (KOC) were calculated from bulk sediment (ng kg−1 dw) and bulk water concentrations (ng L−1) as described earlier [36].The fraction organic carbon (fOC) was determined on the replicate sediment core samples by combustion method (disaggregated samples were dried at 105 °C and burned at 1350 °C in a furnace). Sediment bulk and dry bulk densities were determined from direct measurements on wet and dehydrated samples (see Table S1).Sediment elemental analysis was performed on the replicate sediment core samples (F and G) using Niton XL3t X-ray fluorescence (XRF) analyser (Thermo Fisher Scientific, Waltham, MA, USA) in soil mode. Samples were prepared and analyzed in compliance with the US EPA method 6200 [39]. Dehydrated (freeze-dried) sediment samples were homogenized, weighed and distributed into XRF sample cups (Premier Lab Supply, Port St. Lucie, FL, USA). In XRF cups, sediment was compressed between Prolene film (Chemplex Industries inc., Palm City, FL, USA) and glass fiber filter (Advantec, Tokyo, Japan). Polyester fiber wool was used as a dumper material (for details see Table S2).To improve volumetric representation, each sample was scanned two times with X-ray beam collimated at the centre of the sample, and deviating from the center (5–7 mm). To secure repeatability, the XRF instrument was locked in a fixed position and all samples were centered and scanned in an exact manner.In total, there were twenty-seven sediment samples analyzed. Negative blank and positive reference samples (standard reference material NIST2709a, certified by Rigaku, Tokyo, Japan) were applied for every fifth sample (n = 16) and each core sequence (n = 2), respectively. Measurements of positive reference samples showed a good agreement with certified concentrations for NIST2709a (see Table S3). The sediment sample concentrations were accordingly adjusted to the averaged negative blank levels.For measured PFAS concentrations, the relationships with sediment fraction organic carbon were studied using Pearson’s pair-wise correlation (at 95% confidence interval) on data (n = 44) including sediment Cores E–G. The relationships between measured sediment elemental content and PFAS concentrations were studied using Pearson correlation, Spearman correlation and Principal Component Analysis (PCA) on standardized data (n = 26) including sediment Cores F and G. Related calculations and data evaluation were compiled in MATLAB (MathWorks, Natick, MA USA).In the Lake Sänksjön water (Locations F and G), 8 of 26 investigated PFAS were detected (i.e., PFHxA, PFHpA, PFOA, PFNA, PFBS, PFHxS, PFOS and 6:2 FTSA) (Tables S4 and S5). The ΣPFAS levels were similar at Locations F and G (95 and 100 ng L−1, respectively) and dominated by PFSAs, with a contribution of 84–90% for sum of PFHxS and PFOS (including both linear and branched forms).In sediment samples, the ΣPFAS concentration ranged 3.4–25, 4.9–38 and 3–61 µg kg−1 dw in the sediment Cores E–G, respectively (Tables S6–S8). The PFAS composition was similar in the sediment cores and dominated by PFHxS and PFOS (sum of linear and branched isomers) with the average contribution (n = 44) of 32 ± 11% for PFHxS and 22 ± 16% for PFOS, followed by 6:2 FTSA (21 ± 20%) and PFHxA (14 ± 9%) (Figure 2).For the remaining PFCAs (PFHpA, PFNA, PFDA, PFUnDA and PFDoDa) and FOSAAs (MeFOSAA and EtFOSAA), the overall contribution was insignificant, as these compounds were inconsistently present in top sediment layers (0–12 cm) with total concentration of 0.6 (E), 2.4 (F) and 2.1 (G) µg kg−1 dw.The field-derived sediment–water partitioning coefficient Kd and organic carbon normalized KOC were calculated for sediment samples from Cores F and G. The sediment–water partitioning coefficients (Kd) ranged 0.6–2.3 L kg−1 for individual perfluoroalkyl carboxylates (PFCAs) and 0.9–5.6 L kg−1 for individual perfluoroalkane sulfonates (PFSAs). Overall, log Kd and log KOC values for PFHxA, PFOA, PFBS, PFHxS and PFOS were consistent between F and G, except for 6:2 FTSA (Table S9). For PFSAs, Kd and KOC values showed an increase with perfluorocarbon moiety as PFBS < B-PFHxS < L-PFOS ≈ B-PFOS, except for L-PFHxS. There was no observed relation with perfluorocarbon chain length for PFCAs.Sediment samples were identified as fine detritus gyttia (Core E) and coarse detritus gyttia (Cores F and G). Sediment densities were consistent in sediment Cores E–G with a mean density of 1.0 ± 0.06, 0.97 ± 0.05 and 0.98 ± 0.03 kg L−1, respectively. The mean fraction of organic carbon (fOC) was 0.96 ± 0.03, 0.95 ± 0.04 and 0.95 ± 0.07 for sediment Cores E–G, respectively. The mean dry bulk density (ρdry bulk), however, was slightly lower in Core F (0.03 ± 0.01 kg L−1) than in Cores E and G (0.04 ± 0.01 and 0.04 ± 0.02 kg L−1, respectively). Overall, the sediment cores were similar and considered as a homogeneous media mainly represented by organic matter (Table S1).Based on XRF analysis, the sediment mineral content in Cores F and G was represented by sulfur (12,000 ± 2900 and 13,000 ± 1900 mg kg−1 dw, respectively), iron (9100 ± 1600 and 9100 ± 2300 mg kg−1 dw, respectively) and calcium (6800 ± 1500 and 14,000 ± 2800 mg kg−1 dw, respectively) (Tables S10 and S11).The relationship between individual PFAS concentrations and sediment fraction organic carbon, densities and moisture content was studied for four PFCAs (PFHxA, PFHpA, PFOA and PFUnDA), three PFSAs (PFBS, L-PFHxS, B-PFHxS, L-PFOS and B-PFOS), MeFOSAA and 6:2 FTSA (Table S12). Out of eleven studied PFAS, only long-chained PFUnDA (r = −0.8, n = 8) and L-PFOS (r = −0.4, n = 43) showed a negative correlation (p < 0.05) with fraction organic carbon.All PFAS and PFOA showed a weak negative correlation with bulk sediment density (Table S12). For the dry bulk density, PFHxA (r = −0.4, n = 44), PFOA (r = −0.8, n = 43) and PFBS (r = −0.5, n = 44) showed a negative correlation. There was no significant correlation observed with moisture content.The correlation between individual PFAS concentrations and sediment elemental content was studied for PFHxA, PFOA, PFBS, PFBS, L-PFHxS, B-PFHxS, L-PFOS, B-PFOS and 6:2 FTSA (n = 26) (Tables S13 and S14). All PFSAs showed a positive correlation (p < 0.05) with sulfur (rs = 0.5–0.6) and titanium (rs = 0.5–0.6); moreover, for long-chained PFSAs (i.e., PFHxS and PFOS), a positive correlation was found for sediment lead (rs = 0.6–0.7), arsenic (rs = 0.6–0.7) and iron (rs = 0.5–0.6). For PFCAs, PFHxA showed a positive correlation with sediment rubidium, (rs = 0.4) lead (rs = 0.6) and arsenic (rs = 0.6) and a negative correlation with calcium (rs = −0.5). PFOA was positively correlated with sediment lead (rs = 0.5), arsenic (rs = 0.4) and titanium (rs = 0.4).Form the principal component analysis on standardized parametric data (n = 26, sediment Cores F and G), the overall data variability was sufficiently explained as 43%, 24% and 12% within the first three component spaces (see Figure S1 for contribution to the variance by each component and parameters). Within the first component space (of 43% explained) (Figure 3), the long-chain PFSAs (L-PFHxS, B-PFHxS, L-PFOS and B-PFOS) followed by sediment arsenic and lead had the significant contribution to the data variability. However, the short-chain PFBS, sediment rubidium and titanium contributions were identical on lower level. L-PFOS, sediment iron and sulfur had a similar contribution to the variance within both the first and second component spaces (with <67% variance explained). PFCAs and contravariant dry bulk density contributions were relevant within first and (on greater level) second component spaces. Contribution of the fraction organic carbon and moisture content was relevant within second component space only and represented by <24% data variability.Overall, PFSAs variation showed an alignment (correlation) with sediment arsenic, lead, rubidium, titanium and sulfur. PFCAs were negatively correlated to dry bulk density. There was no clear correlation between individual PFASs and fraction organic carbon found.The PFAS concentration and composition in water samples from Locations F and G were similar which indicates a spatially uniform emission source and an overall equilibrium of the PFAS masses in the lake water. Considering the lake volume (approximately 12 × 104 m3), the total mass of PFAS in the water phase can be estimated as 12 g absolute.The ΣPFAS concentrations in water detected in the present study were in the same range as reported for PFAS levels in surface water across Sweden [8]. The PFAS composition profile (predominated by PFHxS and PFOS) was similar to reported for sites affected by AFFF release at F18 airfield in Tullinge (suburb of Stockholm, Sweden) [7], Arlanda Stockholm Airport, Sweden [40] and Schiphol Amsterdam Airport, Netherlands [41].The vertical distribution of PFAS in the sediment column was considered as relatively even and dominated mainly by PFHxS and PFOS. However, the 6:2 FTSA concentration in sediment samples was elevated in the top layer of Core G and ubiquitously present in sediment Core E (corresponding to south and center of the lake, respectively) (Figure 1). This, in agreement with elevated 6:2 FTSA concentration in the corresponding water samples (8.6 ng L−1 at Location G) may indicate the recent emission or source located in south of the lake (Figure 1).There is limited data available for PFAS concentrations in lake sediment and comparison by both concertation and composition is difficult. In the present study, PFAS concentrations (dw) in sediment were similar to those reported for Schiphol Amsterdam Airport, whereas PFHxS concentrations were one order of magnitude higher [41]. The PFAS composition was similar to sediments from Lake Halmsjön [40]. The elevated 6:2 FTSA levels were similar to sediment from Lake Langavatnet [42]. ∑PFAS concentrations in surface sediment (dw) were 1–10 times higher than reported for Laurentian Great Lakes [26].The vertical distribution data can be very useful for understanding of the sorption and transport processes. However, an exact interpretation of the mass fluxes requires the knowledge on the sediment accumulation rates and temporal boundaries. Unfortunately, the temporal distribution was not possible to determine in present study. There were two independent radioisotope analysis attempts carried out: Pb-210/Ra-226 and Pb-210 analysis (in 2016 and 2017). Due to very low radioisotope signal in sediment, the analysis results were insufficient for an adequate interpretation. The gradient in PFAS concentrations from the bottom to the top of the core indicates both increase in contaminant input (over time) and sorption (retardation) in the vertical transport processes. The MAD (average (mean) absolute deviation) in PFOS and PFHxS concentrations (2.4 and 1.9 (E), 2.8 and 2.3 (F) and 2.5 and 2.1 (G)) was one order of magnitude higher than in PFBS (0.6 (E), 0.12 (F) and 0.1 (G)). This indicates a certain sorption (retardation) in the vertical transport process, with a tendency related to chain the lengths as PFOS > PFHxS > PFBS.In the present study, field-derived log Kd values were slightly higher than reported for corresponding PFAS in lake sediments at Stockholm Arlanda Airport [40] and river sediments at Schiphol Amsterdam Airport [41]. For PFOS, log Kd values were about two times higher than reported for marine sediments [25,43]. It is important to note that PFAS concentrations in sediment can be a result of a recent release or historical emission [7,27,44]. Hence, unless the spatial and temporal conditions are established, field-derived Kd values should be considered with a certain precaution and the local equilibrium conditions have to be considered.The KOC values were generally higher than Kd values (Table S9), indicating an affinity of PFAS to organic carbon, which agrees with previous studies [25,26,29]. However, it was suggested to consider the field-derived KOC values with a certain precaution. Due to the surfactant nature of PFAS and related interaction mechanisms with surfaces, the organic carbon normalized KOC might not fully represent the media [28,35].The sediment inorganic content was measured in bulk dried samples and can represent both matter of the solids and corresponding metal–ligand complexes in aqueous phase [34]. In the present study, the measured sediment elemental content was primarily subjected to the solid phase due to no significant correlation with sediment dry bulk density and moisture content. Previous studies have shown that sorption of PFAS is impacted by pH and suggested to decrease in Kd values with increasing pH, which is most likely due to pH depending changes in zeta potential of the solid surface [33,45]. It was suggested that the partitioning of PFAS (in particular long-chained PFSAs) in natural media could be affected by the presence of metal oxides and metal-ligand complexes (carbonate, sulfate or phosphate ligands) in aqueous phase. This is however beyond the scope of the present study.In studied sediments, long-chain PFCAs and PFSAs showed significant correlation with sediment lead, arsenic, iron, titanium and sulfur (p < 0.05, Tables S13 and S14). This, in agreement with previous studies, indicates the major effect of the electrostatic interaction with the mineral content of the media [29,31]. The slightly stronger correlation with sediment dry bulk densities (rs = −0.6–0.7) (Table S13) might also indicate the association with aqueous phase and the mechanical impact of media pore space on PFAS distribution.The PFAS interaction with the solid organic matter (in dissolved–solid organic matter interface) has been previously reported as hydrophobic [29]. Although, the PFAS sorption on organic porous media can be affected by the organic matter composition [30]. In the present study, derived Kd and KOC predictors for PFAS, reflected an increased association with organic matter (Table S9). However, further investigation has shown that the organic matter (expressed as fOC) had no significant effect on PFAS distribution regardless of the functional group or chain-length. Moreover, PFAS distribution was strongly affected by the sediment inorganic content representing <5% of the bulk media. To the best of the authors’ knowledge, this is the first comprehensive study addressing the influence of the sediment inorganic vs. organic content on PFAS distribution in sediment.Ultimately, the association of PFAS with the solid phase is a complex process that is impacted by the physicochemical properties of PFAS, ionic composition of the aqueous phase, porous media surface charges, structure and composition [30,31,45]. Furthermore, a better understanding is needed in PFAS sorption mechanism, in particular on intermolecular interaction with natural media surfaces and structure.PFAS concentrations in the water and sediment core samples were studied for the Lake Sänksjön. In total, eight (out of 26 investigated) different PFAS were detected in the lake water and thirteen different PFAS in the sediment core samples. The PFAS composition in sediment and water was predominated by PFOS and PFHxS; 6:2 FTSA was inconsistently present in water and sediment core samples. The field-derived PFAS partitioning coefficients (log Kd and log Koc) were identical between compared lake sediment cores (F and G) and showed no apparent trends in relation to perfluorocarbon chain length.Studied sediment cores were identical in density and high organic matter content. The sediment inorganic content was represented by sulfur, iron, calcium, titanium, lead, arsenic and rubidium. PFAS distribution in sediment was strongly connected to the sediment mineral content (i.e., Fe, Pb, Rb and As). This, in connection to previously reported hydrophobic interaction with media organic matter, indicates the significance of the electrostatic interactions (with media inorganic matter) in sorption processes for PFAS. It was suggested that PFAS sorption on natural heterogeneous media should be addressed with a precaution, in particular for considerations of hydrophobic sorption on media organic matter. Although the overall mass distribution can be connected to the mechanistic retardation processes, the actual physical sorption can be affected by the number of external and media specific parameters. Present findings are of importance for assessment of the PFAS distribution and retardation factors in the natural heterogeneous media as well as for far-field transport estimates.The following are available online at https://www.mdpi.com/1660-4601/17/16/5642/s1, Table S1: Densities, fraction organic carbon and moisture content in sediment cores E, F and G respectively, Table S2: Materials description for XRF analysis samples, Table S3: XRF analysis measurement agreement for measured and certified standard, Table S4: PFAS concentrations in water from location F, Table S5: PFAS concentrations in water from location G, Table S6: Individual PFAS concentrations in sediment core E, Table S7: Individual PFAS concentrations in sediment core F, Table S8: Individual PFAS concentrations in sediment core G, Table S9: Sediment-water partitioning (Kd) and organic carbon normalized (KOC) coefficients in sediment core F and G, Table S10: Elemental composition of the sediment core F, Table S11: Elemental composition of the sediment core G, Table S12: Correlation matrix for Pearson correlation and corresponding p-values on PFAS vs. sediment densities, fraction organic carbon and moisture content, Table S13: Correlation matrix for Spearman’s correlation and corresponding p-values on sediment elemental content vs. PFAS, sediment densities, fraction organic carbon and moisture content, Table S14: Correlation matrix for Pearson correlation and corresponding p-values on sediment elemental content vs. PFAS, sediment densities, fraction organic carbon and moisture content, Figure S1: PCA: variance explained by components and contribution of parameters to component variance.Conceptualization, D.M. and L.A.; methodology, D.M., L.A., K.N. and T.I.; formal analysis, D.M., L.A. and T.I.; investigation, D.M.; writing—original draft, D.M.; writing—review & editing, K.M.P., R.B., L.A. and K.N.; supervision, K.M.P., R.B., L.A. and K.N. All authors have read and agreed to the published version of the manuscript.The research was partially funded by Japan Society for the Promotion of Science (JSPS) (ID: SP19502). The APC was funded by the Division of Water Resources Engineering, LTH Lund University.The authors would like to thank Kent Broström (Ronneby Miljö & Teknik AB), Maria Enarsson and Linda Petersson (Blekinge Air Force Wing, Ronneby) for help with field investigation and access to the sites, and Mats Rundgren (Department of Geology, Lund University) for help with identification of the sediments and suggestions with analysis.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.Study site description: (a) site location; (b) area topography and hydrogeology (including groundwater flow boundaries (purple dot-dash), flow direction (blue arrows) and reservoir (blue area)), airfield territory (solid orange), AFFF emission sources (fire station and fire training facility) and lake location (black box); and (c) lake bathymetry and sampling locations. GIS data: GSD Terrängkartan vektor, Lantmäteriet (base map); GSD Höjddata, Lantmäteriet (topography) and SGU Grundvattenmagasin and Grundvatten, SGU (hydrogeology).PFAS concentration (µg kg−1 dw) and relative composition profile of lake sediment Cores E–G; “not detected” includes concentrations of PFHpA, PFNA, PFDA, PFUnDA, PFDoDa, MeFOSAA, EtFOSAA and 0.5 MDL for the undetected PFAS.Bi-plot on parameter contribution the data variance within the first (43% explained) and second (24% explained) component spaces, including PFCAs and FTSA (red); PFSAs (blue); sediment iron, lead, rubidium, arsenic and titanium (black); sediment sulfur and calcium (yellow); sediment densities (ρdry bulk and ρbulk) and moisture content (MC) (magenta); and fraction organic carbon (fOC, green).

Med-MDPI/ijerph_5/ijerph-17-16-05643.txt

@@ -0,0 +1 @@
+Background: Aerobic exercise is known to reduce arterial stiffness; however, high-intensity resistance exercise is associated with increased arterial stiffness. Stretching exercises are another exercise modality, and their effect on arterial stiffness remains unclear. The purpose of this study was to determine whether stretching exercises reduce arterial stiffness in middle-aged and older adults, performing the first meta-analysis of currently available studies. Methods: We searched the literature for randomized controlled trials (RCTs) and non-RCTs published up to January 2020 describing middle-aged and older adults who participated in a stretching intervention vs. controls without exercise training. The primary and secondary outcomes were changes in arterial stiffness and vascular endothelial function and hemodynamic status. Pooled mean differences (MDs) and standard MDs (SMDs) with 95% confidence intervals (CIs) between the intervention and control groups were calculated using a random effects model. Results: We identified 69 trials and, after an assessment of relevance, eight trials, including a combined total of 213 subjects, were analyzed. Muscle stretching exercises were shown to significantly reduce arterial stiffness and improve vascular endothelial function (SMD: −1.00, 95% CI: −1.57 to −0.44, p = 0.0004; SMD: 1.15, 95% CI: 0.26 to 2.03, p = 0.01, respectively). Resting heart rate (HR) and diastolic blood pressure (DBP) decreased significantly after stretching exercise intervention (MD: −0.95 beats/min, 95% CI: −1.67 to −0.23 beats/min, p = 0.009; MD: −2.72 mm Hg, 95% CI: −4.01 to −1.43 mm Hg, p < 0.0001, respectively) Conclusions: Our analyses suggest that stretching exercises reduce arterial stiffness, HR, and DBP, and improve vascular endothelial function in middle-aged and older adults.Vascular aging results in stiffer arteries and vascular endothelial dysfunction, and may have a role in the development of cardiovascular disease [1]. In particular, excess reactive oxygen species production by mitochondria is a key mechanism of aging and age-related vascular dysfunction [2]. Physical inactivity is an independent risk factor for the deterioration of vascular function, atherosclerosis, and cardiovascular diseases [3]. Regular exercise training leads to the prevention of cardiovascular diseases and mortality. Aerobic exercise is known to significantly reduce large artery compliance, one of the parameters of arterial stiffness, in middle-aged and older humans [4]. On the other hand, a previous meta-analysis has shown that resistance exercise does not decrease arterial stiffness in middle-aged subjects [5].Stretching exercises are another exercise modality along with aerobic and resistance exercises. Stretching exercise has been widely utilized as a warm-up prior to main exercise sessions [6]. A study reported that stretching exercises reduced arterial stiffness and improved endothelial function in middle-aged women [7]. Although the results of a few studies have been published to date, no meta-analysis has been conducted, and it remains unclear whether stretching exercises reduce arterial stiffness in middle-aged and older adults.The purpose of this study was to determine whether stretching exercises improve select markers of vascular function, including endothelial function and arterial stiffness in middle-aged and older adults, performing the first meta-analysis of currently available studies.Studies evaluating the effect of stretching exercises on arterial stiffness, endothelial function, and/or hemodynamic status were searched through databases (MEDLINE and EMBASE) using web-based search engines (PubMed and OVID). We started to search the databases on 10th January 2020. In addition, relevant studies were manually identified from the references of initially identified articles and relevant reviews and commentaries. We used the following search terms: stretching, stretching exercises, stretching training, stretching intervention, arterial stiffness, vascular stiffness, pulse wave velocity, pulse wave analysis, endothelial function, and hemodynamic status. The search was limited to human studies in English. Furthermore, conference proceedings were also searched manually in an attempt to identify relevant unpublished studies.There was no ethical approval because this study did not include confidential personal data and did not involve patient intervention.Studies meeting the following criteria were included: the designs were randomized controlled trials (RCTs) or non-randomized controlled trials (non-RCTs); the study populations were middle-aged and older adults; exercise intervention groups were allocated to perform stretching exercise training as a physical exercise therapy with no other combined training; the control groups received instruction to continue their regular lifestyle habits; the durations of interventions were ≥4 weeks; outcomes included arterial stiffness, endothelial function, or hemodynamic status. We defined middle-aged and older adults as subjects whose age was 40 years or older. Three reviewers (M.K., Y.K., and T.T.) each reviewed all the eligible trials and determined whether they fulfilled the selection criteria. Disagreements were resolved by discussion.The following data were extracted from each report: study design, the number of subjects, baseline participant characteristics including age, gender, and body mass index (BMI), and details of the stretching exercise intervention (stretched muscle, time per session, frequency, duration of intervention, and intensity). Data were extracted in duplicate by two investigators (A.K. and M.K.) and verified independently by a third (H.T.).The primary outcome was changes in arterial stiffness. Brachial–ankle pulse wave velocity (baPWV), cardio-ankle vascular index (CAVI), or augmentation index (AIx) was used as a parameter of arterial stiffness [8,9,10]. The secondary outcomes were changes in endothelial function, hemodynamic status, and muscle flexibility. The reactive hyperemia peripheral arterial tonometry (RH-PAT) index or flow-mediated dilation (FMD) was used as a parameter of endothelial function, while heart rate (HR) and blood pressure were used as parameters of hemodynamic status [6,11]. A sit-and-reach test was used for measuring muscle flexibility.The risk of bias for each study was assessed by two investigators (M.K. and H.T.) using the risk of bias tool in the Cochrane Handbook for Systematic Reviews of Interventions for RCTs and the Risk of Bias Assessment tool for Non-randomized Studies (RoBANS) for non-RCTs [12,13].Continuous outcome measures were expressed as a change in the mean ± standard deviation (SD) from baseline to follow-up and were pooled as the mean difference (MD) or standardized mean difference (SMD) with a 95% confidence interval (CI). The SMD is used as a summary statistic in meta-analyses when the studies all assess the same outcome, but it is measured in a variety of ways (for example, if the primary outcome is arterial stiffness, studies use different parameters such as baPWV, CAVI, or AIx). Statistical heterogeneity was evaluated according to the Higgins I2 statistic. I2 values of 0 to 24.9%, 25% to 49.9%, 50% to 74.9%, and 75% to 100% were considered no, low, moderate, and high statistical heterogeneity, respectively [14]. A sensitivity analysis was performed to assess the contribution of only RCTs to the pooled estimate by excluding non-RCTs and recalculating the pooled SMD estimates for the remaining studies. In addition, other sensitivity analyses were performed as needed to enhance the results. Funnel plots were used to examine the presence of publication bias. A p-value of less than 0.05 was considered statistically significant. Analyses were carried out using Review Manager (version 5.3; The Cochrane Collaboration, London, UK).According to the search strategies, a total of 702 records were identified. Studies that did not meet the inclusion criteria, duplicate studies, studies with no reported change in outcome, and studies for which the full text was unavailable were excluded. From 69 potentially relevant citations retrieved from electronic database searches and manual searches of the reference lists, six RCTs [7,15,16,17,18,19] and two non-RCTs [20,21] fulfilled the inclusion criteria (Figure 1). The included studies had a total of 213 participants. All the studies included were designed to compare stretching exercises without combined training to the continuation of the subjects’ regular lifestyle habits (the control).The baseline patient characteristics of the included studies are presented in Table 1. The sample sizes, mean age, and mean BMI ranged from 13–50, 44–71 years old, and 21.2–34.1 kg/m2, respectively. The proportion of females ranged from 0 to 100%. One of the eight studies included subjects with stable chronic heart failure [18], one included subjects with stable peripheral artery disease [19], and one included subjects with metabolic syndrome and chronic diseases [21]. The remaining five studies included subjects without cardiovascular disease or lifestyle-related diseases.Table 1 also shows the details of the stretching exercise interventions of the included studies. The training types were an active static stretching intervention for major muscle groups in six studies, passive static intervention using a splint in one study, and combined intervention in one study. The training intensities were moderate (11 to 15 on the rating of perceived exertion (RPE) scale; the point of minimal discomfort, without pain) in six studies, and high (the point of maximal exertion defined as RPE > 18) in two studies. Training session duration, frequency, and the duration of the exercise training intervention ranged from 15−60 min/session, 3−7 times/week, and 4−12 weeks, respectively.The risk of bias is summarized in Table 2. Although the subjects of all eight studies were not blinded, we considered that arterial stiffness, the primary outcome in this analysis, was not likely to be influenced by the lack of blinding. Therefore, we judged that the risk of bias related to the “blinding of participants and personnel” and “blinding of outcome assessment” was low.Five studies evaluated the primary outcome, the effect of stretching exercises on arterial stiffness, and were included in the analysis (Table 3). One study assessed two different measurements of arterial stiffness, baPWV and AIx [15]. These values were extracted separately as Wong-1 and Wong-2. Similarly, another study measured baPWV and CAVI, which are both measurements of arterial stiffness [20]. These values were also extracted separately as Nishiwaki-1 and Nishiwaki-2. Primary analysis for arterial stiffness was conducted with five studies. For two studies that used more than one measurement for arterial stiffness, we used the data for baPWV because baPWV is the standard assessment for arterial stiffness. Four of the five studies in the primary analysis used baPWV and one study used CAVI. Arterial stiffness was significantly reduced by stretching exercises compared with the control (SMD: −1.00, 95% CI: −1.57 to −0.44, p = 0.0004, I2: 52%, Figure 2).As Nishiwaki’s and Kim’s studies were non-RCTs, sensitivity analysis-1 was performed excluding these two studies. Sensitivity analysis-1 with only RCTs also showed that stretching exercises significantly reduced arterial stiffness compared with the control (SMD: −0.90, 95% CI: −1.63 to −0.16, p = 0.02, I2: 63%; Figure 3). Sensitivity analysis-2 was conducted using the non-baPWV measurements for the two studies with more than one arterial stiffness measurement. Five studies were included in sensitivity analysis-2 with two studies using baPWV, two studies using CAVI, and one study using AIx. Sensitivity analysis-2 also showed that stretching exercises significantly reduced arterial stiffness compared with the control (SMD: −1.19, 95% CI: −1.81 to −0.58, p = 0.0001, I2: 58%; Figure 4).We performed sensitivity analysis-3 with only studies using baPWV. Sensitivity analysis-3 showed that stretching exercises also significantly reduced baPWV compared with the control (MD: −77.50 cm/sec, 95% CI: −135.48 to −19.53 cm/sec, p = 0.009, I2: 94%; Figure 5).Finally, we performed sensitivity analysis-4 with only female subjects to explore the potential gender influence. Arterial stiffness was significantly reduced by stretching exercises compared with the control among female subjects (SMD: −0.86, 95% CI: −1.41 to −0.31, p = 0.002, I2: 47%; Figure 6).Regarding the effect of stretching exercises on endothelial function, one of the secondary outcomes, three studies were included in the analysis (Figure 7, Table 3). Two studies evaluated the RH-PAT index and one study evaluated FMD. Stretching exercises were shown to significantly improve endothelial function compared with the control (SMD: 1.15, 95% CI: 0.26 to 2.03, p = 0.01, I2: 68%).To evaluate the effect of stretching exercises on hemodynamic status, another secondary outcome, six studies were included in the analyses (Figure 8, Table 4). Stretching exercises significantly decreased HR and diastolic blood pressure (DBP) compared with the control (MD: −0.95 beats/min, 95% CI: −1.67 to −0.23 beats/min, p = 0.009, I2: 0%; MD: −2.72 mm Hg, 95% CI: −4.01 to −1.43 mm Hg, p < 0.0001, I2: 0%, respectively). There was no significant difference in systolic blood pressure between the exercise and the control groups.To evaluate the effect of stretching exercises on muscle flexibility, four studies were included in the analysis. Stretching exercises significantly increased muscle flexibility compared with the control (MD: 7.18 cm, 95% CI: 1.33 to 13.02 cm, p = 0.02, I2: 87%; Figure 9).Funnel plots of the primary outcome generated were symmetric, indicating that the results of those meta-analyses were not influenced by publication bias.The results of the present meta-analysis demonstrated that stretching exercises reduced arterial stiffness and improved endothelial function compared with the control. Furthermore, stretching exercises decreased resting HR and DBP. These data are derived from a meta-analysis of six RCTs and two non-RCTs enrolling a total 213 subjects. To our knowledge, this is the first meta-analysis to assess the potential benefits of stretching exercise on arterial stiffness.The present study showed that stretching exercises improved endothelial function. Increased cardiovascular disease in aging is partly a consequence of the vascular endothelial cell senescence and associated vascular dysfunction [22]. Exercise training modifies blood flow, luminal shear stress, and tangential wall stress, all of which can improve endothelial function [23]. A recent meta-analysis demonstrated that aerobic and combined aerobic and resistance exercise increased FMD by 1.21% and 2.49%, respectively [11]. We confirmed that stretching exercises also improved endothelial function assessed by the RH-PAT index and FMD in our meta-analysis. A single session of stretching exercises improved endothelial function and peripheral circulation in patients with acute myocardial infarction. This improvement was potentially induced by increased NO production from vascular endothelial cells [6]. Cyclic stretch stimuli have been shown to upregulate the expression of endothelial nitric oxide synthase (eNOS) messenger ribonucleic acid in vascular endothelial cells in vitro [24]. Hotta et al. reported that 4 weeks of daily muscle stretching enhanced the endothelium-dependent vasodilatation of resistance arterioles in the skeletal muscle of aged rats [25]. In addition, in vitro studies showed that stretch stimuli on skeletal muscle cells increase intracellular superoxide dismutase (SOD) in skeletal muscle cells [26], resulting in decreased ROS generation. Increased eNOS and decreased ROS lead to the improvement of endothelial function [27]. Therefore, regular stretching exercises can improve endothelial function by increasing eNOS and intracellular antioxidants in skeletal muscle cells. Early atherosclerosis-related changes involve the impairment of endothelial function as an important initial step in the atherosclerotic process [28]. The impairment of endothelial dysfunction is a natural process with aging, and although its mechanism is largely unknown, a recent study shows that it may be related to the availability of epoxyeicosatrienoic acids and NO [29]. The change in endothelial function by stretching exercises may be helpful to prevent the progression of atherosclerosis.The measurement of arterial stiffness is the most common non-invasive examination method for the detection of atherosclerosis-related changes [8]. Our meta-analysis demonstrated that stretching exercises reduced arterial stiffness with not only primary analysis but also sensitivity analyses. An improvement in arterial stiffness has been reported in previous studies immediately after performing short-term static stretching [30,31]. Furthermore, yoga practice, including muscle stretching, is reported to be effective in preventing or reducing arterial stiffness in elderly hypertensive patients [32]. These studies are consistent with our results, confirming the impact of stretching exercises on arterial stiffness. According to a previous meta-analysis, aerobic exercise had a significant effect on reducing baPWV (MD: −67 cm/sec, 95% CI: −97 to −38 cm/sec) [33]. In our meta-analysis, baPWV was significantly reduced after stretching exercises compared with the control (MD: −77.50 cm/m, 95% CI: −135.48 to −19.53 cm/sec). These results suggest that stretching exercises might lead to a comparable reduction in arterial stiffness compared with aerobic exercises. Yamato et al. reported that the reduction in baPWV following stretching exercises may have been the result of a change in peripheral arterial stiffness rather than a change in central arterial stiffness, which was the case with aerobic exercises [30]. The responses of arterial stiffness were different between aerobic and stretching exercises because aerobic exercise affects all arteries via the systemic circulation, whereas stretching exercise stimulates only vasculature in the muscle mechanically [30]. Although the mechanisms of the PWV decrease is not clear, it might be associated with the improvement of endothelial function following stretching exercises. There is a strong relationship between endothelial function and the arterial stiffness index in patients with atherosclerosis. In addition, arterial stiffness is functionally determined by the vascular tone of the artery, which is partially regulated by sympathetic nerve activity [4]. Eight weeks of stretching exercise has been reported to improve cardiac autonomic modulation by decreasing sympathetic activity and increasing parasympathetic nerve tone [16]. Therefore, decreased sympathetic nervous activation induced by stretching exercises could be one of the mechanisms underlying the reduction of arterial stiffness. Arterial stiffness is reported to be significantly and positively correlated with intima–media thickness in middle-aged subjects with and without diabetes [34], indicating that hyperplastic and/or hypertrophic change in vascular smooth muscle cells may have an impact on arterial stiffness. The effect of muscle stretching on vascular wall thickness remains unknown; however, one in vitro study reported the stretch-induced hypertrophic change of vascular smooth muscle cells [35]. It remains unknown whether muscle stretching causes such vascular morphological changes; however, at least, skeletal muscle stretching does not seem to have any unfavorable effects on arterial stiffness.The present meta-analysis also demonstrated that stretching exercises decreased resting HR and DBP after the intervention period. Similarly, a study by Logan et al. showed that HR and DBP decreased by 3.31 bpm (t = 2.17, p = 0.05) and 2.13 mm Hg (t = 1.93, p = 0.07), respectively, following 20-min stretching exercises in healthy pregnant women, though they did not reach statistical significance [36]. HR and BP are controlled by the two branches of the autonomic nervous system, the sympathetic nervous system and parasympathetic nervous system. Regular exercise, like endurance training and yoga, is well known to enhance parasympathetic output [37]. As we described before, stretching exercises change cardiac autonomic modulation by increasing parasympathetic nerve tone [16]. Muscle stretching exercises are thought to be effective for regulating hemodynamics associated with changes in autonomic nerves.The present meta-analyses have several limitations. First, the numbers of included studies and patients were relatively small, with only eight studies and 213 subjects, which may not be sufficient for a valuable meta-analysis. In future meta-analyses, it is important to assess more trials and subjects to confirm the effect of stretching exercises on arterial stiffness. Second, two of the eight studies in our analyses were non-RCTs. We included these two non-RCTs in the primary analysis because only few studies were available in this area. We also conducted a sensitivity analysis with only RCTs (sensitivity analysis-1), which showed similar results with the primary analysis. Third, the evaluation methods for the primary outcome, arterial stiffness, were not uniform, including baPWV, CAVI, and AIx. As supplementary analyses, we performed several sensitivity analyses to enhance our results (sensitivity analysis-2, -3, and -4). These results were all similar to the result of the primary analysis. Fourth, we were unable to evaluate whether stretching exercise is superior to aerobic exercises in reducing arterial stiffness. As we stated in the discussion, different stimulations from aerobic and stretching exercises may have different effects on arterial stiffness. Fifth, it is unclear whether stretching exercises have long-term effects on arterial stiffness in this meta-analysis because the durations of the included studies were mostly 4–8 weeks. Shinno et al. reported that stretching exercises decreased baPWV after intervention, though a significant increase in this value was observed after 6 months of not training. Therefore, stretching exercise might not have long-term effects on arterial stiffness once it is discontinued. Finally, the present study focused on the effect of stretching exercises on arterial stiffness but did not evaluate whether this effect leads to the prevention of diseases related to arteriosclerosis. Therefore, further studies will be needed to compare the effects of stretching exercises on disease prevention.Our meta-analyses demonstrated that stretching exercises reduced arterial stiffness, HR, and DBP, and improved endothelial function, which are crucial parameters of arteriosclerosis in middle-aged and older adults.For research articles with several authors, a short paragraph specifying their individual contributions must be provided. The following statements should be used “Conceptualization, M.K.; methodology, M.K., H.T.; software, M.K., H.T.; validation, M.K., T.T., and A.K.; formal analysis, M.K., F.N.G.; investigation, M.K., A.K., T.T., Y.K.; resources, M.K., Y.K.; data curation, M.K., F.N.G.; writing—original draft preparation, M.K.; writing—review and editing, F.N.G., H.T.; K.H., A.K.; visualization, M.K., T.T, Y.K.; supervision, A.K., H.T.; project administration, M.K.; funding acquisition, M.K. All authors have read and agreed to the published version of the manuscript.The authors disclosed receipt of the following financial support for the research, authorship, and publication of this article: this research was supported by JSPS KAKENHI (grant number 18K17693).The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.Flow chart of the systematic literature search for the meta-analysis.Forest plot comparing changes in arterial stiffness following stretching exercises to non-exercise controls (primary analysis for primary outcome). SD, standard deviation; IV, inverse variance; CI, confidence interval; Std, standardized.Sensitivity analysis with only RCTs comparing changes in arterial stiffness following stretching exercises to non-exercise controls (sensitivity analysis-1 for primary outcome). SD, standard deviation; IV, inverse variance; CI, confidence interval; Std, standardized.Sensitivity analysis with different assessments of outcome comparing changes in arterial stiffness following stretching exercises to non-exercise controls (sensitivity analysis-2 for primary outcome). SD, standard deviation; IV, inverse variance; CI, confidence interval; Std, standardized.Sensitivity analysis with only studies using baPWV comparing changes in arterial stiffness following stretching exercises to non-exercise controls (sensitivity analysis-3 for primary outcome). IV, inverse variance; CI, confidence interval; Std, standardized.Sensitivity analysis-4 with only female subjects comparing changes in arterial stiffness following stretching exercises to non-exercise controls (sensitivity analysis-4 for primary outcome). IV, inverse variance; CI, confidence interval; Std, standardized.Forest plot comparing changes in endothelial function following stretching exercises to non-exercise controls. SD, standard deviation; IV, inverse variance; CI, confidence interval; Std, standardized.Forest plot comparing changes in hemodynamics following stretching exercises to non-exercise controls. SD, standard deviation; IV, inverse variance; CI, confidence interval.Forest plot comparing changes in muscle flexibility following stretching exercises to non-exercise controls. SD, standard deviation; IV, inverse variance; CI, confidence interval.Characteristics of included studies.RCT, randomized controlled trial; BMI, body mass index; wk, week; s, second; baPWV, brachial–ankle pulse wave velocity; AIx, augmentation index; NR, not reported; CAVI, cardio-ankle vascular index; RH-PAT, reactive hyperemia peripheral arterial tonometry; CHF, chronic heart failure; PAD, peripheral artery disease; FMD; flow-mediated dilation; RPE, rating of perceived exertion. Data are shown as mean ± SD.Risk of bias.Low, low risk; High, high risk.Changes in parameters of arterial stiffness and vascular endothelial function.SE, stretching exercises; CON, control; baPWV, brachial–ankle pulse wave velocity; AIx, augmentation index; CAVI, cardio-ankle vascular index; sec, second; RH-PAT, reactive hyperemia peripheral arterial tonometry; FMD; flow-mediated dilation; NR, not reported. Data are shown as mean ± standard deviation.Changes in parameters of hemodynamic status.HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; SE, stretching exercises; CON, control; NR, not reported. Data are shown as mean ± SD. Generated funnel plots of the primary outcome were symmetric, indicating that the results of those meta-analyses were not influenced by publication bias.

Med-MDPI/ijerph_5/ijerph-17-16-05644.txt

@@ -0,0 +1,2 @@
+In Italy, the COVID-19 epidemic curve started to flatten when the health system had already exceeded its capacity, raising concerns that the lockdown was indeed delayed. The aim of this study was to evaluate the health effects of late implementation of the lockdown in Italy. Using national data on the daily number of COVID-19 cases, we first estimated the effect of the lockdown, employing an interrupted time series analysis. Second, we evaluated the effect of an early lockdown on the trend of new cases, creating a counterfactual scenario where the intervention was implemented one week in advance. We then predicted the corresponding number of intensive care unit (ICU) admissions, non-ICU admissions, and deaths. Finally, we compared results under the actual and counterfactual scenarios. An early implementation of the lockdown would have avoided about 126,000 COVID-19 cases, 54,700 non-ICU admissions, 15,600 ICU admissions, and 12,800 deaths, corresponding to 60% (95%CI: 55% to 64%), 52% (95%CI: 46% to 57%), 48% (95%CI: 42% to 53%), and 44% (95%CI: 38% to 50%) reduction, respectively. We found that the late implementation of the lockdown in Italy was responsible for a substantial proportion of hospital admissions and deaths associated with the COVID-19 pandemic.In early January a novel strain of coronavirus, SARS-CoV-2, a virus which follows a human-to-human transmission, was identified in the Hubei province of China as the causative agent for a new disease later defined as Coronavirus Disease 2019 (COVID-19), a respiratory disease which is often characterized by influenza-like symptoms but which can also evolve (3–5% of the cases) into acute respiratory distress syndrome, or even sepsis, and multi-organ failure which might lead to death [1]. Starting from an outbreak in China, the scale of the emergency has rapidly grown globally, leading the World Health Organization (WHO) to declare the pandemic status on March 11th, 2020 when many countries had already introduced unprecedented physical distancing and containment measures to various extents [2]. As of May 28th, 2020 almost six million of COVID-19 cases and 361,836 deaths have been recorded worldwide [3].The effect of containment measures in curbing the COVID-19 epidemic varied among countries [4,5,6,7,8,9]. While a combination of stringent policies together with wide early-phase testing coverage and effective contact tracing strategies was effective in halting the COVID-19 epidemic in countries such as mainland China, Hong Kong, and South Korea, in others the epidemic slowed only recently [3,5,6,8]. Factors explaining differences in time patterns might be found in the readiness of government responses and in the degree of compliance of the population to the implemented policies [5,6,7,8,9].Italy, which has passed 232,000 confirmed cases and 33,000 deaths [10], is one of the most affected countries in the world so far and the first in Europe where the public health emergency rapidly escalated at the national level. On March 9th, 2020 the government ordered a national lockdown, a measure including: (a) strict home confinement of the entire population; (b) closure of all non-essential commercial activities; (c) mobility restrictions related to the involved municipalities [11]. The lockdown remained in place until May 3rd, when a slowdown of the epidemic in the different Italian regions allowed its release [12].Compared with China, Italy introduced containment measures later in the course of the national epidemic, about one month after the first COVID-19 case was reported in the country. Italy’s lockdown was enforced 13 days after the one in Hubei, when normalizing for the time when the outbreak hit 50 cases in both countries [8]. This prompted a debate, in Italy and abroad, on the causes of such a delay and on how many COVID-19 cases could have been avoided, had the lockdown been implemented earlier [13]. A formal investigation into possible government mismanagement of the COVID-19 crisis is currently ongoing [14]. The aim of this study was to evaluate the health effects of late implementation of the lockdown in Italy. For this reason, we estimated the number of deaths and hospital admissions for COVID-19 that would have occurred if the lockdown had been implemented one week earlier than it was actually enforced.In the present analysis we used data on the daily number of COVID-19 cases, hospitalized patients, and deaths recorded in Italy from February 24th, the first day national data were made available, to May 3rd, the last day of implementation of the national lockdown. Figures were provided by the official website of the Italian Department of Civil Protection [10].First, we evaluated the effect of the Italian lockdown using interrupted time series (ITS) analysis. We modeled the time-series of daily new cases, Yt, using the following quasi-Poisson regression model, accounting for the possible overdispersion of data:log(Yt) = α + β1T + β2Xt + β3T2 + et
+      where T is the time elapsed since the start of the study; T2 is the time elapsed since the implementation of lockdown (set to 0 before the lockdown); X is a dummy variable indicating the pre-lockdown period (coded 0) or the post-lockdown period (coded 1); Y is the logarithm of the number of new cases at time T; α is the intercept of the model; β1 represents the trend of new cases before the lockdown; β2 is the step change following the lockdown; β3 is the slope change following the lockdown; and et is the error term of the model. Preliminary analysis of the data suggested that no adjustment was required for autocorrelation of the error terms et. We also assumed a two-week lag between the implementation of the lockdown (March 9th) and the start of its effects (March 23rd), to take into account the COVID-19 incubation period and the diagnostic delay after symptoms onset [15].Second, we evaluated the effect of an early lockdown on the trend of new cases, creating a counterfactual scenario where the lockdown was implemented one week in advance (i.e., on March 2nd instead of March 9th).Third, based on the expected number of new cases, we predicted the corresponding number of intensive care unit (ICU) admissions, non-ICU admissions, and deaths, using a previously published mathematical model [16]. Briefly, the model simulates the progress of infected individuals between different compartments during the course of an epidemic: isolated at home, admitted in a non-ICU ward, admitted in ICU, recovered, dead. Finally, we compared the number of hospital admissions and deaths under the actual and counterfactual scenarios. All the analyses were performed using the R software (R Core Team (2013). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/).From February 24th to May 3rd, 210,717 cases of COVID-19 were observed in Italy. There was an exponential increase in the number of new COVID-19 cases until March 22nd, followed by a sharp reduction (Table 1; Figure 1). Table 1 reports estimated coefficients, while related predictions are plotted in Figure 1 together with the expected number of new cases under the counterfactual scenario. On May 3rd, the number of new cases under the counterfactual scenario was less than half than that estimated under the observed scenario.Figure 2 shows differences in the total number of cases, non-ICU admissions, ICU admissions, and deaths under the two scenarios. The plots show that an early implementation of the lockdown would have averted about 126,000 COVID-19 cases, 54,700 non-ICU admissions, 15,600 ICU admissions, and 12,800 deaths. On the relative scale, this corresponds to a reduction of 60% (95%CI: 55% to 64%), 52% (95%CI: 46% to 57%), 48% (95%CI: 42% to 53%) and 44% (95%CI: 38% to 50%), respectively (Table 2).Moreover, the maximum hospital demand would have been much lower under the counterfactual scenario. The peak number of non-ICU admissions would have been 14,336 rather than 29,010 (−51%; 95% CI: −45% to −56%). A similar reduction would be expected for ICU admissions as well (2300 vs. 4068 beds; −44%, 95%CI: −38% to −49%).In Italy, the COVID-19 pandemic led to the implementation of containment measures at the highest level, with a national lockdown enforced on March 9th, 2020. Despite this, by the time the epidemic curve started to flatten, the health system had already exceeded its capacity in different areas of the country, raising concerns that the public health response was indeed delayed. We found that if restrictive measures had been enforced one week earlier, this would have had a significant impact on the evolution of the epidemic in terms of hospital admissions and deaths. By May 3rd, we estimated that there would have been a 60% reduction of COVID-19 cases and 44% of confirmed deaths would have been averted.The COVID-19 pandemic is threatening public health preparedness and medical response capacity globally. Our findings add to a growing body of evidence supporting the need for rapid responses to contain the current COVID-19 pandemic and similar threats that could occur in the future [5,6,12]. Besides Italy, other European countries profoundly impacted by the pandemic such as Spain, France, and the UK, as well as the US, also hesitated to enforce containment measures in a timely manner [8], with a consequent health, economic, and societal impact that still needs to be fully assessed. Lack of collaboration between national health systems, as well as delayed communication by international organizations might be some of the factors explaining the late response to the emergency. Public health intelligence at both the international and national level should identify all barriers and challenges associated with the current pandemic to improve response in the future. This is particularly necessary in this phase of the pandemic, as a possible second wave of infections is expected in the next months. As most European countries are gradually lifting restrictions, there is a need to enhance the existing surveillance systems and develop strategies for timely reactions to a new increase in the number of infections.To our knowledge, this is the first study assessing the impact of the delay in the implementation of containment measures on the spread of COVID-19 epidemic, and the associated burden on the health system. However, several caveats merit discussion. First, analyses were conducted using publicly available data on confirmed cases, which did not account for the proportion of undetected cases, estimated to be high in Italy, especially in the regions more affected by the epidemic [17]. This means that, on the absolute scale, our estimates should be regarded as conservative. On the other hand, assuming that the timing of the lockdown is not associated with the detection rate, which seems plausible, the relative estimates provided are expected to be unbiased. Second, we did not take into account how the different hospital demand under the two scenarios affected the treatment of critical patients. In the actual scenario, hospitals in the worst-hit areas often exceeded their capacity and experienced ventilator shortages [8,16]. This affected their capacity to deliver effective care to all critical patients. On the other hand, under the counterfactual scenario the maximum hospital demand would have been about 50% lower. For this reason, we probably underestimated the positive effects of an early lockdown in terms of reduced ICU admissions and deaths.The COVID-19 pandemic has been requiring unanticipated and extraordinary containment measures, which has raised concerns about public health preparedness of health systems globally. The late implementation of the lockdown in Italy was responsible for a substantial proportion of hospital admissions and deaths associated with the COVID-19 pandemic. Understanding factors contributing to such a delayed response is fundamental to strengthen public health preparedness and timing in response capacity.Conceptualization, R.P., J.B., F.B.-A.; methodology, R.P., J.B., F.B.-A.; writing—review and editing, R.P., J.B., L.R., F.B.-A. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The authors declare no conflict of interest.Predicted number of new cases of COVID-19 under different scenarios. Solid line represents the actual scenario (lockdown implemented on March 9th), and dashed line the counterfactual scenario (lockdown implemented on March 2nd).Total number of cases of COVID-19, non-ICU admissions, ICU admissions, and deaths under different scenarios. Solid line represents the actual scenario (lockdown implemented on March 9th) and dashed line the counterfactual scenario (lockdown implemented on March 2nd). ICU = intensive care unit.Interrupted time series analysis. Estimated regression coefficients.Changes in the number of cases of COVID-19, non-ICU admissions, ICU admissions, and deaths under the counterfactual scenario (lockdown implemented on March 2nd), compared to the actual scenario (lockdown implemented on March 9th). ICU = intensive care unit.

Med-MDPI/ijerph_5/ijerph-17-16-05645.txt

@@ -0,0 +1 @@
+The Australian Asthma Handbook does not recommend use of fixed dose combination (FDC) controller medicines for asthma in children aged ≤5 years. FDCs are only recommended in children and adolescents (aged 6–18 years) not responding to initial inhaled corticosteroid (ICS) therapy. Using Pharmaceutical Benefits Scheme dispensing claims from 2013–2018, we examined the annual incident FDC dispensing and the incident FDC dispensing without prior ICS up to 365 days. We also determined cost of FDCs to government and patients. During 2013–2018, there were 35,635 FDC initiations and 31,368 (88%) did not have a preceding ICS dispensing. The annual incidence of FDC dispensing declined from 14.7 to 7.2/1000 children. Incidence of FDC dispensing/1000 children without a preceding ICS declined from 2.1 to 0.5 in children aged 1–2 years, 7.2 to 1.7 in 3–5 years, 14.8 to 5.1 in 6–11 years, and 18.6 to 11.9 in ≥12years. The cost of FDCs was 7.8 million Australian dollars (AUD); of which 4.4 million AUD was to government and 3.3 million AUD was to patient. Despite inappropriate dispensing of FDCs in children aged ≤5 years, incidence of FDC dispensing and more importantly incidence without a preceding ICS is declining in Australia.Asthma is the most common chronic childhood disease. In Australia, the prevalence of childhood asthma is higher than many other high-income countries [1,2,3]. It is estimated that 20.8% of Australian children aged 0–15 years have ever been diagnosed with asthma, while 11.3% of children have a current diagnosis [4]. The annual national hospitalisation rate for this disease is 495/100,000 children aged 0–14 years [5], costing the Australian health care system ~$200 million [6]. This high burden of asthma is in part due to variation in the clinical management of asthma resulting in low value care [7]. The appropriate management of asthma includes correct diagnosis, asthma self-management education, removal of modifiable triggers, and appropriate medication.Several national and international guidelines for the management of paediatric asthma have been created in an attempt to reduce variability, standardise clinical care across different health care providers, and to improve health outcomes for patients. Physicians across Australia are encouraged to use the freely available Australian Asthma Handbook developed by National Asthma Council, Australia [8]. The Australian Asthma Handbook (AAH version 1 and 2) does not recommend the use of inhaled fixed dose combination (FDC) controller medicines, which include a combination of inhaled corticosteroids and a long acting β2-agonists (LABAs), in children aged ≤5 years [8]. Additionally, AAH recommends use of FDCs in children ≥6 years only as a step-up controller therapy if the initial use of daily inhaled corticosteroid (ICS, anti-inflammatory) fails to control symptoms. Prior to 2019, the international Global Initiative for Asthma (GINA) guidelines recommended increasing daily dose of ICS as a step-up controller therapy in children aged 6–11 years and use of FDCs as a step-up controller therapy after an initial trial with ICS only in adolescents (≥12 years) [9].Data from the USA [10,11] and UK [12] suggest that, despite these established guidelines, the inappropriate use of FDC is actually increasing in children. A similar trend was also observed in the Australian Capital Territory [13] and in the 2014 Pharmaceutical Benefits Scheme (PBS) post-market review of medicines used to treat asthma in children [14]. However, there have been no national studies examining the dispensing pattern of FDCs since the 2014 post-market review and our understanding of how these medicines are dispensed in contemporary practice remains limited. Therefore, the objectives of our study were to assess the patterns of asthma FDC controller medicines dispensed to Australian children using a national, 10% sample of PBS dispensing data. As FDCs are the most expensive asthma controller medicines and are not recommended for children aged ≤5 years, we also aimed to calculate the cost of these medicines to the health system. We further investigated the sequence of dispensing of FDCs with the aim of determining whether or not their use adhered to the AAH step-up recommendations. These data are helpful to quantify the extent of appropriateness in asthma controller dispensing in children, with the goal of improving asthma management for children and reducing burden (including cost) on the health care services.Australia has universal healthcare arrangements for all Australian citizens and eligible residents. The PBS is a program of the Australian Government that provides subsidised prescription drugs to all residents of Australia, as well as certain foreign visitors covered by a Reciprocal Health Care Agreement [15]. We conducted a population-based, retrospective cohort study using the 10% PBS sample dataset—a standardised dataset provided by Services Australia (servicesaustralia.gov.au). This 10% sample PBS dataset is a longitudinal nationally representative random sample of the PBS-eligible, Australian population. The PBS data has records of 23 million Australian citizens. The patient population for the dataset is selected for the sample based on their unique, randomly assigned Medicare ID. The data collection includes all dispensing records of prescription medicines for the sample [15].Our study population consisted of all children and adolescents aged 1–18 years of age who were dispensed at least one FDC between January 2013 and December 2018.The PBS data set does not include details about how the diagnosis of a specific condition was made. However, FDCs are only prescribed to children with asthma. The names of FDCs that are available in Australia and were included in the analysis are listed in Table 1.We calculated annual age-stratified (1–2 years, 3–5 years, 6–12 years, and >12 years age groups) incidence/1000 children per year of FDC dispensing. We estimated incident (new) use by identifying children with a dispensing record for an FDC within a given calendar year and without any dispensing of an FDC in the preceding 12 months. We further estimated incident use of FDC without any dispensing of ICS in the preceding 12 months. We used the Australian Bureau Statistics (ABS) midyear population estimates [16] for each age group, divided by 10 to correspond to our 10% sample, as the denominator for all incidence estimates.The total cost of FDC, including cost to government and patients based on the dispensed price and patient co-payment, over the entire study period as well as for each calendar year was estimated.The New South Wales Population and Health Services Research Ethics Committee granted ethics approval for this study (approval number 2013/11/494).The study involved analyses of routinely collected data and did not involve any direct patient participation or recruitment.During 2013–2018, 31,149 children and adolescents aged 1–18 years were dispensed at least one FDC. There were 35,635 FDC initiations and 31,368 (88%) did not have a preceding ICS dispensing. The median annual number of FDC dispensing/patient (interquartile range (IQR)) was 1 [1,2,3]. For children with two or more FDC dispensing in a year, the median time between dispensing was 70 days (IQR 37–151 days). The most commonly dispensed FDC was fluticasone and salmeterol preparation (Table 1).During 2013–2018, the overall incidence of FDC dispensing in children and adolescents declined from 14.7–7.2/1000 children (Figure 1). The incidence of FDC dispensing /1000 in children aged 1–2 years ranged between 2.6–0.6; 8.8–2.5 in children aged 3–5 years; 16.8–6.6 in children aged 6–12 years; and 19.5–13.1 in adolescents aged >12 years (Figure 1).Incidence of FDC dispensing/1000 children without a preceding ICS dispensing was between 2.1–0.5 in children aged 1–2 years; 7.2–1.7 in children aged 3–5 years; 14.8–5.1 in children aged 6–12 years and 18.6–11.9 in adolescents aged >12 years (Figure 2).The overall cost of FDC for 2013–2018 in our cohort was AUD 7.8 million; of which AUD 4.5 million was to the government and AUD 3.3 million was to the patient (Table 2).This nationally representative, population-based study suggests that while there is inappropriate dispensing of FDC in pre-school children there is a steady declining trend in the annual dispensing of FDCs across all age groups. The observed declining trend in FDC initiations for children contrasted findings from a previous study conducted in the Australian Capital Territory over a different timeframe. That study found a 12% increase in use of FDC between 2002 and 2005 [13]. Despite the observed declining trend across more recent years, if we extrapolate our estimates to the wider Australian population, a large number of children (>50,000) were initiated on FDC therapy without a prior trial of ICS. Although there is some evidence that maintenance and reliever therapy in children with budesonide and formoterol may be beneficial [17], at the time of the study there was insufficient evidence to recommend FDC before a trial of ICS in children ≥6 years [18]. Such practice was also not supported by national and international clinical practice guidelines [8]. However, in 2019 the GINA guidelines updated their recommendations and suggested use of daily low dose of ICS or FDC (budesonide-formoterol) as needed as the first line of controller therapy in adolescents [19]. It is expected that the national guidelines will also be updated to reflect this change and thus our study will provide baseline data in terms of evaluating the change in the dispensing pattern of FDCs following this change.In our cohort, 3500 children aged ≤5 years across Australia were inappropriately [8] initiated on FDC annually which amounted to a cost of ~AUD 500,000 to the government and patients. Such inappropriate use represents wastage of health funds. Additionally, the high cost of these medicines is a significant barrier to compliance with asthma medications [20]. The most commonly dispensed FDC was the combination of fluticasone and salmeterol. This is likely because only the combination of fluticasone and salmeterol is listed on the PBS for use and reimbursed in children aged 4 years and over.It is pleasing to note the trend in reduced dispensing of FDC during the period of the study. Whilst we could not look into the reasons for this, this time period coincides with a significant increase in education to the prescribing community about the appropriate use of FDCs by the Australian asthma peak bodies and the Australian Paediatric Respiratory Medical Group [21] following concerns of tachyphylaxis caused by long acting beta agonists [22].In March 2018, the Pharmaceutical Benefits Advisory Committee (PBAC), an independent expert body of doctors, health professionals, health economists, and consumer representatives appointed by the Australian Government to recommend new medicines for listing on the PBS, considered a three-year evaluation report conducted following the 2014 post market review of asthma medicines in children. Following the meeting PBAC concluded that the proportion of use of FDC outside clinical guidelines remained “unacceptably” high and recommended that listing of all FDC for asthma should be streamlined authority [23]. When prescribing a streamlined authority item, a doctor needs to ensure that the prescribing of the medication is in line with the PBS restrictions criteria for the medication and is required to add the respective streamlined authority code on the prescription [24]. This recommendation was made to promote prescribing ICS as the first line of controller therapy [23]. The data from our study will help monitor the effectiveness of this policy change over time.There are several limitations of our study. The 10% PBS sample includes the year of birth for all patients based on dates of birth that have been perturbed by up to six months to protect individual privacy. As such, some children in our cohort would have been less than one year and greater than 18 years of age. Our data contain records of asthma prevention medicine dispensing, but lack information on adherence to medicine. While non-compliance is associated with sub-optimal management of asthma symptoms [2] we cannot assess this aspect of asthma management. Our data also lack information regarding the type of prescriber, however studies suggest that >90% of asthma preventers are initiated by primary care providers [14]. Finally, our data did not include information on treatment indication and we were unable to investigate why prescribing was not in keeping with the national guidelines.In conclusion, we have demonstrated that both FDC dispensing and initiation are decreasing in Australian children, which is a promising trend. However, children aged <12 years are often prescribed FDCs without an initial therapy with ICS which is inconsistent with National and International Guidelines. Clinical practice guidelines standardize clinical care, reduce wastage of health resources, and improve the value of healthcare. There is a need to understand factors associated with guidelines adherence in order to develop appropriate interventions to improve health care professionals’ awareness of guidelines and appropriate prescribing practices.N.H., A.J., S.P. and B.D. conceived and designed the study; N.H. was responsible for drafting the manuscript; B.D. performed the statistical analyses for the study; A.J., S.P. and B.D. provided technical feedback with drafting of the manuscript. All authors have read and agreed to the published version of the manuscript.The study was funded by Rotary Club of Sydney Cove.The authors would like to thank Sydney Children’s Hospital Foundation and Rotary Club of Sydney Cove for their continued support in our research endeavours. The authors thank the Services Australia for providing the data. This work was supported by The Rotary Club of Sydney Cove. The funding organization had no role in the study design, analyses, or drafting of the manuscript. NH is funded through NHMRC research fellowship (APP1158646), B.D. is funded by a NHMRC postgraduate award (ID: 1094325) and Centre for Research Excellence in Medicines and Ageing (CREMA) PhD top-up award.The authors declare no conflict of interest.Access to the dataset analysed during the current study is not permitted without the express permission of the approving human research ethics committees and data custodians. There is no additional data available.Year and age specific incident dispensing of any fixed dose combination medicines in 10% PBS sample of Australian children aged 1–18 years, 2013–2018.Year and age specific incident dispensing of any fixed dose combination medicine without a preceding dispensing of inhaled corticosteroid in 10% PBS sample of Australian children aged 1–18 years, 2013–2018.Annual fixed dose combination product specific incidence/1000 children per year dispensed in 10% PBS sample of Australian children aged 1–18 years, 2013–2018.Costs of fixed dose combination medicines by year in 10% PBS sample; 2013–2018, Australia.

Med-MDPI/ijerph_5/ijerph-17-16-05646.txt

@@ -0,0 +1 @@
+Adolescence and youth are stages of exploration and experimentation, when the consumption of psychoactive substances for recreational or experimental purposes often begins. The general objective of this study was to explore youth consumption habits in nightlife settings and associated factors in Andalusia (Spain). To this end, we took into account young people’s perceptions about patterns of drug polyconsumption in nightlife settings and the perceptions and actions of health and teaching professionals towards this issue. We carried out a qualitative methodology with 24 in-depth interviews and 3 discussion groups with Andalusian girls and boys aged between 16 and 22 (n = 45) and 13 in-depth interviews with social agents (health and teaching professionals). We performed narrative discourse analysis and triangulation of identified categories and measured the units of analysis. The results show information relating to gender, age of initiation, most commonly consumed substances, motivation and effects, peer group pressure and how they obtained the substances, and the perceptions held and main activities carried out in the educational institutions and health centers.Adolescence and youth are exploratory stages in life characterized by a wide range of biological, psychological and social transformations, many of which generate conflicts, crises and contradictions [1]. The United Nations defines youth as the age cohort between 15 and 24 years, and within this, the first stage from 15 to 19 years, is called adolescence [2]. Although it is difficult to delimit where these terms begin and end, in view of the multiple socio-cultural diversities which exist [3], it is essentially a flexible concept whose meaning is constructed and reformulated according to the cultural, political and economic situation in each place and historical period [4].It is important to point out that it is an exploratory stage, characterized by an increase in risk behavior, sensation-seeking and risk-taking [5,6] and experimental consumption of legal and illegal psychoactive substances, which link to certain habits and lifestyles, and have a major impact on an individual’s future health status [7,8].We identified three levels of organization (individual, social and cultural) in relation to what motivates young people to follow risky consumption behavior. At the individual level, it is relevant to note that “sensation-seeking” is a personality trait defined by the need for varied, novel experiences and sensations [9], and the eagerness to take physical and social risks, and to accept the risk associated with any likely negative outcomes [10]. At a social level, there are the emotional stimuli of social interaction, the feeling of living “for the moment”, in which sex, alcohol and drugs are all “part of the fun” [11]. In the extensive, ephemeral sociability of adolescents’ shared leisure time, they play down risks, and everything revolves around the group: they go out, socialize, take drugs and experience the rites of passage to adulthood in a group [12]. By ‘youth nightlife’, we mean when young people and adolescents go for a night out, to a nightclub, disco or a bar, usually in the company of friends, or meet up to drink or take drugs in other areas such as parks, fields, abandoned lots or even a friend’s house [13]. This need to belong and feel part of a group is at its most intense in the stages of adolescence and youth, where individual and collective identity is formed [14,15,16]. On a cultural level, it mainly takes place on weekends, which are geared towards play and hedonistic pleasures [17]. Studies of youth culture have shown how leisure time has become the new measure of happiness in Western culture, and youths in particular understand it as an essential part of a desirable lifestyle [18]. For adolescents, nightlife equates to a break from the everyday routine, and a sense of freedom [12].These individual, social and cultural elements explain how, in the context of adolescent leisure time, there is no single cause, but rather a combination of factors which leads to risky behavior, a mix of euphoria and collective excitement [11,15,17], where alcohol and drugs are a common denominator, which facilitates disinhibition and relationships between individuals, and in turn promotes risky situations [19].Alcohol consumption is among the leading public health problems worldwide. Drinking alcohol is associated with a risk of developing health problems such as mental and behavioral disorders, including alcohol dependence, major non-communicable diseases, some cancers and cardiovascular diseases, among others [20]. Initiation to drinking alcohol, in societies where it is part of their culture, is universal. Drinking is used to mark social rites of passage of all kinds, including access to adult life, but it is also traditionally associated with learning self-control. However, in Western European countries, the mode of consumption is changing, with the globalization of products and patterns: young people from Anglo-Saxon countries have taken on the daily consumption of wine, so typical of Mediterranean countries, while young people from Southern Europe have started to consume more drinks with a high alcohol content and to binge drink [21,22]. Alcohol has become the universal fuel of the nightlife economy and its consumption is considered pleasant, fun and sociable [21,23,24,25].When we consider the consumption of psychoactive substances, there are different degrees of social acceptability depending the type and intention—in other words, how dangerous the drug is considered, and where it is consumed. Heroin is considered the most marginal and stigmatizing, cocaine is seen as attractive, pills dangerous, cannabis natural and innocuous, and alcohol is the most socially accepted and omnipresent [6,26]. Consumption that takes place in leisure time and for recreational purposes, seeking fun and pleasure, is considered more acceptable [27,28].In addition, drug polyconsumption is occurring, which is defined as the use of two or more psychoactive substances at the same time [29]. Some authors have established three modes of drug polyconsumption: pattern A (alcohol and tobacco), pattern B (cannabis with alcohol and/or tobacco) and pattern C (cannabis with alcohol and tobacco and at least one other illegal drug) [30]. These last two patterns increase the risks, enhance the effects of some substances on others, interfere with diagnosis and make treatment more difficult [29]. A previous intervention in 3882 Spanish adolescents pointed out that, in the short term, the consumption of more than one substance led to a higher probability of consuming other illegal substances, risky or abusive consumption of alcohol (binge drinking), traffic accidents, injuries, risky sexual relations with a greater probability of sexually transmitted diseases, fights, problematic use of the Internet or cyberbullying [31], criminal behavior and other acts of violence [32,33,34]. In terms of medium- and long-term consequences, it is associated with physical problems, low cognitive performance, depression and suicide. To make matters worse, the more substances consumed, the worse the clinical outlook and the greater the dependence [30,35].At a global level, there are major differences in consumption patterns. The latest report on drugs stated that consumption among young people differs from one country to another according to social and economic circumstances, with two extreme categories: first, in high-income countries, where consumption occurs at rave parties, university cafeterias, concerts, and in recreational contexts to enhance sensations and experiences, using ecstasy, methamphetamines, cocaine, ketamine, LSD and gamma hydroxybutyrate (GHB), among others; secondly, in countries and areas with a low socioeconomic level, it is linked to young people living in marginal conditions, who consume drugs to escape from the harshness of their circumstances, and the substances most consumed are solvents, gasoline, paint, correction fluid and glue [36].In Europe, addictive drugs are those that are capable of inducing habit, tolerance and physical dependence, where the doses need to be increased to achieve the same effect and whose deprivation produces a specific, manifest, measurable and observable syndrome, with harmful consequences for the individual and society [35]. Although there is no consensus over the legality or illegality of each substance in the legal systems of the different countries, studies are unanimous in including tobacco and alcohol, because of the habit and addiction they produce, even though they are mostly considered legal substances [37]. The data indicate that the most prevalent illegal substance is cannabis, followed by cocaine, 3,4-metilendioxi-metanfetamina (MDMA) and amphetamine [35]. In terms of gender, men have used these drugs over twice as much in the past year as women, with more intense and regular patterns. In terms of age, the prevalence is highest among young adults aged 15–34 [35], the onset of substance abuse is most pronounced in the period of early (12–14) and late (15–17) adolescence, and substance abuse peaks among persons aged 18–25 [36].In Spain, the drugs most consumed by young people between 14 and 18 years old are alcohol and tobacco, followed by cannabis and sedative-hypnotics, with or without a doctor’s prescription. The ESTUDES Survey [38] showed that in late adolescence, there is a normalization and generalization of the abusive consumption of alcohol in both sexes, associated with weekend leisure (Spanish Observatory on Drugs and Addictions [38]). The average age of initiation into consumption is 13–16 years old, with no significant differences by gender. Consumption is “generalized” by type of substance, with women opting more for legal substances (alcohol, tobacco and sedative-hypnotics), and men for illegal ones [38], with an increasing consumption of a range of new psychoactive substances, such as spice, ketamine, ayahuasca, mephedrone, psychedelic sage (Salvia divinorum), hallucinogens, volatile inhalants, heroin, magic mushrooms, ecstasy, amphetamines and methamphetamines [29].Andalusia follows the same pattern as Spain in terms of initiation, age, gender and type of substance consumed. However, there is a higher percentage of male polyconsumers, with the highest rates among users of heroin (average of eight substances), gamma hydroxybutyrate (GHB) (7.5), Ketamine (6.5) and magic mushrooms (6.1) [39].As discussed above, substance use among adolescents leads to negative outcomes and risk behavior. Here, several studies pointed to the need to identify and respond early to reduce the associated risks in adolescence and adulthood [40,41,42]. For instance, Levy et al. 2016 stated that there needs to be an integrated public health approach to Screening, Brief Intervention and Referral to Treatment (SBIRT), involving not only health professionals but also educational institutions [43,44,45].To date, various studies of substance consumption in adolescents and/or young people associated with risk factors have stressed the importance of detection and intervention programs. However, there is no research that jointly addresses the polyconsumption habits of young people and adolescents in nightlife settings and the perception of social agents related to education and health, which warrants the need to carry out this research using a qualitative methodology.This study therefore aimed to explore youth drug consumption habits in nightlife settings and associated factors in Andalusia (Spain), taking into account the perception of young people regarding polyconsumption patterns in nightlife settings and the perceptions and action of health and education professionals to deal with this problem.We adopted a qualitative exploratory and descriptive design with a phenomenological approach [46]. The study also followed an ethnographic approach, since this allowed us to explore a particular topic in a specific context and in a specific subgroup [47]. This approach is characterized by (a) a conceptual orientation provided by a team of researchers; (b) a focus on a discrete community or group; (c) a focus on a problem within a specific context; (d) a limited number of participants; (e) the use of participants who may have specific knowledge; and (f) limited observation of participants [48].The total number of participants was 58. We set up 24 interviews and 3 discussion groups (DG) (n = 21) with young people between the ages of 16 and 22, from the Andalusian provinces of Granada and Seville (Andalusia, Spain). In addition, we conducted 13 interviews with educational staff, promotion and prevention workers and healthcare employees.For the selection of the sample, we used intentional sampling, through a process of searching for independent networks in different contexts to maximize the capture of different experiences, as well as to achieve greater control over bias in the selection of participants.The inclusion criteria for the sample of young people were being aged between 16 and 22, with the following profile: high school or university students, consumers of alcohol and attending nightlife venues frequented by young people. As for the social agents, they had to have permanent and direct contact with young consumers in the fields of health and education.Researchers from the anonymous project run in schools in Granada and Seville in 2019 carried out the semi-structured interviews with the young people. To recruit participants for the discussion group (DG) and agent key group, the 3 researchers from the team contacted professionals from the different areas involved, and, after informing them about the study, interviewed them.For the interviews and DG with the young people, we used an interview script, structured in two thematic blocks: Recreational and leisure environment and Recreational drug use patterns (Table 1).For the interviews with the key actors, we followed another interview script which included specific questions from the areas to be explored: education, health care and health promotion (Table 2).We collected sociodemographic variables, such as environment, gender and age, to ensure the heterogeneity of profiles.In addition, we held discussion groups to explore the issues raised in the interviews. This allowed us to find out the motivating factors that were expressed only in the group dialogue, and to establish contextual interpretations for the responses registered [49].A final report was prepared with the statements of interview and DG informants, which is coded as follows: R (respondent); HC (health care employees); EE (educational employees); HPE (promotion health employers)-number. The analytical process was performed using the QSR NVivo 12 program.The interviews lasted on average 60 min and the DG, 45 min, and we recorded and transcribed both. We performed a narrative analysis of discourse by coding and categorizing the themes emerging from the interviews and DG, read the scripts, and made a first identification of codes and categories. After this first phase, another member of the research team carried out a triangulation of the identified categories.After repeated readings of the transcripts, we identified the units of analysis and identified and grouped the categories, maintaining criteria of reflexivity and flexibility throughout the process. This whole analytical process was performed using the QSR NVivo 12 program. We completed the data collection process following the principle of theoretical saturation. This research followed the criteria of the Consolidated Criteria for Reporting Qualitative Studies (COREQ). The methods used to ensure validity were the triangulation of data, including participants with different socio-demographic characteristics, and the triangulation of data analysis through different researchers (Table S1).The study was approved by the Research Ethics Committee of the University of Granada (CEI-UGR/883). All persons participating in the project voluntarily agreed to be interviewed, received information about the study with a letter from the research team guaranteeing confidentiality, and signed the informed consent form. The participation of minors also required the signature of their legal representatives in each case.The sample consisted of 45 adolescents (41.6% males and 58.4% females). Twenty-three were resident in Granada and 22 in Seville. The average age was 17.7 years old. All had consumed alcohol and/or other drugs/substances in nightlife settings. There were five educational employees (40% men and 60% women) with an average age of 36.2 years, who were working at that time in secondary schools, vocational training (n = 4) or universities (n = 1). There were also four promotion and prevention employers, who carried out specific programs with adolescents and young people in educational centers, specifically the Forma Joven Program, (25% men and 75% women), with an average age of 55 years, who taught the Forma Joven Program in middle schools, from the primary care network of the Autonomous Government of Andalusia, or from the Health Promotion Unit of Seville City Council. In the case of the healthcare workers, the sample was composed of two men and two women (50% men and 50% women), with an average age of 39.7 years, who worked at the Critical Care and Emergency Unit (DCCU) and in primary care for the Public Health Service of Andalusia (Spain).Among the motivating factors for alcohol, tobacco and cannabis use, young people stress experimentation as the main reason to start consuming, as well as for pleasure, meeting new people, relaxing and escaping from problems (37 participants). We recorded the following statements:R8: “I’d say that drugs, apart from freeing you from your inhibitions and problems, make you feel superior, they make you feel like you’re God.” (R21, R23)R14: “There, all my mates who usually meet up smoke—we all smoke. We started smoking together and it’s not like I’m going to get addicted, no. One of my mates comes along one day, we get together, he always rolls a joint and that’s it.” (R22)Educational professionals report that young people see consumption as a way of escaping or relaxing, or simply of integrating into a group.EE2: “At times in class when the subject comes up, they say they smoke joints because it relaxes them.”EE5: “They do it to be accepted by the group, we see that at school too.”Among the responses reported by young people, there is evidence of a variety of substances and medicines consumed, from the commonest triad of alcohol, tobacco and cannabis, to the use of medicines or other drugs such as MDMA. Most of the participants (31 interviewees) admitted to having consumed cannabis, resins and oils derived from marijuana, and designer drugs, obtained on the illegal market. The following comments were recorded:R29: “The substance young people consume most, after alcohol and tobacco, is cannabis. To a lesser extent, pills and cocaine.”R3: “So much cocaine, joints, MDMA, pills… pot, GHB (…). The liquid ecstasy thing I told you about. (…) I mixed it with more drugs. I mixed it that night with MDMA, with Speed, with cocaine… when I went out I didn’t just use one drug, maybe I mixed several.” (R3, R4a, R4b, R5a, R12, R13, R19, R24).R41: “It’s a pill that makes your mouth really dry and makes you quite dizzy, because as people don’t buy alcoholic drinks, but they go there to take the drugs, the clubs are now forced to sell bottles of water for twelve euros.”On the other hand, educational professionals point out that, in school, the substances most consumed are tobacco, hashish and cannabis at break and when classes change over. The same substances are also mentioned by health promotion employers, with the addition of pills.EE1: “The most common substance is cannabis, consumed at breaks or outside the classroom (…) they talk to each other before starting the class, and on some occasions I’ve heard how they pool their money to buy alcohol, and even a kind of powder.”EE2: “Tobacco and hashish in the breaks, and when they come in and out of the school.”HPE1: “Alcohol, cannabis, marijuana and pills.”These can also be drugs bought in pharmacies, according to several of our informants:R6: “Do you know what it is? A cough syrup that has codeine in it, which is something, let’s see, how can I tell you… it has a substance that’s also found in heroin. And that’s in the cough syrup to sooth your cough. You mix that with Sprite and you drink it with ice, and it gives you a high like a reefer, and it tastes good too.” (R5b, R8.)Likewise, the most commonly used drugs, according to the four health care staff, were tobacco, alcohol, cannabis, cocaine and synthetic drugs:HC1: “I couldn’t tell what the typical profile is: we deal with young polyconsumers with very diverse characteristics, although they tend to be young people with a low level of purchasing power and culture and who tend to come from unstructured families.”HC2: “I don’t think there’s a typical profile; many young people today are polyconsumers without seeming that way.”Respondents differentiate between “safe” and “addictive” consumption.R16: “I can go out without having to smoke—like playing, going to training, or going to a party and not smoking. It’s not something I depend on. I like doing it, but I’m not addicted to it. In fact, people consider marijuana a drug, I don’t consider it as a drug because tobacco is a drug too.”R19: “Who doesn’t like temptation? But I control myself, I know that it’s bad for me and I don’t do it.”However, others mentioned that experimental consumption can turn into a habit:R32: “For me, marijuana is something that I think will be around for the rest of my life, I don’t know, I can’t deny it.”The young people interviewed were not unanimous in their opinion about gender-related differences in consumption.R2: “Boys smoke more cigarettes and girls more joints.” (R1, R4)R35: “It doesn’t matter if you’re a woman or a man. It could be that it affects women more… you know what I mean, it’s more or less similar, but for each person, each substance is different.”Here, educational professionals did not perceive a difference in relation to gender. Their opinion was shaped by anecdotes from girls about drug consumption after the weekend.EE2: “The consumption we can see here in the school is not a reflection of what they may have taken at the weekend or at night.”EE4: “I’m not sure… they don’t usually tell us what they do when they speak in class, but then, first thing Monday morning is crazy-as they see you as a young person, they tell you about it, and I can sometimes hardly believe what I hear.”Among the young people interviewed, sales and distribution happen through a network of “contacts” known to them. Sometimes the young people grow drugs for their own consumption and sell the surplus to get some extra money.We observed in this study that it is relatively easy to obtain substances of all kinds. For pharmacological substances, they go to pharmacies, and we have found one informant who can obtain it in his family environment, since one of his parents is a consumer.R7b: “Not other things, but unfortunately drugs are everywhere and you can get whatever you want.”R22: “If I have weed and I have it to spare, I sell it and make a bit of cash.”R2: “There are certain people we know who are the typical pushers, the ones who always sell everything, and yes, they drop by.” (R18, R7a, R24)In the educational field, it is difficult to obtain information about this, but some of our informants report that they have heard students talking in class about how they are going to get drugs for a party.EE4: “As the class starts or when you leave, a typical group forms, and they talk about people they buy from”… I heard once, “Go on, send him a WhatsApp and get him to save us some.”EE5: “They have several, and I once heard “Well, if X doesn’t have any, we’ll get the car and go to Y.” Another one answered, “Yes but that one’s more expensive and I’m broke.””Health promotion professionals agree with the young participants about the usual way of obtaining substances, and say they get them through friends, family and drug dealers.EE3: “Colleagues and drug pushers.”EE5: “Friends and family.”As for the places where they consume drugs, it is usually in parks, outdoors or in bars, or sometimes at a friend’s house.R3: “We’ve been smoking in a park or quietly at someone’s house.”R36: “For example, where I work…the people…in the bathroom we’ve seen little bags, so I asked my workmates what they were and they said “Well, that’s the people, they sometimes come here to the bars and take drugs..” Because once I saw an older girl having a beer who was falling asleep like that and I said “What’s up with this one?” Then they said “She’s been to the bathroom…” and I was like, “OK, right!”We also wanted to know the opinion of the health care employees about the places where they administer health care, whether in the public sector, at private homes or in emergency wards:HC1: “We see these young people mainly at homes, or in the street.”HC2: “In my case, I see them at the health center.”The lack of any health care attention at these nightlife settings was also noted, as opposed to other options in other Spanish cities:     HC1: “We don’t go there, but I don’t think it’s necessary.”HC2: “There isn’t, but it would be a good idea.”HC3: “No, we don’t go there, we’re at the emergency wards and that’s where they usually are. In other cities like Ibiza, there are health staff at the clubs, but not here yet.”Educational staff point out that Thursday-night parties have become increasingly popular and the students usually buy the tickets before, which comes with a free drink. They also mention that consumption increases at the weekend.EE1: “On Friday when I leave work in the afternoon-evening, there’s an open field nearby, and they’re in the car with music and drinking, the other days of the week I don’t see them, some of them even go to our school.”EE3: “As the weekend approaches, consumption seems to increase, and their attitude is usually a bit more laid back/slower.”EE5: “Here at college we hear about university parties on Thursdays and Fridays. There are students who sell tickets with a free drink.”EE4: “Especially on Thursdays, there are parties that they organize themselves… They talk about the ‘botellones’ (large groups of young people drinking in public places) and what each one is going to buy when the weekend comes.” (EE10, EE11).In our interviews with young people, there was an association between polyconsumption and violence, mainly attributed to alcohol, as well as undesired effects and psychic symptoms, sometimes requiring treatment from health care givers.R11: “The most I’ve ever smoked was thirty joints. But I dropped dead that day.” (R2a)R13: “I tried it in cakes and things like that, but it was too strong, so I don’t want to try it anymore… I remember being in bed, because I was in my house, lying on the bed as if I was asleep but I couldn’t react, although I could hear everything. I don’t know, it was just too much, so I don’t want to do it anymore.”R9: “Mix any drug with alcohol, and all hell breaks loose.” (R2b)R27: “When you take x substances what it does is… it doubles or triples whatever you feel… everything, so, it doubles or triples the violence, doubles or triples your sex-drive, but alcohol already does that, doesn’t it? That’s another drug, the only thing that it’s legal (Laughs). But these substances do it to you too.”Here, key health promotion actors in the field (four participants) consider that when short-term adverse effects arise, they primarily turn to friends rather than to health professionals. The adolescents and youths we interviewed are aware of the risks of substance use and the possible consequences of addiction, but the peer pressure from the group of friends is stronger than their individual decision.HP2: “They turn to friends, older relatives, and finally their parents.”HP3: “They know the risks, but they follow the tribe.”Health care workers point out negative experiences they have experienced during health care in emergency situations.HC1: “Care usually consists of monitoring, knowledge of the clinical situation, stabilization, and assessment whether to transfer them to hospital.”Once the emergency is over, the agents point out that there is no protocol to follow up these young people:HC3: “It’s hardly ever done… it’s always through primary care consultations, and in more serious cases, through Detoxification Centers.”HC4: “No follow-up, an emergency is dealt with, and that’s it.”To avoid these situations, the healthcare workers interviewed point out that healthcare centers take some measures to prevent polyconsumption (for example through information campaigns, interventions with the population, among others).HC2: “[We run] workshops in schools, with the emphasis on primary care consultation.”HC4: “Information on posters.”In the educational field, there is a notable lack of education and training expressed by these professionals to address this problem. Some of them refer to training acquired from previous experiences or through personal interest.EE1: “I have no training” (EE4–EE8; EE10)EE3: “None. The only training I’ve had is my experience in life”They also refer to the lack of prevention programs aimed at students and that the educational institution has few protocols for action.EE1: “There’s no specific plan. The only things in the Regulations are the type of punishments for consuming drugs in the school (the sanctions model).”EE5: “We have no prevention programs: the problem’s just addressed by dealing with cross-curricular subjects such as healthy living habits, physical exercise, anatomy or topics which lend themselves to making comparisons between healthy and harmful substances due to addiction, etc.”Health promotion professionals in the area of health have a higher level of training, and these workers claim to have received specific training courses given in continuous education courses by the District Health Organization.HP1: “I get training from my health district training unit.”HP4: “Courses, scientific publications… I’m qualified in the Methadone Program, ‘Proyecto Hombre.” (‘Proyecto Hombre’ is a project set up in Spain in 1984 which tackles addiction, from prevention to treatment, rehabilitation and reinsertion into society/work. https://proyectohombre.es/quienes-somos/#:~:text=Proyecto%20Hombre,as%C3%AD%20como%20a%20sus%20familias.&text=Tambi%C3%A9n%20trabaja%20en%20la%20prevenci%C3%B3n,de%20un%20mill%C3%B3n%20de%20personas.)For more information on the themes and distribution of verbatim quotations, please consult Table A1.The objective of this study was to explore the consumption habits of young people in nightlife settings and associated factors in Andalusia (Spain), through the perception of young people and professionals working at educational and health centers, as well as the main actions carried out.The main results shown are that the use of substances was not related to gender, and that the main motivation factor for substance use was for pleasure, escaping from problems, relaxation and integration into the peer group. The most commonly used substances were tobacco, alcohol and cannabis, and their use was sometimes linked with unpleasant experiences and negative effects, such as violence or psychological symptoms. The setting in which they consume ranges from parks or bars to their own homes. The substances are sold and distributed via word of mouth and at key places known to the peer group.The professionals generally coincide with the adolescents interviewed in most aspects. It is noteworthy that very few prevention programs are run in the areas of education, and when they are, it is thanks to the health professionals associated with the ‘Forma Joven’ program at the request of the school or institution. We also found little training to address this issue taking place in the educational sphere, which is not the case with health professionals, who receive continuous training in this respect.In relation to the influence of the gender variable, several studies highlighted that there are no significant differences between men and women in the consumption of legal substances such as alcohol and tobacco [31,50,51,52], but the percentage of consumption of cannabis and other drugs (ecstasy, amphetamines or hallucinogens) is higher in men [31,52,53].Previous research linked changes in gender roles and the consumption of certain substances; in the case of alcohol, there may be a direct relationship between these social changes in young people and alcohol consumption [21,53,54,55]. Nevertheless, evidence from previous research reported greater polyconsumption in young men than in women [56]. Fernandez et al. (2018) stated that adolescents construct their masculinity through alcohol abuse, and that violence is a trait associated with “proving masculinity”. The interviewees share this idea, linking excessive alcohol consumption and violence and relate them to masculinity [57]. However, despite teachers referring to a higher consumption among young men (as reported by them), this figure may be biased because young men are more likely to talk about their experiences. Males are more explicit in their responses because they understand that they can freely exercise activities related to nightlife, such as enjoying going out, partying, and alcohol and drug use, without these activities damaging their reputation or undermining their masculinity. The opposite happens with girls, who tend to take the stance and declare in their statements that “they are good girls”, who repress their desire to drink alcohol and get drunk at parties, so as not to damage their “feminine” image, which is why they also follow rules of legal consumption [21].The motivation factors which drive consumption include feeling euphoric, having hallucinations and increasing self-esteem [52]. Low resistance to peer pressure appears to be a risk variable in consumption. Consumption behavior also seems to be influenced by different psychosocial aspects, such as feeling integrated and accepted by the group, or to facilitate socialization [58], a view which educational professionals share. Previous studies have pointed out that among the factors which influence the initiation and stabilization of substance use in youths, peer group pressure is paramount, since relationships with peers are more important in this period and they spend more time with their peer group. Thus, there is a general consensus over the idea that one of the most powerful predictors for substance use is being accompanied by friends who engage in this type of practice [58,59,60]. On the other hand, Romo-Avilés in 2015 identified that the peer group functions as a protective community which looks after its members’ safety when risky consumption takes place [55].Polyconsumption has become another common practice. Among the possible combinations of psychoactive substances which characterize polydrug use, the most common according to several studies are alcohol, tobacco and cannabis [3,27,30,61]. Occasionally, in special situations, this may include other substances such as cocaine, ecstasy or designer drugs [30].There is a clear distinction here between “safe” consumption and “addiction”. This idea coincides with the literature, in which the figure of a weekend consumer of legal and illegal drugs is considered perfectly natural and normalized and is associated with the typical recreational experimentation of young people [27], to such an extent that young teens often consider sporadic consumption harmless, which stresses the relationship between consumption and low risk perception [61].Strategic points of supply and consumption exist in most cities where there are certain neighborhoods, usually found on the city outskirts, linked to marginal culture and with low socioeconomic levels, where substances are trafficked, as Ferrer pointed out in one of her interventions [62]. The players involved in this process are the traffickers, who supply the pushers, who sell to the consumers, as well as other intermediaries who perform support tasks (vigilance, raising the alarm if the police appear, etc.) [63]. The educational professionals confirm the existence of these figures, and that young people usually have their phone numbers so they can confirm the purchase before the event or simply to know where to pick it up.This contrasts with other articles, which show that in the vast majority of cases, the sale, distribution and purchase of illegal substances occurs online, so Internet users are able to evade the authorities [64], and new buying and selling practices also occur using collective appointments made by phone [63].For promotion, prevention and action, the study by Levy et al. 2016 stated that an integrated public health approach of Screening, Brief Intervention and Referral to Treatment (SBIRT) is needed [41], involving not only health professionals but also educational institutions [43,44,45]. We have included the figure of these social agents in our study because of their key role in identifying and responding early to reduce the risks in adolescence and adulthood, as reported in other studies conducted in the area of adolescent drug use [40,41]. As verified in this research, a successful SBIRT promotes behavioral changes by helping these adolescents to resolve the ambivalence of their changes in behavior, through empathetic interview styles, with guided debates about the perceived harm of drug use and the benefits of abandoning the use of substances [65].In addition, our study, together with the one conducted by Kristen et al. 2019, shows that there are gaps in our knowledge in the educational field and our capacity to provide evidence-based prevention and early intervention. Most of the teachers had not received any kind of training in this area, and more training is certainly needed [41].This study has certain limitations. We chose our sample on the basis of convenience, which makes it difficult to extrapolate wider conclusions from the results obtained. Nevertheless, the results can serve as a reliable approximation of the real situation, and the qualitative methodology used allowed for a precise approach to these issues. Another limitation is that we did not measure the previous information that the participants have about the topic, and this would have been interesting to include as a variable of the study in order to obtain more detailed results about the level of previous knowledge and the relationship with the answers obtained.We believe that analyzing the discourse of young people, adolescents and social agents is an appropriate way to continue working on these issues to improve our knowledge and ways of addressing this social problem. This research can be considered as an approach to a contemporary reality which is of considerable relevance to social, political and health issues. We need to create a wide range of action protocols, and adolescents and young people, health professionals and teachers should work together to design evidence-based prevention and intervention programs.In our study related to gender, the inclusion of young women as consumers in the same percentages as their male peers and beginning in adolescence will lead to a serious health problem in this decade. As for the types of drug consumed, we found that alcohol is omnipresent at all ages and types of consumption, and we have observed a “generalized” use of substances and their consumption. Women remain within the parameters of “legality” with poly-consumption of tobacco, alcohol and sedative-hypnotic drugs dispensed in pharmacies, without crossing the lines of legal risk. Meanwhile, in addition to legal drugs, if men risk accessing the illegal market to obtain other types of substances, the most commonly consumed drug is cannabis, but there are others such as cocaine, speed, MDMA, ecstasy or GHB. The ease of socializing, relaxation, getting euphoric and escaping from problems in a fun, party-like environment, generally at weekends, are the main motivating factors. Sometimes, undesirable effects such as drowsiness and distortion of perception appear, mainly related to excess consumption of cannabis, or aggressiveness related to the excessive consumption of alcohol or a mixture of this with other substances.The peer group is one of the most influential factors in initiating and maintaining consumption. The sale or distribution of illegal substances occurs through known contacts at strategic points. On the other hand, drugs such as codeine, generally in the form of syrups mixed with alcohol, are available in pharmacies with or without a prescription.In the area of health and education, health professionals provide more education, training and prevention, detection and intervention programs, whereas educational professionals do much less: teachers should be more involved in delivering such programs, since adolescents and young people spend a large part of their lives in educational institutions.We believe that it is vital for teachers, health care workers and parents to continue working on these aspects through a qualitative approach with young people, adolescents and social agents to improve our knowledge of this social problem. We hope this research will help these professionals to understand more about a contemporary issue which has enormous relevance in social, political and health fields. They should plan new lines of action to address them, and adolescents and young people, health professionals and teachers alike should work collectively to design evidence-based prevention and intervention programs.The following are available online at https://www.mdpi.com/1660-4601/17/16/5646/s1, Table S1: Consolidated criteria for reporting qualitative studies (COREQ): 32-item checklist.Conceptualization, M.Á.G.-C.-M. and L.T.-C.; Methodology, R.d.D.-C.; Software, R.d.D.-C.; Validation, R.d.D-C.; Formal Analysis, M.Á.G.-C.-M., R.d.D.-C and L.T.-C.; Investigation, M.Á.G.-C.-M., R.d.D.-C. and L.T.-C.; Data Curation, M.Á.G.-C.-M., R.d.D.-C. and L.T.-C.; Writing–Original Draft Preparation, M.Á.G.-C.-M., R.d.D.-C. and L.T.-C.; Writing–Review & Editing, M.Á.G.-C.-M., R.d.D.-C. and L.T.-C. All authors have read and agreed to the published version of the manuscript.This research was funded by FEM2016- 77116-C2-1-R, MINECO/ERDF, EU.This study is part of the project “Gender and Interpersonal Violence in Adolescent Recreational and Leisure Settings (VIGEA)”. R + D + I projects of the State Program for Research, Development and Innovation oriented to the Challenges of Society. State Plan for Scientific and Technical Innovation Research. Ministry of Industry, Economy and Competitiveness. State Research Agency. 2016–2019. (Reference FEM2016- 77116-C2-1-R, MINECO/ERDF, EU).The authors declare no conflict of interest.Description of themes and distribution of verbatim quotations.Interviews and discussion group (DG) guide.Preferred days for going out at night.Most common habits and settings chosen for nightlife.Models of organization within the group (relationship between peers).Preferences for venue (public or private), time schedules.Substances consumed when out at night: types, quantity, frequency, consumption over the last month, how it is bought.Reasons for drug polyconsumption.Differences between genders.Expectations over consumption and related experiences.Script for social informant interviews.How does your educational institution (college, university) address the prevention of substance use (alcohol, tobacco and other drugs)?Describe the times when substance use by adolescents increases. What are the most common substances they use?What is the protocol to follow when you detect cases of drug consumption in the college or when students are seen to have consumed drugs?Are there particular days or times in the year when you detect that adolescents consume more? What usually happens those days?What training have you received to address this problem?Have you detected any contacts (e.g., contact person who supplies the drugs) in the institution, or outside it? How are these contacts made?Is the motivation to consume drugs dealt with in the FORMA Joven 1 health promotion program?When are the drugs consumed (daily, during the week, at weekends)?What types of substances do they consume?What are the short-term adverse health effects? Who do they turn to if they suffer adverse effects?In the medium and long term, do they reflect on addiction and adverse health effects?Where do they go to buy the substances, and who do they buy them from?What primary health care measures are taken to prevent drug polyconsumption, for example through awareness-raising campaigns, interventions with the target population, etc.?Are any kind of health workers present at nightlife settings?In cases of emergency: where are young people treated, what care is given to them, and how do they get there?In your experience, what is the typical profile of the young drug polyconsumer? What substances/drugs are most frequently detected?After emergency care, is there any kind of follow-up of these young people?1 FORMA Joven’, is a Program of the Andalusian Regional Government, with strategies to promote healthy life habits among adolescents and young people. https://www.formajoven.org/.

Med-MDPI/ijerph_5/ijerph-17-16-05647.txt

@@ -0,0 +1,15 @@
+In recent years, urbanization has been developing rapidly. However, it is also accompanied by land management problems, such as low land use efficiency. In this research, we manage to explore the temporal and spatial evolution laws as well as characteristics of the coupling and coordinated development between urbanization and land use benefits. Through this, it is possible for us to provide policy recommendations for the sustainable development of the urbanization in Fujian Province. In this study, we take prefecture-level municipal districts and county-level cities in Fujian as the research subject. We construct an index system, based on data in 2002, 2005, 2010, 2015, and 2017, to evaluate the urban land use benefits and urbanization. Besides, we leverage the Gini coefficient weighting method to give weight to each index and calculate the value of its benefits. Moreover, it is the relative development degree and the coupling coordination degree model that we comprehensively leverage to study the spatiotemporal evolution law of the coupling coordination degree (CCD). The results show that: (1) Urban land use benefits and urbanization level are positively correlated with the regional administrative level and economic development status; (2) The CCD of urban land use benefits and urbanization level in various regions of Fujian is still low. However, the overall development direction is good; (3) From the perspective of spatial distribution, the CCD owns a “center-periphery” pattern that is based on the law of diminishing CCD power from three central cities of Fuzhou, Xiamen, and Sanming. Consequently, it requires governments to take action. Firstly, they should promote the intensive land use in the urbanization process. Meanwhile, they should also pay attention to ecological environment protection. Besides, it is recommendable to give full play to the radiating and leading effect of central cities on surrounding ones. Finally, they are required to provide appropriate policies and resource support to peripheral cities.As a basic carrier of human activities and urban operations, land resources are a basic guarantee of urban socioeconomic stability. Their effects on urbanization have been gradually strengthening. The land resources are not only closely related to the development of urbanization itself, but also to the food security and social stability [1,2,3]. In recent years, the rapid progress of China’s urbanization has brought huge economic benefits. Nevertheless, it has developed at the expense of the ecological environment. More specifically, it has brought various environmental problems, such as ecological deterioration, resource scarcity, land degradation, overall climate change, and a comprehensive decline in urban land use. Even in some rural areas, there has been a phenomenon of abandoned farmland [4,5,6,7,8,9]. As a result, these problems have heavily affected the sustainable development of the regional economy and society as well as human well-being. What is worse, this accelerated urbanization process demands a lot of land resources [10]. Urban expansion in the peripheral areas of major city centers generally takes place at the price of prime farmlands [11]. China’s land for construction is still increasing with rapid urbanization [12]. Consequently, how to strike a balance among land uses, expand the benefits of land use while achieving healthy development is a hard problem lying in the process of China’s urbanization.The land use benefits refer to economic, social, ecological, and environmental benefits. These kinds of benefits are directly generated by the utilization of the unit land area in a certain period and area [13]. They are the comprehensive benefits of all four benefits [14]. Nowadays, research on land use benefits is no longer only confined to the evaluation of land use benefits. However, they also include the relation between land use benefits and other factors. Foreign scholars’ research on urbanization and land use benefits is mostly independent. The majority regard cities as the research background on land use benefits, or as a way to optimize the urban environment. They mainly studied the impact of changes in the land use structure on cities under urbanization. They also focused on the interrelation between urban land use and urban growth [15,16,17,18,19,20,21,22]. Some also analyzed changes and benefits of land use by methods such as Geographic Information System (GIS), remote sensing technology, and space measurement [23,24,25]. Comparatively, in China, there are many studies on correlation. These studies mainly focused on the coupling relation between intensive land use and urbanization level [26,27,28,29,30], coordinated relation between intensive urban land use and socioeconomic development [31,32,33], as well as coupling and coordination relation between urbanization and urban land use benefits, etc.Coupling is a concept in physics, referring to the motion of two (or more than two) systems affecting each other through interplay [34]. we often leverage it to measure the extent of the interplay among systems or motion. Comparatively, the coordination degree refers to a measure of the coordination degree among systems [35]. It can reflect whether the systems promote each other at a high level or restrict each other at a low one. Liao et al. first combined the degree of coupling with that of coordination. He proposed a model to measure the coupling and coordination degree (CCD) in a system or among multiple ones. What he has done made up for the defect that the coupling degree can only reflect the degree of interaction among systems, but not the level of development [35]. From then on, the CCD model has been widely used to evaluate the relationship between urbanization and land use benefits. Some scholars performed a time series analysis to study the coupling and coordination relation between urbanization and urban land use benefits. In their research, they took the major cities in China, such as Wuhan, Xining, Jinan, Shenzhen, as the research object [36]. However, most conducted research based on specific areas, such as the economic belt of Bohai Rim [37], Beijing–Tianjin–Hebei [38,39], Shaanxi–Gansu [40], three provinces in the Northeast [3], Middle Yangtze River Region [41], Guangxi Northern Gulf Economic Zone [42], as well as provinces of Shanxi [43], Zhejiang [44], Shenzhen [45], Jinan [46], etc. The main research methods include: entropy method [40,46], coefficient of variation method [3], gray correlation model [47], analytic hierarchy process (AHP) [45,48], data envelopment analysis (DEA) [49], Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) model [43] global principal component analysis [29,32], etc. It is found that the land use benefits and urbanization in many cities are still uncontrolled coordination. It is of significance to study the coupling and coordination relationship between the two. That is because it can, to some extent, prevent the urban area from developing in the form of “BIG PIE”. Consequently, it can also help realize the maximum comprehensive benefits of urban society, economy, ecology, and environment management. From the above analysis, we can conclude that in terms of research areas, researchers mainly focused on the more developed regions on the southeast coast, northeast as well as central and western regions. There are fewer studies on the relatively less advanced regions along the southeast coast. Fujian is the core area of the 21st Century Maritime Silk Road supported by the Party Central Committee [50]. Excellent geographical location and policy support in Fujian promote its development greatly. Nevertheless, its unique geographical feature of “eight mountains, one river, and one field” demonstrates the lack of available land resources. This defect severely limits the socioeconomic development and urbanization process of Fujian. Consequently, the intensive utilization of land, which contributes to the effective coordinated development of land use benefits and urbanization, is particularly significant for promoting the socioeconomic advancement of Fujian.In view of the above theoretical and practical background, this study, taking prefecture-level municipal districts and county-level cities in Fujian Province as the research object, builds an index system. This system is based on statistical data in 2002, 2005, 2010, 2015, and 2017, to evaluate the urban land use benefits and urbanization level. We also comprehensively leverage the Gini coefficient weighting method and the CCD model to study the spatial and temporal evolution of the coupling coordination degree (CCD). The contributions of this paper are as follows. Firstly, the utilization of the new weight determination method, the Gini coefficient weighting method (GCWM), can reflect the information more fully. Secondly, this paper, taking county-level cities and prefecture-level municipal districts as the research object, reflects the local situation more accurately.Fujian Province is situated at 23°33′ N–28°20′ N and 115°50′ E–120°40′ E southeastern China. It is on the coast of the East China Sea, across the Taiwan Strait, and opposite to Taiwan. It is an important estuary in mainland China (see Figure 1), adjacent to Zhejiang Province in the northeast, crossing the Wuyi Mountains and bordering with Jiangxi Province in the northwest, connecting with Guangdong Province in the southwest, as well as connecting the Yangtze River Delta and the Pearl River Delta. As of the end of 2017, Fujian Province has jurisdiction over 9 prefecture-level cities of Fuzhou, Xiamen, Quanzhou, Zhangzhou, Putian, Longyan, Sanming, Nanping, and Ningde (for convenience, we all call the districts of prefecture-level cities below according to the names of prefecture-level cities. For example, we call Fuzhou city as Fuzhou). There are a total of 29 districts. Fujian spans about 124,000 square kilometers of land, with 14 county-level cities and 44 counties. It has a permanent population of 39.11 million [51]. The territory of the regional city mainly consists of mountains and hills, accounting for more than 80% of the total area. River valleys and basins are covered with dense forests [52]. The sea area of Fujian, reaching 136,000 square kilometers, is slightly larger than the land. The coastline is long and winding, with unique harbor resources. The weather in Fujian is mostly the subtropical marine monsoon climate, which is warm and humid. Fujian’s forest overage rate has ranked first in the country for 40 years consecutively. It reached as high as 65.95% in 2017, thanks to its superior geographic location and natural conditions [51].In recent years, the economy in Fujian has developed rapidly. As shown in Table 1, Gross Domestic Product (GDP) increased from 0.45 trillion yuan in 2002 to 3.22 trillion yuan in 2017, with an annual growth rate of 14% [51,53]. Economic development has promoted the improvement of people’s living standards. From 2002 to 2017, the per capita disposable income of urban residents increased from 9189 yuan to 39,001 yuan, multiplied by nearly three times [51,53]. The per capita disposable income of rural residents increased from 3539 yuan to 16,335 yuan, multiplied by 3.6 times [51,53]. With economic development, the urbanization process is also steadily advancing. In 2002, the urban developed area was about 502.58 square kilometers, while increasing to 1516.88 square kilometers by 2017 [51,53]. The urban area has been swallowing huge amounts of land for agricultural and environmental uses. The scarcity of available land resources has gradually limited the development of urbanization. With the implementation of the “One Belt One Road” initiative in China and the deepening of economic globalization, Fujian has gained an increasingly prominent status. Consequently, it is of great significance to strengthen the research on the relationship between the urbanization in Fujian and the land use benefits. It is also necessary to explore the temporal and spatial evolution of its coordinated development proposing feasible plans. In this way, it is possible for us to promote the coordinated development and comprehensive management of cities in Fujian. Besides, we are also likely to avoid human-land conflicts and problems brought by urbanization. Finally, we can create a city cluster that is reasonably and scientifically organized.In view of data use, previous studies often used prefecture-level municipal districts data. However, it was difficult to accurately reflect the relationship between local land use benefits and the urbanization level. Moreover, compared with data of county-level cities and prefecture-level municipal districts, those of counties are incomplete. Consequently, we finally leverage the data of county-level cities and prefecture-level municipal districts in Fujian for this research. In terms of time, due to the missing data in some years, we select 2002, 2005, 2010, 2015, and 2017 as the time points for valuation. Among them, 2005, 2010, and 2015 are the end year of China’s 10th, 11th, and 12th Five-year Plan, respectively. They are of significance to the whole Five-year Plan. Besides, 2002 and 2017 are the earliest and latest data we can obtain. Many changes took place in some administrative regions within the study’s time interval. These changes include the abolishing of Jianyang City and the establishment of the Jianyang District of Nanping City in 2014, as well as the abolishing of Changle City and the establishment of Changle District of Fuzhou City in 2017 [51]. Consequently, the number of research objects were also changed. That is to say, there were 23 research objects in 2002, 2005, and 2010, 22 in 2015, and 21 in 2017. Considering this, we adjust the data accordingly. The data mainly come from the “Fujian Statistical Yearbook in 2003, 2006, 2011, 2015, 2017”, “China City Statistical Yearbook in 2003, 2006, 2011, 2015, 2017”. Note: China’s yearbooks record the previous year’s data, for example, “Fujian Statistical Yearbook 2003” records the data of Fujian Province in 2002.Missing data are supplemented by the statistical bulletin and statistical yearbook on the official website of the statistical bureau of each city.To mitigate the impact of different dimensions or magnitude orders of indices, we first perform dimensionless processing to the data. We divide the indices into two categories, positive index and negative index according to their effect on the system [54]. Then, we standardize the data by Equation (1).
+        (1)xstk={(Xstk−mtk)/(Mtk−mtk)(Positive    indicator)(Mtk−Xstk)/(Mtk−mtk)(Negative    indicator)The Xstk is the value and xstk is the standardized value. s is the number of regions, ranging from 1 to 23 (23 represents 2002, 2005, 2010; 22 represents 2015; and 23 represents 2017, respectively). t is the year, ranging from 1 to 5, representing 2002, 2005, 2010, 2015, and 2017, respectively. k is the number of indexes, ranging from 1 to 22. mtk refers to the minimum value of the k-th index in the t-th year of all regions, while Mtk refers to the maximum value of the k-th index in the t-th year of all regions. Except for indices of urban population density, the rest indices are positive ones. Besides, all indices fall in the interval [0,1] after standardization.Constructing a scientific index evaluation system is a prerequisite for evaluating the coupling degree of urban land use benefits and urbanization [55]. To ensure scientificity, integrity, hierarchy, and operability, this research, based on the idea of synergy in physics and previous research, constructs a comprehensive evaluation index system of urban land use benefits and urbanization in Fujian. Urban land benefits are the four in one of the economic, social, ecological, and environmental benefits [14]. To comprehensively reflect this connotation, instead of only focusing on economic benefits, we evaluate urban land use benefits from the four dimensions of economic, social, ecological, and environmental benefits. Urbanization has a multidimensional meaning. It mainly includes the four interacting aspects of population migration, economic development, spatial expansion, and improvement of living standards [56]. Therefore, we perform a comprehensive evaluation of the four dimensions of population, economic, social, and spatial urbanization. In this way, we can avoid the one-dimensional view of the spatial transfer of the rural population that the government often thinks.Specifically, to the index layer, the economic benefits of urban land use refer to the value of products and services produced per unit area of land. They can be measured by GDP, industrial output value, and fixed asset input [40]. Social benefits are mainly measured by population density, urban road area per capita, and developed area per capita [3]. For ecological benefits, the per capita park area, as well as the green coverage and area rate of the developed area, can fully reflect the urban ecological status. For environmental benefits, we usually use the pollutant compliance and removal rate for reflection. Due to the defects of statistical work in the statistical department, a large number of indices, such as industrial waste-water compliance rate and industrial solid comprehensive utilization rate, are missing. Consequently, we can only leverage the sewage and harmless treatment rate of domestic garbage to measure environmental benefits [57]. In the urbanization system, population urbanization refers to the process of transforming the agricultural population into the urban ones. Therefore, we leverage the rate of population urbanization and the number of nonagricultural populations to reflect the process of population urbanization [40]. In the process of urbanization, the economy is accumulating in the secondary industry and the tertiary one. Besides, the proportion of the tertiary industry keeps expanding. Therefore, we select per capita GDP, per capita industrial production value, and the proportion of tertiary industry to reflect economic urbanization [58]. In addition, social urbanization is a process that public service facilities continue to improve. Consequently, we choose indices, such as the number of hospital beds per 10,000 people, buses per 10,000 people, ordinary teachers per 10,000 people, and the total wages of urban employees, to reflect the process of social urbanization. We also reflect the process from multiple angles such as medical treatment, transportation, education, and wage level. The spatial urbanization is represented by the expansion of urban construction land, which is measured by urban construction land area and proportion of construction land [59].Finally, we determined 8 primary indices and 22 secondary indices, as shown in Table 2.The weighting of indices is the core of multiattribute decision-making. The weighting method mainly includes subjective and objective types. The subjective weighting method mainly weights by expert experience. However, this method is likely to be interfered with by subjective factors. Besides, it also does not fully utilize the information in the data. Comparatively, the core idea of the objective one is to give indices weight by comparing the content or differentiation of the data in indices. The more obvious the change in a certain index is, the richer the information the index contains, and the heavier the weight is [60]. Overall, the objective weighting method is more appropriate. Nevertheless, it is also a problem to choose the most suitable one from various objective weighting methods.The Gini coefficient is an important analysis index in economics. It is internationally leveraged to comprehensively measure the difference in the distribution of income among residents. Its method of calculation coincides with the core idea of the objective weighting method. For both, the greater the degree of data differentiation is, the larger the value is. The Gini coefficient weighting method draws on this method, weighting each index by deriving the Gini coefficient value. It comprehensively reflects the difference between any two data in the same index. It fully utilizes the value of information. It scientifically and objectively reflects the differentiation (distinction) of the data in a certain index. In addition, the definition of the Gini coefficient itself eliminates the effect of dimensions. Consequently, there is no need for us to standardize data in advance. As a result, it avoids the loss of information in data processing. It simplifies the calculation. Besides, its applicability is stronger and order preservation is better, compared to other objective weighting methods [61]. Therefore, we leverage the Gini coefficient weighting method in our index system. The steps are as follows [61]:
+		Calculate the Gini coefficient value of the evaluation indices, as shown in Equations (2) and (3):
+		(2)Gk=∑i=1n∑j=1n|Xki−Xkj|2n2μk (k =1,2,…,m and μ≠0)
+        (3)Gk=∑i=1n∑j=1n|Xki−Xkj|n2−n (k =1,2,…,m and μ=0)
+        where Gk is the Gini coefficient value of the k-th index. m is the number of evaluation indices. n is the number of samples of the indices. Xki refers to the i-th sample of the k-th index. μk refers to the sample of the k-th index. In particular, when the average value of the index data a not 0, the Gini coefficient value is calculated by Equation (2). When it is 0, the value is calculated by Equation (3).Calculate the Gini coefficient value of the evaluation indices, as shown in Equations (2) and (3):
+		(2)Gk=∑i=1n∑j=1n|Xki−Xkj|2n2μk (k =1,2,…,m and μ≠0)
+        (3)Gk=∑i=1n∑j=1n|Xki−Xkj|n2−n (k =1,2,…,m and μ=0)
+        where Gk is the Gini coefficient value of the k-th index. m is the number of evaluation indices. n is the number of samples of the indices. Xki refers to the i-th sample of the k-th index. μk refers to the sample of the k-th index. In particular, when the average value of the index data a not 0, the Gini coefficient value is calculated by Equation (2). When it is 0, the value is calculated by Equation (3).Calculate the Gini coefficient weight of the evaluation index:After calculating the Gini coefficient values Gk of m indices in the evaluation system by Equation (2) or (3), and then normalizing them, the Gini coefficient weight gk of the k-th index can be derived, as shown in Equation (4):(4)gk=Gk∑i=1mGi,
+        where gk is the weight of the Gini coefficient of the k-th index. Gk is the value of the Gini coefficient of the k-th index. m is the number of evaluation indices. The sum of all weights is 1. The specific weight values are shown in Table 2.The relative development degree is a specific index that reflects the urban land use benefits as well as the relative development degree and status of urbanization in a certain period. The calculation steps are as follows [58].Calculate the overall benefits of urban land use benefits and urbanization system, as shown in Equations (5) and (6):(5)fst(x)=∑k=1m1akxstk    (s=1,2,…,n1 and t=1,2,…,n2),
+        (6)gst(y)=∑k=1m2bkystk    (s=1,2,…,n1 and t=1,2,…,n2),Among them, fst(x) and gst(y) are the overall benefits of the urban land use benefits and urbanization system in the t-th year in the s area. n1 and n2 refer to the sample sizes in areas and years, respectively. xstk and ystk refer to the standardized values of the k-th index in the t-th year in the s-th area, respectively. ak and bk refer to the weights of the corresponding indices, respectively. m1 and m2 refer to the numbers of indices in the urban land use benefits and urbanization system, respectively. The two are added as the indices in the evaluation system, that is, m1+m2=m.Calculate the relative development degree, as shown in Equation (7):(7)Est=fst(x)gst(y)    (s=1,2,…,n1 and t=1,2,…,n2),
+        where Est is the relative development degree of urban land use benefits and urbanization in the s-th area in the t-th year. When Est>1, we call it an advanced city; when Est=1, we call it a synchronous city; when Est<1, we call it a lagging city.Refer to Tang’s studies, we build the CCD model by the following steps [62], as shown in Equations (8)–(10):(8)Cst(x,y)=2fst(x)×gst(y)[fst(x)+gst(y)]2    (s=1,2,…,n1 and t=1,2,…,n2),
+        (9)Tst(x,y)=αfst(x)+βgst(y)    (s=1,2,…,n1 and t=1,2,…,n2),
+        (10)Dst(x,y)=Cst×Tst    (s=1,2,…,n1 and t=1,2,…,n2),Among them, Cst is the coupling degree of urban land use benefits and urbanization of the s-th area in the t-th year, and Cst∈[0,1]. Tst is the coordination degree of urban land use benefits and urbanization of the s-th area in the t-th year. Dst is the coupling coordination degree of urban land use benefits and urbanization of the s-th area in the t-th year, and Dst∈[0,1]. α and β are the contributions of the urban land use benefits system and the urbanization system, respectively. According to the existing literature, we cannot conclude whether the urban land use benefits and the urbanization are more important, so we take 0.5 for both α and β. According to the existing research, we divide the CCD into three stages and 10 categories [63], as shown in Table 3.It can be seen from Table 2 that the difference between the weights of the secondary indices is huge. The largest is 0.1084, while the smallest is 0.0062, 17.48 times as large as the latter. Among the 22 secondary indices, there are three indices, concentrated on economic benefits and spatial urbanization, which are above 0.1. There are five indices between 0.05 and 0.099. Fourteen are below 0.05. As shown in Table 4, among the primary indices, the weight of the economic benefit index is the largest, close to 1/3 of the total. Besides, the difference between the weight of the ecological benefits index and the environmental benefits index is not large, far lower than the weights of other primary indices. To conclude, it shows that the gap between ecological and environmental benefits among regions is narrow. However, the gap between economic benefits and spatial urbanization is huge, comparatively.From Figure 2, in the five years of 2002, 2005, 2010, 2015, and 2017, the relative development degree of nine prefecture-level municipal districts and 14 county-level cities in Fujian Province did not take 1. This shows that the urban land use benefits and urbanization in various regions did not develop synchronously in each year. Considering the overall trend, the type of mainstream cities has transferred from advanced cities to lagging ones. From 2002 to 2017, the number of advanced cities was 19, 15, 12, 12, and 10, respectively, with the proportions of 82.6%, 65.2%, 52.2%, 54.5%, and 47.6%. This situation is more obvious in prefecture-level municipal districts. The number of lagging prefecture-level municipal districts was two in 2002, while reducing to 0 by 2017. Among them, Fuzhou and Xiamen were lagging cities in the past five years.The relative development degree is derived by the ratio of the urban land use benefits and the urbanization level. Considering this, we cannot help asking whether urban land use benefits are lower in areas with a low relative development degree. Taking 2017 as an example, we make a comparison chart of urban land use benefits and urbanization level in various regions, as shown in Figure 3.From Figure 3, the urban land use benefits in areas with relatively low development are higher. Correspondingly, the urbanization level is also higher. From the perspective of the administrative level, the urban land use benefits and urbanization level of prefecture-level municipal districts, such as Fuzhou, Xiamen, Zhangzhou, Quanzhou, Sanming, and Longyan, are generally higher than their county-level cities. From the perspective of the regional economic development level, Fuzhou, Xiamen, Quanzhou, Zhangzhou, and other well-developed regions have higher levels of urbanization. However, the discrepancy of land use benefits between well-developed regions and others is much narrower than that of urbanization levels.From Table 5, we can see that in the five years of 2002, 2005, 2010, 2015, and 2017, regions whose coupling degree C of urban land use benefits and urbanization level were above 0.9 accounted for 100%, 100%, 95.6%, 95%, and 100%, respectively. Only ratios of Xiamen in 2010 and Shaowu in 2015 were below 0.9, while still between 0.8 and 0.9. This variable is relatively stable. The coupling coordination degree D was between 0.2 and 0.6. The change is relatively obvious. From Figure 4, it can be seen that the coupling coordination degree is mainly in four stages. They are moderately uncoordinated, slightly uncoordinated, at the edge of being uncoordinated, and barely coordinated. It is moderately uncoordinated and slightly uncoordinated that dominate in them. These two stages accounted for 74%, 74%, 82.6%, 68.2%, and 66.7% of the previous years, respectively. No region has reached the stage of coordinated development. However, most have moved from moderately uncoordinated to slightly uncoordinated, with an overall trend to develop well. Considering the stage of each region, there are obvious trends and laws of distribution.In view of the time series, from 2010 to 2015, there were eight cities that achieved the improvement of the coupling coordination phase. The number far exceeded the total of other adjacent years. In addition, only the coupling coordination phase in one city had a reduction. This number of reductions was also far lower than that in other adjacent years. From the perspective of spatial distribution, regions with relatively high coupling coordination, such as Xiamen, Fuzhou, Quanzhou, and Shishi, are all coastal regions. Except for Sanming, it is an inland city. Moreover, Xiamen and Fuzhou become the coastal central cities while Sanming is the inland central city. Consequently, there grows a sort of “central-peripheral” development pattern. That is, for the surrounding cities of these three central cities, the closer to the central city they are, the higher their coordination degree is, and vice versa.The type of mainstream cities in Fujian Province has changed from leading cities to lagging ones. It reflects that in the urbanization process, the development of urban land use benefits lags behind urbanization. This also confirms that the rapid development of urbanization has brought many defects, such as the spreading development of urbanization and inefficient land use [64,65]. The main reason is that, in the process of urbanization, local governments blindly expand urban areas. They overpursue urbanization while neglecting the efficient utilization of land. The “BIG PIE” policy has resulted in low urban land use benefits, far behind the development of urbanization. Moreover, the deeper reason is related to the assessment mechanism of government officials. The assessment mainly considers economic indices. Since urban land expansion has a motivating effect on economic development [66,67], government officials are encouraged to expand urban areas. It causes that officials only focus on socioeconomic benefits in land planning, while ignoring the ecological and environmental benefits. As a result, it finally leads to the fact that urban land use benefits are lower than the urbanization level.Besides, we find that the urban land use benefits and urbanization level are related to the administrative level and economic development of the city. It is manifested that areas with higher administrative and economic development levels tend to own higher urban land use efficiency and urbanization level. However, the impact of economic conditions on urbanization is greater than that of the urban land use benefits. The reason is not complex to explain. One the one hand, areas with high administrative level will receive more policy support and resource tilt [68]. As a result, it is more beneficial to urban construction and economic development. Comparatively, economically well-developed areas tend to pay more attention to economic and social benefits. Nevertheless, economic development often comes at the price of the environment. This partially hinders the improvement of overall benefits. Consequently, the difference in urban land use benefits among cities is insignificant.Moreover, it is also found that the CCD is still at a relatively low level in various regions of Fujian. However, it develops in a good direction. There are situations of high coupling and low coupling coordination. This is mainly because the coupling degree only indicates the strength of the effect between urban land use benefits and the urbanization level. However, we do not know whether they promote or inhibit each other. Comparatively, the coupling coordination degree fully considers the coordination degree between them [69]. Early urban land use benefits limited to urbanization. However, the two gradually promote each other as time goes by. This situation does not only occur in Fujian exclusively. Jia et al. have studied the three major urban agglomerations, including the Yangtze River Delta, the Pearl River Delta, and the Beijing–Tianjin–Hebei area, which are relatively better-developed [70]. They obtained the same result. Similarly, Zuo et al. also got the same result when studying the Shaanxi–Gansu-Ningxia region, which is of lower development level [40]. It requires further study of whether this situation is suitable for the whole country. However, the increase in the CCD took place in large numbers from 2010 to 2015. It far exceeded the sum of other years. This may partly benefit from the establishment of the Western Taiwan Straits Economic Zone in 2009 and the peaceful cross-strait relations.Finally, we find that each city generally exhibits a development pattern according to its distance from the three central cities of Fuzhou, Xiamen, and Sanming. That is, the closer to the central city it is, the higher the coordination degree it has; vice versa. What is more, cities with developed coastal transportation have a higher coupling and coordination degree than inland cities. Zhang et al. [3] and Wang et al. [58] studied the three provinces in Northeast China and the Bohai rim area, respectively, to obtain this “center-periphery” pattern. Zhang et al. found that cities with relatively high coupling and coordination degrees are distributed along the Harbin–Dalian Transportation Economic Belt in strips [3]. This finding is so interesting. It shows that the development of regions not only depends on their own influencing factors, but also on the radiation and leading role of the central cities. The farther away the distance is, the weaker this radiation and driving effect is. Meanwhile, developed traffic can also strengthen this driving effect. This provides a way for the government to give full play to the leading role of central cities. However, there are still several questions that require us to perform further research and practice. For example, we should consider how to leverage the central city’s radiation and driving role more efficiently. This is significant to promote the coordinated development of land use benefits and urbanization in other cities.When it comes to research methods, the research methods of this study are applicable to various areas with similar issues, such as urban agglomerations, provinces, and cities. However, the completeness of the data in each region, the difference of regional conditions, and the different perspectives of the target connotation will all affect the construction of the index system. The universal system still needs to be tested in practice. Besides, the selection of the index weighting method will also have an impact on the results. Weights, computed by the objective weighting method, are based on the information of the data itself [61]. Due to the discrepancy in the data of various regions, weights will change accordingly. Consequently, weights obtained in this study cannot be directly applied to other regions.In addition, this research also has several drawbacks, which are likely to be solved in future research. (1) This study only selects 2002, 2005, 2010, 2015, and 2017 as time nodes, causing the time series data to be insufficient. Consequently, we can only tentatively study the spatial and temporal evolution of urban land use benefits and urbanization in various regions. Therefore, future research can leverage continuous-time data to analyze its evolution in more detail. (2) Considering a large number of county statistical data is missing, we only take prefecture-level municipal districts and county-level cities as research samples. However, the county is also an important part of China’s urbanization process. As a result, there are limitations in the spatial analysis of urban land use benefits and urbanization in regions of Fujian; (3) In the construction of the index system, some data are missing. For example, compared with other studies, some important indices, such as the proportion of the tertiary industry’s employed population [59], urban residents per capita use area [57], and the three-waste treatment rate [54], are not included. Consequently, it requires us to add more comprehensive data in future relevant studies to improve the index system, reflecting the spatiotemporal evolution of the CCD of each city more accurately.It is of practical significance to study the law of spatiotemporal evolution of the coupling and coordination relation between urban land use benefits and urbanization level in Fujian. It is likely to promote the healthy development of the urbanization process. Therefore, this study takes county-level cities and prefecture-level municipal districts in Fujian Province as the research object, leveraging the data of 2002, 2005, 2010, 2015, and 2017 to construct the evaluation index system of urban land use benefits and urbanization system. We derive the CCD of the index system and analyze the temporal and spatial evolution law. The conclusions of this study are as follows:
+	  (1)Urban land use benefits and urbanization levels are positively correlated with the regional administrative level and economic development status.(2)The CCD of urban land use benefits and urbanization levels in various regions of Fujian is still low. However, the overall development direction is good.(3)In terms of spatial distribution, the CCD has a “center-periphery” pattern. That is, the closer to the three central cities of Fuzhou, Xiamen, and Sanming it is, the higher the coordination degree it has; vice versa.Urban land use benefits and urbanization levels are positively correlated with the regional administrative level and economic development status.The CCD of urban land use benefits and urbanization levels in various regions of Fujian is still low. However, the overall development direction is good.In terms of spatial distribution, the CCD has a “center-periphery” pattern. That is, the closer to the three central cities of Fuzhou, Xiamen, and Sanming it is, the higher the coordination degree it has; vice versa.The above findings can provide certain implications for policy formulation. First of all, in the process of urbanization, Fujian Province cannot blindly pursue the speed. It is necessary to plan urban expansion rationally and strengthen the intensive use of land, avoiding the development model of “BIG PIE”. It is also significant to add some indices on the ecological environment to the assessment of government officials. It can avoid only concentrating on the economy while neglecting the ecological environment. It can finally promote the overall improvement of urban land use benefits. Secondly, it is necessary to give targeted priority to the improvement of the urban land use benefits or urbanization level, based on the reality of different regions. Finally, it is recommendable to strengthen the radiation and the leading role of central cities to surrounding ones. Appropriate policy and resource support for peripheral cities should be provided, such as financial subsidies for regional construction.Conceptualization, K.W., Y.C. and M.Y.; data curation, Y.C. and L.S.; formal analysis, Y.T.; funding acquisition, Y.T.; investigation, P.W.; methodology, K.W. and X.J.; project administration, L.S. and P.W.; resources, Y.C.; software, Y.C. and L.S.; supervision, K.W.; visualization, P.W.; writing—original draft, K.W. writing—review and editing, Y.T., X.J. and M.Y. All authors have read and agreed to the published version of the manuscript.This research was funded by the National Natural Science Foundation of China (No. 71072066), Sichuan University (No. SKGT201602, No. 2018HHF-42), and the Department of Science and Technology of Sichuan Province (No. 2018JY0594).All authors declare no conflicts of interest.The location of the study area.The relative development degree of urban land use efficiency and urbanization level.Comprehensive value of urban land use benefits and urbanization level by region in 2017.The spatial distribution diagram of the regional coupling coordination degree. (a)—The spatial distribution diagram of coupling coordination degree in 2002.; (b)—The spatial distribution diagram of coupling coordination degree in 2005. (c)—The spatial distribution diagram of coupling coordination degree in 2010.; (d)—The spatial distribution diagram of coupling coordination degree in 2015.; (e)—The spatial distribution diagram of coupling coordination degree in 2017.; (f)—The spatial distribution diagram of Variation type of coupling coordination degree.Basic economic information of Fujian Province.GDP—Gross Domestic Product; PCDIUR—the Per Capita Disposable Income of Urban Residents; PCDIRR—the Per Capita Disposable Income of Rural Residents.Evaluation index system of urban land use benefits and urbanization level.GDP—Gross Domestic Product; CNY—China Yuan.Discriminating standards of the coupling coordination degree.Weights of primary indices.Coupling coordination degree (CCD) of urban land use benefits and urbanization level in various regions.C—the Coupling Degree of Urban Land Use Benefits and Urbanization; D—the Coupling Coordination Degree of Urban Land Use Benefits and Urbanization.

Med-MDPI/ijerph_5/ijerph-17-16-05648.txt

@@ -0,0 +1 @@
+These authors contributed equally to this work.Over the past two decades, there have been two major outbreaks where the crossover of animal Betacoronaviruses to humans has resulted in severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV). In December 2019, a global public health concern started with the emergence of a new strain of coronavirus (SARS-CoV-2 or 2019 novel coronavirus, 2019-nCoV) which has rapidly spread all over the world from its origin in Wuhan, China. SARS-CoV-2 belongs to the Betacoronavirus genus, which includes human SARS-CoV, MERS and two other human coronaviruses (HCoVs), HCoV-OC43 and HCoV-HKU1. The fatality rate of SARS-CoV-2 is lower than the two previous coronavirus epidemics, but it is faster spreading and the large number of infected people with severe viral pneumonia and respiratory illness, showed SARS-CoV-2 to be highly contagious. Based on the current published evidence, herein we summarize the origin, genetics, epidemiology, clinical manifestations, preventions, diagnosis and up to date treatments of SARS-CoV-2 infections in comparison with those caused by SARS-CoV and MERS-CoV. Moreover, the possible impact of weather conditions on the transmission of SARS-CoV-2 is also discussed. Therefore, the aim of the present review is to reconsider the two previous pandemics and provide a reference for future studies as well as therapeutic approaches.Coronaviruses (CoVs) are a group of highly enveloped viruses that are diversely found in humans and wildlife. With their high mutation rate and infectivity, CoVs are important zoonotic pathogens that can infect animals [1,2] and humans, leading to 5–10% of acute respiratory syndromes [3]. Apart from infecting a variety of economically important vertebrates (such as pigs and chickens), six species have been identified to cause disease in humans [4]. They are known to infect respiratory, gastrointestinal, hepatic and neurologic systems with a wide range of clinical features from asymptomatic course to severe disease that require hospitalization in the intensive care unit [4,5]. The first human coronaviruses (HCoVs), HCoV-229E and OC43, shown to be significant respiratory pathogens, were identified in the 1960s [6,7]. However, it is assumed that the first recorded coronavirus-related disease was feline infectious peritonitis (FIP) in 1912 [8]. The “corona”-like or crown-like morphology of these viruses leads to choose the name “coronavirus,” in 1968 [6].Coronaviruses were not considered as highly pathogenic for humans before the beginning of the 21st century. Afterward, two highly pathogenic HCoVs, including severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV), emerging from animal reservoirs, have led to global epidemics of deadly pneumonia in humans with high morbidity and mortality [9,10]. In December 2019, seven years after MERS outbreak, the third pathogenic HCoV emerged in Wuhan, the capital city of Hubei province in China, causing severe pneumonia [11,12]. Considered as agents that are a great public health threat, an epidemiological alert was placed by the World Health Organization (WHO) and this new coronavirus was named SARS-CoV-2 and the related respiratory disease COVID-19 (https://www.who.int). Compared with SARS-CoV, SARS-CoV-2 appears to be more readily transmitted from human-to-human, spreading to multiple continents and the outbreak of SARS-CoV-2 was declared on January 30, 2020 [13] (https://www.who.int). In this review, we will introduce the current knowledge on the origin and evolution of SARS-CoV-2, emphasizing its characteristics and its genetic diversity from previous coronaviruses, with a brief comment on its epidemiology and pathogenesis. We also highlight the environmental factors involved in virus transmission. Knowledge about this novel coronavirus is rapidly evolving, and efforts must be implemented in order to protect the populations by reducing transmission and controlling the spread of this fatal disease.According to the International Committee on Taxonomy of Viruses, CoVs are classified under the order of Nidovirales, a family of Coronaviridae and subfamily of Coronavirinae [14]. Based on previous serologic and recent genomic evidences, the family of Coronaviridae encompasses two subfamilies: subfamily Orthocoronavirinae and subfamily Torovirinae (Figure 1) [7,15]. The subfamily of Orthocoronavirinae consists of four genera: Alphacoronavirus, Betacoronavirus, Gammacoronavirus and Deltacoronavirus [7,16,17].CoVs can be isolated from different animal species, including birds, livestock and mammals such as camels, bats, masked palm civets, mice, dogs and cats [18,19]. Animal CoVs are known to cause acute diseases in several animals and could be responsible for economic losses in domestic animals or birds [20,21]. Domestic animals may play an important role as intermediate hosts that enable virus transmission from one species to humans [17]. The genera Gamma- and Deltacoronavirus infect birds, but some of them can also infect mammals [16]. These animal CoVs include transmissible gastroenteritis virus (TGEV), porcine epidemic diarrhea virus (PEDV), avian infectious bronchitis virus (IBV)—and more recently—swine acute diarrhea syndrome coronavirus (SADS-CoV). However, animal CoVs can also infect humans that can spread the infection through human-to-human transmission [17,22]. On the other hand, Alpha- and Betacoronavirus infect only mammals and usually cause respiratory illness in humans; among these, strains 229E, OC43, HKU1 and NL63 are the most widespread infecting young children, infants as well as elderly individuals [23,24,25]. The high rates of mutation characterizing all RNA viruses [23,26], the evolving nature of CoVs and the simplicity of transmission from one species to another are the most relevant features learned from SARS-CoV and MERS-CoV previous outbreaks [15,23,25]. Importantly, most of Alpha- and Betacoronavirus were found only in bats, and many genetically diverse coronaviruses phylogenetically related to SARS-CoV and MERS-CoV have been discovered in diverse bat species worldwide [17]. Therefore, HCoVs such as SARS- and MERS-CoVs seem to have originated in bats by sequential mutations and recombination, including those occurring in the intermediate hosts, civets and raccoon dogs for SARS-CoV and camels in the case of MERS-CoV, finally acquiring the ability to infect humans [15,17]. Comparative genome studies published in recent papers strongly support the hypothesis that SARS-CoV-2 originated in bats and that pangolins (Manis javanica) acted as intermediate mammalian hosts [11,27] (Figure 2). Indeed, the genetic sequence of the SARS-CoV-2 showed more than 79% nucleotide identity with the sequence of SARS-CoV and 50% with MERS-CoV [17,19]. The high degree of homology of the angiotensin-converting enzyme 2 (ACE2) receptor in several animal species can be considered as an additional evidence to support that SARS-CoV-2 originated from bats [28]. Based on findings from molecular studies, the ACE2 proteins of non-human primates, pigs, cats and ferrets closely resemble the human ACE2 receptor. Therefore, these species may be susceptible to SARS-CoV-2 infection, as has been shown for SARS-CoV. Although a recent study showed that neither pigs nor chickens are susceptible to SARS-CoV-2 by intranasal or oculo-oronasal infections, more evidences are needed to exclude pigs as intermediate host of SARS-CoV-2 [29].Based on the genetic sequence identity and the phylogenetic reports, SARS-CoV-2 is sufficiently different from SARS-CoV; thus, WHO has classified it as a new Betacoronavirus that infects humans [30].The genome of HCoVs is a single-stranded positive-sense RNA (+ssRNA) (~26–32 kb) with 5′-cap structure and 3′-poly A tail, which is among the largest known RNA genomes [31,32,33]. The typical HCoVs gene order is 5′-replicase-S-E-M-N-3′, with numerous (6 to 11) open reading frames (ORFs) encoding accessory proteins scattered among the structural genes [34,35]. The first ORFs (ORF1a and 1b) comprise two-thirds (approximately 67%) of the genome length and encode 16 nonstructural polyproteins (nsps 1–16) and are directly translated from the genomic RNA [17]. There is a −1 ribosomal frameshift between ORF1a and ORF1b, leading to the production of two large replicase polypeptides (pp): pp1a and pp1ab. These polypeptides are further processed by two virally encoded cysteine proteases, the papain-like protease (PLpro) and a 3-chymotrypsin-like protease (3CLpro) into 16 nsps [3,33,36]. There are at least four structural proteins encoded by the coronaviral genome: a spike glycoprotein (S), an envelope protein (E), a membrane protein (M) and nucleocapsid protein (N) with short untranslated regions at both termini, required to produce a structurally complete viral particle [37]. The typical coronavirus virion structure and proteins are shown in Figure 3. The M protein is in higher quantities in comparison to any other proteins in the virus particle; with its three transmembrane domains, it shapes virions, promotes membrane curvature and binds to the nucleocapsid [38,39]. The N protein contains two domains, both of which can bind to nsp3 protein to help tether the genome to replication–transcription complex (RTC) and package viral RNA into the viral particle during viral assembly [39,40]. The E protein is involved in virus assembly and virion release from host cells, while the S protein plays a vital role in attachment to host receptors, viral entry and determines host tropism [41,42]. Additionally, some coronaviruses, such as HCoV-OC43 and HCoV-HKU1, have a hemagglutinin-esterase (HE) gene between ORF1b and S [43,44,45,46]. This hemagglutinin, like the influenza homolog enzyme, binds to sialic acid on host cell-surface glycoproteins and possesses acetyl-esterase activity [47]. Besides coronavirus-conserved genes, the SARS-CoV, SARS-CoV-2 and MERS-CoV genomes contain several specific accessory genes including ORF3a/b, 4a/b, ORF5, ORF6, ORF7a/b, ORF8a/b and 9b (Figure 4) [4,48,49]. All the structural and accessory proteins are translated from subgenomic RNAs (sgRNAs) generated during genome transcription/replication of CoVs [4].Attachment, cell entry, translation of viral replicase, genome replication, translation of structural proteins and virion assembly and release are the phases of coronavirus replication cycle [4,50]. SARS-CoV, MERS-CoV and SARS-CoV-2 bind to different host receptors to gain entry into host cells [4,51,52]. Viral entry is mediated by the transmembrane S glycoprotein that comprises two functional subunits (S1 and S2 subunits) responsible for receptor recognition and viral-host cell membranes fusion, respectively [53,54]. S1 receptor-binding domain (RBD) mediates binding to the cognate host cell receptor; however, the S2 domain mediates the fusion events, between viral envelope and host cell membrane [52,55,56]. As recently found, SARS-CoV-2 uses the same ACE2 receptor [57], as SARS-CoV, whereas MERS-CoV uses dipeptidyl peptidase 4 (DPP4, also known as CD26) receptor (Table 1) [58]. The fusion of the S protein to the plasma membrane of host cell generates a double membrane vesicle in the host cell, thereby allowing release of the nucleocapsid into the cytoplasm, followed by genome transcription [53,54]. Upon entry into the cell, virus-specific RNA and proteins are synthesized, probably entirely in the cytoplasm. Translation starts with the expression of two polyproteins, pp1a and pp1ab, which undergo co-translational proteolytic processing into the proteins that form the replicase complex. This complex is used to transcribe a 3′-coterminal set of nested subgenomic mRNAs, as well as genomic RNA that have a common 5′ “leader” sequence derived from the 5′ end of the genome. New virions are assembled by budding into intracellular membranes of the pre-Golgi compartment and released through the cell secretory mechanisms [4,42,48,50].Virologic as well as genetic studies have demonstrated that bats are reservoir hosts of both SARS-CoV and MERS-CoV, but also that they can use other species as intermediate hosts before spreading to humans [59,60]. The detection of two genomes distinct from known swine in ill piglets were reported by two independent groups [61,62]. The phylogenetic analyses showed that these novel swine enteric Alphacoronaviruses (SeACoVs) were strongly related to the Rhinolophus bat coronavirus HKU2 isolated in Guangdong Province, in southern China [61,62]. This suggests that coronaviruses of bat origin may have ‘jumped’ the barrier of the species to infect pigs as intermediate hosts. The CD26 receptor sequence alignment between humans and pigs demonstrates a 94.5% overlap, which is sufficient for the possible cross-species transmission [63]. It has also been documented that pigs are susceptible to human SARS-CoV [64] and MERS-CoV infections [65]. The large number of mutations within the RBD enabled viruses to infect new hosts, representing a potential threat for both animal and human health. In southern China, the unique climate, the high density of domestic as well as wild pigs, along with the extensive bat distribution and carriage of tremendous quantities of recombinant novel coronaviruses may result in the appearance of more novel coronaviruses in the future [66]. It is generally acknowledged that numerous viruses have existed and were restricted to their natural reservoirs for lengthy times [17]. The consistent spillover of viruses from natural hosts to humans and other species is essentially related to human activities, including urbanization and modern agricultural practices, leading to the constant human exposure to the ever-changing mutant CoVs from their reservoirs [15,17]. The close contact between humans and animals and the practice of eating raw meat are both risk factors for causing a new human CoV outbreak [15]. Hence, COVID-19 should be considered as a zoonotic disease that spread from animals to humans.Following the first SARS-CoV-2 outbreak in seafood and wildlife market in Wuhan, secondary cases started to be identified after ten days. Although these new patients did not have any contact with the market, they had a history of contact with people who attended the market [60]. Therefore, similarly to SARS-CoV and unlikely to MERS-CoV, human-to-human transmission for SARS-CoV-2 has been reported and is currently considered as the main type of transmission worldwide [5,19]. On January 13, 2020, Thailand announced the first non-Chinese case of infection that spread from the Chinese provinces, to the Asian continent [60]. This case was a Chinese tourist who has traveled to Thailand and did not have any epidemiological link to the market [30]. More recently, Forster et al., by using phylogenetic analysis based on nucleotide mutations of 160 complete human SARS-CoV-2 genomes found that three variants of SARS-CoV-2 (A as the ancestral type, plus B and C) represent the bat outgroup coronaviruses. In particular, the A and C types were found mostly in European- American patients, whereas the B type was common in East Asia suggesting that this kind of analysis could help in following the evolution of SARS-CoV-2 [67].It was demonstrated that SARS-CoVs have adapted themselves to bind to human ACE2 receptor and infect human cells effectively [68]. This form of adaptation required a series of amino acid changes in the RBD within the S protein of SARS viruses that circulated in bats [56,68]. Therefore, the human-to-human transmission that was seen in the course of the SARS-CoV outbreaks is directly attributable to the ability of SARS-CoVs to adapt their S protein to efficiently bind to human ACE2, thus infecting ciliated bronchial epithelial cells and type II pneumocytes [15,69]. Similar to SARS-CoV, ACE2 is also used by SARS-CoV-2 as the entry receptor in the ACE2-expressing cells, suggesting that SARS-CoV-2 has a life cycle similar to SARS-CoV [56,68]. As outlined before, SARS-CoV S protein regulates the receptor binding and membrane fusion activities determining host tropism and transmission capacity. Several evidences highlighted a higher binding affinity of SARS-CoV-2 RBD to the ACE2 receptor. In particular, molecular and in silico analyses demonstrated that SARS-CoV-2 RBD conformation and amino acid composition enhance the bonding between the S protein and ACE2 receptor [51,70,71]. A recent biophysical and structural analysis of the SARS-CoV-2 S protein showed that it binds to ACE2 receptor with about 10- to 20-fold higher affinity than the S protein of SARS-CoV [52,72]. This high affinity could account for its extreme infectivity among human populations. Another feature of the powerful infectivity of SARS-CoV-2 is that the shedding pattern of viral particles in SARS-CoV-2 diagnosed patients is similar to that of influenza patients in which viral loads at the time of symptom onset are higher and gradually decrease within days; interestingly, this pattern seems to be different from that reported for SARS-CoV patients where the highest shedding is reported 10 days after the onset of symptoms [20,73,74]. These results indicate that SARS-CoV-2 can spread more easily than SARS-CoV in the community even in the absence of symptoms or when only initial mild symptoms are present [75].The human-to-human transmission of SARS-CoV-2 mainly occurs by inhalation of respiratory droplets spread by coughing or sneezing from an infected individual, but also by direct contact of contaminated surfaces and then touching the nose, mouth and eyes [24,76,77,78]. The virus was shown to remain stable in favorable atmospheric conditions on different surfaces for days [79]. Additionally, transmission in an unventilated environment or closed spaces due to high aerosol concentrations has been suggested [76,77]. In agreement, the presence of SARS-CoV-2 in the surfaces of the houses of confirmed patients was reported, further strengthening this mode of contact transmission. Moreover, live viruses were also found in the stool of COVID-19 patients, as previously found for both SARS-CoV and MERS-CoV [77]. Given its capacity for survival in feces and the expression of ACE2 within intestine, it was demonstrated that SARS-CoV-2 can infect these tissues and can be released in feces; therefore, water supply contamination and fecal-oral route transmission is also hypothesized [24,80]. However, at present, there have not been reported cases of fecal-oral transmission of the virus. Studies have also indicated that SARS-CoV-2 transmission via ocular surfaces should not be overlooked, as contaminated droplets and body fluids could easily infect the human conjunctival epithelium [81]. SARS-CoV-2 is also responsible for cluster transmission, in particular within family clusters [77]. In some cities, 50% to 80% of all reported cases of COVID-19 accounted for cluster transmission [82]. Based on the current information, there is no evidence for transplacental transmission from infected pregnant women to their fetus, who underwent caesarean section [24,78]. Therefore, whether transmission during vaginal birth can occur remains to be established, neonatal COVID-19 disease as postnatal transmission was documented [83]. Although, SARS-CoV-2 may definitely infect infants, it has been reported that neonates, infants and children develop significantly milder forms of the disease than their adult counterparts [24,84].Coronaviruses are responsible for 5–10% of acute respiratory illness while it has been estimated that 2% of the population is deemed as an asymptomatic carrier of these viruses. The first discovered HCoV was IBV that causes respiratory disease in human whereas, HCoV-229E and HCoV-OC43, which cause the common cold in humans [15,26,85]. They were not considered to be highly pathogenic for humans until the outbreak of SARS in Guangdong state of China in 2002 and 2003. SARS-CoV infected more than 8000 people worldwide and caused 916 deaths (Table 2), representing a mortality rate by around 10% [86]. Ten years later in 2012, MERS-CoV emerged in Saudi Arabia and infected over 2494 people with 858 deaths, accounting for a mortality rate approximately of 35% [9,24,87]. Starting in China in December 2019, there were reports of patients presenting severe viral pneumonia [15,88,89]. This public health concern resulted in many unknown pneumonia cases who were admitted to local hospitals [22,78] (https://www.who.int). Primary etiologic investigations performed in those patients showed that they were epidemiologically linked to a Huanan wholesale seafood market that also traded live animals and wildlife [17,24,90]. By January 7th, 2020 Chinese authorities announced that a new type of coronavirus was isolated [60,91]. The epicenter of infection was probably linked to a zoonotic pathogen being present in the seafood and exotic animal wholesale market [60,91]. The rapid increasing numbers and rate of fatalities indicated a second mode of transmission, from human-to-human, that allowed viral spreading primarily in other Asian countries such as South Korea and Iran followed by many countries such as Italy, Spain, Germany, France, Brazil and USA [24,60]. It is very intriguing to note that the SARS epidemic in southern China in 2002 and the current outbreak of COVID-19 had peaked in mid-February due to exposure to live animals sold in markets. Furthermore, similar to the SARS outbreak, this epidemic has occurred during the Spring Festival in China, as the most famous traditional countrywide festival in China, gathering nearly three billion people from different areas. These favorable conditions caused the wide transmission of this fatal pneumonia and severe difficulties for disease control and prevention of the epidemic [92].Based on clinical data of diagnosed patients during the SARS-CoV-2 outbreak, the basic reproduction number (R0) is estimated to range between 2 and 6.47 in various modeling studies [76,93]. The SARS-CoV-2 R0 is in line with the one estimated for SARS-CoVs and MERS-CoVs (from 2 to 5) [94,95]. Currently, increasing countries are experiencing clusters of cases and community transmission following SARS-CoV-2 pandemic. Since its emergence, the COVID-19 has drawn well-deserved attention from authorities in order to protect their community and stop or slow down transmission of this disease. At the time of this review, according to the daily report of the World Health Organization, SARS-CoV-2 has affected over 17,889,134 people with around 228,611 daily new cases and killed more than 686,145 people all over the world, by August 3rd, 2020 (the up to date fatality rate is reported from https://covid19.who.int). We must take into consideration that these data are relative to laboratories and clinically confirmed cases while the actual number including asymptomatic cases, infected undiagnosed and death patients would be much higher than reported cases.The transmission of seasonal respiratory coronaviruses can be affected by several climate parameters such as temperature and humidity [96]. Therefore, understanding the relationship between weather and transmission of COVID-19 is the key to forecast the intensity and end time of this pandemic. To this regard, emerging evidence suggests that whether climate conditions may influence the transmission of the SARS-CoV-2 by boosting the spread (much of the data have not been peer-reviewed yet). To date, COVID-19 has had a significant expansion in the Northern Hemisphere (NH) belt, given that it covers cities and populated areas; conversely, in the belt of the Southern Hemisphere (SH), which covers very low population and landless areas, COVID-19 has not been reported yet.Based on climate and ERA-interim reanalysis dataset in NH belt from November 2019 to March 2020, we compared the average rate of humidity and temperature between five cities in European Countries with significant community transmission of COVID-19 versus five cities of North Africa which are expected to be less exposed to COVID-19. The information recorded by the meteorological stations has been used, since these are more accurate than satellite data [97]. As shown in Table 3, the average amount of humidity is very close between European and African selected sites. The main reason is the proximity of these cities to the Mediterranean Sea coastline. In addition, the north wind, which blows from northern Europe to European cities (ECs), increases the humidity of these cities. Conversely, there are temperature differences between considered ECs and North African cities (Table 3). Thus, temperature and humidity should be considered parameters involved in the transmission of COVID-19. Up to know, few studies have investigated the association of temperature and humidity with COVID-19 incidence and death rates. The first meteorological study was done in 100 different Chinese cities each having more than 40 cases of COVID-19 in a 3-day period during the end of January [98]. This group showed that high temperature and humidity significantly reduces the transmission of COVID-19. Their results indicate that the increases of 1 °C in temperature and 1% in relative humidity lower R by 0.0225 and 0.0158, respectively [98]. A preprint study on confirmed COVID-19 cases collected from 429 cities showed that every 1 °C increase in the minimum temperature of higher-temperature cities reduced the disease incidence and death rates by 0.86 [99]. Another preprint study suggested that the average increase of 1 °C in temperature correlates negatively with the predicted number of cases worldwide [100]. These results are in accordance with Wu Y et al. who showed that among all confirmed COVID-19 new cases and new deaths from 166 countries (excluding China), a 1 °C increase in temperature is associated with a 3.08% reduction in daily new cases and a 1.19% reduction in daily new deaths, whereas a 1% increase in relative humidity was associated with a 0.85% reduction in daily new cases and a 0.51% reduction in daily new deaths [101]. A recent study conducted in Italy showed a positive correlation of SARS-CoV-2 spreading and weather conditions including temperatures ranging 4–12 °C and relative humidity of 60–80% [102]. In a geographic and population modeling study conducted in five largest cities in Colombia, the transmission of SARS-CoV-2 seems to be comodulated by temperature and humidity. Their observation revealed a strong reduction of transmission in climates with temperatures above 30 °C and relative humidity below 78% which may comodulate the infectivity of SARS-CoV-2 within respiratory droplets [103].Overall, these meteorological analyses support that the combination of temperature and humidity could represent a direct influence on the transmission of the COVID-19. It can be assumed that the arrival of summer and rainy season in the NH can effectively reduce the transmission of the COVID-19. The distribution of COVID-19 across different longitudes and latitudes with a range of temperatures and humidity may help to predict the prevalence of this disease in terms of environmental characteristics. This could lead to a better understanding of how the virus spreads around the world (Figure 5). It should be noted that apart from the capability of SARS-CoV-2 to persist on environmental surfaces under favorable atmospheric conditions, the duration of its persistence may be affected by temperature and humidity. However, caution is needed when considering the implications of these findings, which may be subject to confounding. Although warmer climates may slow the spread of SARS-CoV-2, relying on weather changes alone to slow the transmission of COVID-19 are unlikely to be enough. However, using this type of dataset and climate analysis modeling is possible to identify areas that are most likely to be at risk of significant COVID-19 cases in the future and serve as an alarm signal to various government departments and agencies to adopt the necessary measures to prevent virus spread [104]. Moreover, more data are gathering around the world due to the change of the season and all authors agree that the association between temperature/humidity and SARS-CoV-2 is an appreciable hypothesis, but not a certainty yet.The most common symptoms of patients at onset of COVID-19 disease are defined as fever, dry cough, fatigue and less often, symptoms of sputum production, headache, sore throat, myalgia; hemoptysis, dyspnea, diarrhea and lymphopenia were also observed [15,24,28] (Figure 6). The spectrum of clinical features of COVID-19 has been found ranging from an asymptomatic state to severe respiratory failure and multiorgan dysfunction [24,76]. Symptomatic people are considered to be more contagious, similar to most viral-related respiratory diseases. However, individuals who remain asymptomatic may also transmit the virus and cases infected by an asymptomatic individual in the prodrome period of COVID-19 have also been reported [76]. Asymptomatic infections can occur because of weakened immune responses and subclinical manifestations or also because the virus is waiting for opportunities to invade and reproduce. A recent study has shown that a viral load detected in an asymptomatic patient was just like to the one observed in symptomatic patients, indicating the capability of transmission in asymptomatic patients [74]. According to disease presentation, COVID-19 can be classified as mild, moderate, severe and critical (Table 4) [57,76,105].The symptoms of SARS-CoV-2 infection appear after an incubation period of 1 to 14 days, similar to those of SARS- and MERS-CoV infections (median approximately 5.2 days in different studies) and 95% of patients are likely to experience symptoms within 12.5 days from contact [24,106,107,108]. However, in an asymptomatic carrier the incubation period was 19 days, complicating the challenge to contain the outbreak [109]. The median time between onset of symptoms and dyspnea is 5 days, 7 days for hospitalization and 8 days for acute respiratory distress syndrome (ARDS) (Figure 7) [24] (https://www.epicentro.iss.it/coronavirus/sars-cov-2). Patients at this stage in intensive care unit (ICU) with quarantine facilities may require mechanical ventilation. Moreover, bacterial infections can cause a secondary pneumonia [108]. In addition, the period from the beginning of COVID-19 symptoms to death varied between 6 and 41 days with an average of 14 days [110]. This period depends on immune system status and the patient’s age, being shorter in 70-year-old subjects compared with those younger [78,110]. In people with compromised immune systems and in elderly patients with underlying health problems, SARS-CoV-2 is able to infect the lower respiratory tract leading to severe pneumonia [111]. In 25–30% of patients presenting acute lung injury, shock, ARDS and acute kidney injury, ICU admission was absolutely required [24]. Recovery started within the 2nd or 3rd weeks with the median duration of hospitalization of 10 days. The virus appears to be more fatal in individuals with underlying co-morbidities (50–75% of fatal cases) [24,111]. Available dataset was obtained from Italian Istituto Superiore di Sanità (ISS) on 34,026 patients dying in-hospitals (Figure 7). The mean number of diseases was 3.3 (median 3, SD 1.9). Overall, 4.0% of the reported cases has no co-morbidities, 14.0% with a single comorbidity, 20.6% with 2 and 61.4% with 3 or more (https://www.epicentro.iss.it/coronavirus/sars-cov-2).SARS-CoV-2 infections revealed some unique clinical characteristics that include targeting the lower airway which is evident by symptoms of upper respiratory tract including rhinorrhoea, sneezing and sore throat [78]. Chest computed tomography (CT) scans revealed pneumonia in most SARS-CoV-2 infected patients and several cases showed an infiltrate in the upper lung lobe, which is related to increasing dyspnea with hypoxemia [28,78,112]. Table 4 describes the full picture of COVID-19 clinical manifestation. Atypical symptoms include RNAemia, acute cardiac injury, ARDS and grand-glass opacities that lead to death [113]. It should be noted that COVID-19 manifestations such as fever, dyspnea, dry cough and bilateral ground-glass opacities in chest CT scans are similar to the previous Betacoronavirus-related diseases [113,114]. Although gastrointestinal symptoms such as diarrhea were reported in SARS-CoV-2 infected patients, the similar gastrointestinal distress occurred in only a small percentage of MERS-CoV or SARS-CoV patients (Figure 6) [78]. It was shown that severe cases were characterized by an increased inflammation due to both systemic and localized immune response activation [78,115]. Higher leukocyte numbers, significantly high blood concentrations of cytokines and chemokines were noted in these cases [28,78]. Hence, it is now accepted that high levels of proinflammatory cytokines could worsen the prognosis [28,113].COVID-19 clinical evaluation is focused mainly on epidemiological data, clinical symptoms and clinical and laboratory tests. Although the scenario is continually changing, several approaches were selected as standard laboratory methods for COVID-19 diagnosis. Lab tests, differently from clinical-based analyses, immediately reveal SARS-CoV-2 infected patients. This was particularly important for diagnosis due to the difficulties in detecting specific clinical signs and symptoms in COVID-19 patients. Moreover, atypical manifestations revealed by pulmonary imaging [116] and the huge number of different clinical signs and symptoms forced the development of molecular-based laboratory tests [117,118]. Lastly, the analysis of personal history of each patient played a fundamental role in COVID-19 diagnosis and up to now is considered one of the key information for detecting infected patients also in the early phases of infection. Therefore, the epidemiological history together with clinical and laboratory tests are all required for the diagnosis of COVID-19. A detailed description focused on clinical diagnostic methods was reviewed by Taisheng Li [119]. Herein, we present an updated overview of the principal techniques used for COVID-19 diagnosis.High-throughput sequencing and real time quantitative polymerase chain reaction (RT-qPCR) are the best nucleic acid detection techniques for SARS-CoV-2. However, in clinical diagnosis, the application of high-throughput sequencing technology is limited due to high cost and its equipment dependency [114,120,121]. Moreover, to speed up the development of standardized analytic kits for diagnostic application, the quantification of viral load was not considered. Therefore, the RT-PCR method was chosen as the gold standard for the detection of SARS-CoV-2 infections from the commonly used samples such as naso- and oropharyngeal swabs [106,107,121,122]. This molecular method relies on the amplification of up to three SARS-CoV-2 specific targets including the RNA-dependent RNA polymerase (RdRp), E and N genes [121]. The WHO has released numerous RT-PCR protocols for the detection of SARS-CoV-2 RNA at https://www.who.int/emergencies/diseases/novel-coronavirus-2019/technical-guidance/laboratory-guidance (Accessed March 15, 2020). Three of those protocols are mentioned below. The US centers for disease control and prevention (CDC) developed an RT-PCR that includes three sets of oligonucleotide primers and probes recognizing three regions of the virus N gene (named N1, N2 and N3) and an additional primer/probe set to detect the human RNase P gene (RP) representing an internal control for RNA extraction and retro-transcription. Moreover, the positive control consisting in retro-transcribed viral RNA is also available at CDC. To report the positivity for SARS-CoV-2 two out of three N regions must be positive. The Chinese Center for Disease Control and Prevention (China CDC) recommends the use of specific primers and probes targeting the ORF1ab and N gene regions for SARS-CoV-2 detection by RT-PCR [123]. The positivity is confirmed when both targets are detected. Available online: http://ivdc.chinacdc.cn/kyjz/202001/t20200121_211337.html (accessed on 21 January 2020).Overall, the WHO summarized all the primer pairs and probes that can be used to detect SARS-CoV-2 in clinical specimens with the description of RT-PCR settings and the specificity. Apart from the possibility to perform the RT-PCR in house using the selected primer pairs and probes, several ready to use kits were developed for RT-PCR performing. One of the most used is the Allplex 2019-nCoV (Seegene, Seoul, South Korea) which includes three different viral targets and a positive control [124]. Another example is represented by the BGI’s real-time fluorescent RT-PCR Kit for detecting SARS-CoV-2 that includes one SARS-CoV-2 specific target and an internal control of the reaction (BGI, Cambridge, MA, USA). Both companies declared a sensibility of 100–150 viral copies per mL and a high specificity that excludes most respiratory tract viral and bacterial pathogens. The recommended samples for both in-house and ready-to-use RT-PCR kits include upper and lower respiratory specimens such as throat, nasal nasopharyngeal (NP) and/or oropharyngeal (OP) swabs, lower respiratory tract aspirates, sputum, bronchoalveolar lavage (BAL) fluid and nasopharyngeal wash/aspirate or nasal aspirate. It was observed that samples of the lower respiratory tract provide the higher viral loads [74]. On the other hand, it was shown that in the early stage of infection, the positive rate of RT-PCR was reported to be about 60% for throat swab samples [125]. Indeed, although being the gold standard, the RT-PCR presents some drawbacks. One of the most important is related to the sensibility because it was extensively reported that in the presence of low viral load this technique fails in detecting viral genome leading to false negative results [126]. Due to this problem, clinical governance as well as kit troubleshooting indicate to retest all the samples showing only single positive target along with patient resampling. To this respect, it should be underlined that operator skills or sampling sources can profoundly affect RT-PCR testing results [22]. Finally, during this pandemic several microbiologic labs worldwide are experiencing scarce availability of RNA extraction as well as ready-to-use RT-PCR kits increasing the timing of diagnosis confirmation through molecular approaches. Very recently, it was reported that the Allplex 2019-nCoV and the RealStar SARS-CoV-2 RT-PCR kits can amplify the target genes bypassing the RNA extraction step for a faster diagnosis [127].Although RT-PCR is specific for the diagnosis of COVID-19, its false-negative rate cannot be overlooked due to the severe consequences of missed diagnosis. Clinicians have demonstrated the usefulness of CT and chest radiography for the diagnosis of COVID-19 associated pneumonia [128]. Moreover, the ability of radiologists to diagnose COVID-19 pneumonia from chest CT evaluations has been reported to be very high [129]. Then a combination between RT-PCR and CT imaging represents the best approach for the correct COVID-19 diagnosis. In particular, for early detection and assessment of disease severity, the high-resolution CT (HRCT) of the chest is considered necessary [130,131]. One study analyzed the consistency and diagnostic value of RT-PCR test compared with chest CT in 1014 patients with suspected SARS-CoV-2 infection. Findings indicated that the chest CT sensitivity in suspected patients was 75% based on negative RT-PCR results and 97% based on positive RT-PCR results [132]. Moreover, Salehi et al. confirmed the higher sensibility of pulmonary imaging with respect to RT-PCR for COVID-19 diagnosis and showed a positive correlation between specific CT findings with the different stages of the disease and its severity [116]. The collection of numerous CT images has opened the possibility to build a database of pulmonary images from COVID-19 patients. Interestingly, the recent progress in integrating artificial intelligence (AI) with computer-aided design (CAD) software for diagnostic imaging revealed that AI could be used to support disease diagnosis [133,134]. Ito et al. reviewed the literature on the use of AI for lung diagnostic imaging of COVID-19 patients. Among the 15 selected studies, 11 used AI for CT and 4 used AI for chest radiography. The number of datasets ranged from 106 to 5941, with sensitivities ranging from 67–100% and specificities ranging from 81–100% for prediction of COVID-19 pneumonia. This study revealed the usefulness of AI approach to support the diagnosis of COVID-19, but also for future emerging diseases [134]. All the collected knowledge on lung lesions revealed some characteristic CT findings of COVID-19 pneumonia: the pulmonary ground-glass opacities in a peripheral distribution and the consolidation referring to an increase in pulmonary parenchymal density [135,136,137]. However, chest CT manifestations can vary in different patients and stages of infection, highlighting certain shortcomings of this approach. Apart from atypical manifestation that cannot be recognized by radiologists, several lung images are common in viral pneumonia leading to misdiagnosis [138]. Soon after the beginning of SARS-CoV-2 spreading, infected patients underwent antibody research for both basic research and clinical applications. One of the first studies reported the seroconversion of 100% of infected patients (n = 285) within 19 days after symptom onset. Seroconversion for IgM and IgG occurred simultaneously or sequentially and both immunoglobulins titers plateaued within 6 days after seroconversion. Importantly, the application of serology testing in surveillance in a cluster of 164 close contacts of COVID-19 patients identified 4.6% of positive patients showing negative RT-PCR results [139]. Hence, several studies underlined the recommended usage of serology to promote the detection of SARS-CoV-2 infections where NP swab specimens were improperly collected, molecular assays were unsatisfactorily carried out and for determining asymptomatic infections [122]. Based on these data, several companies developed kits for IgM/IgG testing showing a high detection rate of infected patients. Basically, there are two different testing methods: the rapid IgG-IgM test and the classical enzyme-linked immunosorbent assay (ELISA)-based test. The rapid test consists in a lateral flow qualitative immunoassay on a strip to detect the presence of both anti-SARS-CoV-2-IgM and anti-SARS-CoV-2-IgG in human specimens such as whole blood, serum and plasma. This IgG-and IgM-combined antibody test kit has a sensitivity of 88.66% and specificity of 90.63%. Results are obtained in 15 min leading to its useful application as point-of-care testing and in supporting RT-PCR-based diagnostic [140]. On the other hand, several ELISA-based kits are now commercially available, and their sensitivity and specificity were compared showing an overall high specificity, but a variable sensibility [141]. Differently from the rapid tests, the ELISA-based test should be performed on serum or plasma samples collected from venous sampling. Interestingly, the authors showed the neutralizing capacity of SARS-CoV-2 specific antibodies on Caco-2 cells directly incubating the sera from patients with the cell monolayers [141]. This assay is extremely important for the plasma-based therapies that are successfully used to treat seriously ill patients (see below). Finally, recently published papers described the seroconversion of COVID-19 patients including the evaluation of IgA that seems high in the early stages of infection (about 4 days’ post symptom development) [142,143]. Another interesting application of antibody detection is represented by the fluorescence immuno-chromatographic assay for the detection of SARS-CoV-2 nucleocapsid protein in human specimens such as NP swab [144]. It shows the fastness of rapid tests (results in 10 min), the possibility to use the same type of sample that is commonly used for RT-PCR-based diagnosis and high sensibility (detection of the nucleoprotein in all positive samples tested). Although these methods were suggested for COVID-19 diagnosis, the extent of antibodies production by infected patients is greatly variable. Moreover, the delay of antibodies production with respect to the onset of symptoms affects the use of this approach for diagnosis. Vice versa, it is reported that several governments, included Italy, are using serologic test for population screening to assess the proportion of people that have developed an immunological response against SARS-CoV-2 (http://www.salute.gov.it/portale/nuovocoronavirus). This screening will help also to detect asymptomatic and/or paucisymptomatic subjects.The rapid spread of SARS-CoV-2 raises an urgent requirement for effective therapeutic strategies against COVID-19. Although many efforts have been intended to develop vaccines against HCoVs infections in recent years, there is no official and effective treatment against SARS-CoV-2. However, different considerable options have been applied for possible vaccine validity, efficacy and safety along with speeding up other ongoing searches to discover valuable modalities for dealing with the emerging COVID-19 [12,145,146,147,148].Most of the drugs that are being used to cope with COVID-19 epidemic are directed towards specific viral molecular targets and biologic processes through which the virus spreads damaging the host. In line, all available experimental therapies for COVID-19 management are based on previous experiences in treating SARS-CoV and MERS-CoV infections, such as inhibitors of SARS-CoV-2 fusion/entry/replication, anti-viral agents against main viral proteases, regulators of SARS-CoV-2 induced host inflammatory response and direct administration of human monoclonal antibodies (mAbs) (Figure 8) [149]. Apart from all these possible therapeutic approaches, it has been reported that the Chinese medicine products, as Lianhuaqingwen and ShuFeng JieDu capsules may be helpful for SARS-CoV-2 treatment [12,150]. Indeed, this product is mainly used to treat upper respiratory tract infections such as the flu, swelling and pain in the throat, mumps and strep throat [151,152]. Moreover, four COVID-19 cases have been described to gain improvement after taking combined Chinese and Western medicine [153]. Notably, encouraging progress in deciphering SARS-CoV-2 genome will lead to new potential therapeutic targets. Likewise, more prospective, rigorous population studies are urgently required to confirm the therapeutic effect as well as the safety of new potential therapeutic strategies in order to further implement robust preventive and control measures against SARS-CoV-2 spread.As outlined above, multiple strategies are aimed at developing CoVs vaccines, most of which are headed for the surface-exposed spike (S protein) glycoprotein as the major virus–host cell membrane interactor. To this aim, vaccines under study are based on full-length S protein, S1-RBD, expression of virus-like particles (VLP), DNA or viral vectors [42,154,155,156,157,158]. As outlined above, the S1 includes the RBD that interacts with its host cell receptor, ACE2, whereas the S2 mediates fusion between the virus and host cell membranes promoting the entry and subsequent replication of the viral RNA into the cytoplasm [158]. The ACE2 receptor, as a specific biologic target for vaccine development, is under study in a controlled pilot clinical trial to investigate the effect of recombinant human ACE2 (rhACE2; GSK2586881) in patients with severe COVID-19 (NCT04287686) (Figure 8I) [159,160]. Vice versa, both recombinant proteins containing RBD and the recombinant vectors encoding RBD can be used to generate the effective SARS-CoV vaccines given the capability of this domain to induce neutralizing antibody [156]. Indeed, the first available SARS-CoV-specific human monoclonal antibody with neutralizing activity against SARS-CoV, named CR3022, was found to bind potently to SARS-CoV-2 RBD, in agreement with the high homology shared by this domain with SARS-CoV homolog [161]. However, it must be taken into account that more than 85% of the RBD antibody epitopes in SARS-CoV-2 show implicit noticeable changes, indicating the necessity to develop more specific monoclonal antibodies for SARS-CoV-2 [162].Angiotensin receptor blockers (ARBs), such as losartan, valsartan, telmisartan, usually assumed for treating high blood pressure, heart and kidney failure in people with diabetes, have been recently proposed as a novel therapeutic approach to block SARS-CoV-2 RBD binding to ACE2- expressing cells binding, similarly to ACE inhibitors [163].Additional targetable epitopes that should be considered are the heptad repeat 1 (HR1) and heptad repeat 2 (HR2) in SARS-CoV-2 S protein. In fact, the HR2-derived peptides (HR2P) and EK1 (a modified OC43-HR2P peptide), exhibit effective fusion inhibitory activity towards SARS-CoV-2, suggesting a promising strategy in treating SARS-CoV-2 infection, although further studies are required to strengthen these hypotheses (Figure 8I) [164,165].Lately, immuno-informatics have been employed to identify significant cytotoxic T lymphocyte (CTL) and B-cell epitopes in SARS-CoV-2 S protein, such as the nucleocapsid (N) protein as well as the potential B cell epitopes of the E protein of MERS-CoV as likely immunoprotective targets [166,167].Reverse genetic strategies have been successfully used in live-attenuated vaccines to inactivate the exonuclease effects of non-structural protein 14 (nsp14) or to wipe out the envelope protein in SARS [154]. A recent study also revealed that the invasion process requires the priming of the S protein which is facilitated by the host cell produced serine protease TMPRSS211. The clinically demonstrated serine protease TMPRSS2 inhibitor Camostat mesylate, which partially blocks SARS-CoV-2 entry into host cells, was shown to be a good target to significantly reduce pulmonary infection in COVID-19 affected individuals (Figure 8I) [168] Moreover, it has been suggested that coronavirus entry also involves pH and receptor-dependent endocytosis [169,170]; thus, targeting endocytosis may be another assessable option for fighting SARS-CoV-2 (Figure 8I). In this view, throughout AI technology, a group of approved drugs, such as the Janus kinase (JAK) inhibitor baricitinib [171] targeting the AP-2-associated protein kinase 1 (AAK1) regulating clathrin-mediated endocytosis, has been developed (Figure 8I) [172]. Furthermore, other drugs such as arbidol (ChiCTR2000029621), a haemagglutinin inhibitor and chloroquine phosphate, a traditional antimalarial drug, have been added to the National Health Commission of the People’s Republic of China (NHC) guidelines for COVID-19 treatment (Figure 8I) (http://www.nhc.gov.cn). In particular, in vitro studies have demonstrated that chloroquine as well as hydroxychloroquine could impair the endosome-mediated viral entry or later stages of viral replication [173]. Combination of hydroxychloroquine and azithromycin has also been suggested as a valid approach since it showed more rapid resolution of infection than hydroxychloroquine alone [174]; however, the combined use of azithromycin and hydroxychloroquine seems to be associated with at increased risk of arrhythmias. Available online: https://www.acc.org/latest-in-cardiology/articles/2020/03/27/14/00/ventricular-arrhythmia-risk-due-to-hydroxychloroquine-azithromycin-treatment-for-covid-19 (accessed on 29 March 2020).To date, several attempts have also been made in targeting viral main enzymes; in fact, many inhibitory drugs targeting the coronavirus main proteinase 3C-like protease (3CLpro) have been validated in clinical trials (e.g., Lopinavir/Ritonavir; ChiCTR2000029387, ChiCTR2000029468, ChiCTR2000029539) (Figure 8II) [175]. Moreover, four additional molecules including prulifloxacin, tegobuvir, bictegravir and nelfinavir, detected by high-throughput screening, showed reasonable binding conformations with the viral main protease [176]. Moreover, a recent study by performing a virtual screening using a three-dimensional model of the SARS-CoV-2 3C-like protease (3CL), identified 16 biologic candidates that deserve further consideration. Among these, the antivirals Ledipasvir or Velpatasvir proved to be particularly attracting as therapeutics to combat the new coronavirus showing optimal anti-viral activity and minimal side effects, such as fatigue and headache; also, Epclusa (velpatasvir/sofosbuvir) and Harvoni (ledipasvir/sofosbuvir) are promising antivirals, not only for their effective and synergic inhibitory activities against two viral enzymes, but also for their minimized possibilities to develop resistance [177].A certain number of clinical trials on antiviral drugs aimed to arrest SARS-CoV-2 replication are currently in progress, such as Remdesivir (NCT04252664, NCT04257656) Favipiravir (ChiCTR2000029600, ChiCTR2000029544) and ASC09 (ChiCTR2000029603) (Figure 8III). Among these, Remdesivir was recently approved for medical use in America and European Union and seems to be the most promising antiviral for fighting SARS-CoV-2 [178] (http://www.who.int), as in vitro studies demonstrated that this molecule, a mono-phosphoramidate prodrug of an adenosine, effectively inhibited SARS-CoV-2 RNA synthesis [179]. Targeting the SARS-CoV-2 RNA genome could, therefore, be another potential strategy. In fact, a CRISPR/Cas13d technology, which is an RNA-guided RNA-targeting CRISPR system, has been employed to specifically chew up SARS-CoV-2 RNA genome. In this system, a Cas13d protein and guide RNAs-containing spacer sequences are used to specifically complement the virus RNA genome (Figure 8IV). Furthermore, RNA genome can be packaged into one adeno-associated virus (AAV) vector, making the CRISPR/Cas13d system more efficient for virus elimination and resistance prevention, taking into account that AAV has serotypes highly specific to the lung, the main organ infected by SARS-CoV-2 [180].In addition to antiviral therapy, a new treatment strategy having a significant impact on clinical outcomes is utmost required. Immunomodulatory therapy to downregulate the cytokine storm may provide great benefit to the treatment of COVID-19. Recently, researchers focused on targeting specific molecular markers involved in inflammatory cytokines-receptors interactions, their correlation in health and disease and drugs in use that can activate or block their actions. A higher concentration of cytokines has been found in the plasma from COVID-19 patients in ICU compared with the ones from non-ICU COVID-19 patients, suggesting that the cytokine storm could be linked to the severity of the disease [113]. Corticosteroids are among the most commonly used drugs for immunomodulatory therapy of infectious diseases. However, the use of corticosteroids in the treatment of COVID-19 can cause host immune suppression and delay of viral clearance. A recent study on 201 patients with ARDS showed that treatment with methylprednisolone decreased the risk of death (hazard ratio 0.38, 95% confidence interval 0.20–0.72). These findings indicate that using corticosteroids does not influence viral clearance time, length of hospital stays or duration of symptoms in patients with mild COVID-19 [181]. Thus, the use of corticosteroids is considered beneficial in severe cases of COVID-19 (especially in patients with ARDS), but not in mild cases. Accordingly, a recent retrospective study showed the potential benefits from low-dose corticosteroids treatment in a subset of critically SARS-CoV-2 patients [182]; these data are in contrast with NHC guidelines that highlight that systematic use of corticosteroids is not recommended for these cases, due to their immunosuppressive effects. However, administration of corticosteroids has been indicated for specific reasons such as exacerbation of asthma or chronic obstructive pulmonary disease (COPD), septic shock or severe acute respiratory distress syndrome (ARDS). Further studies are required to find out how and when it is appropriate the use of corticosteroids for COVID-19, as there are no available data on the benefits of corticosteroid treatment in SARS-CoV or MERS infection [183].Apart from corticosteroids, IL-6 pathway inhibitors such as sarilumab, siltuximab and tocilizumab have been proposed as experimental approach considering the increased IL-6 levels that have been observed in patients with severe COVID-19 [184]. Tocilizumab is a recombinant, humanized monoclonal antibody commonly used for treating patients with rheumatoid arthritis, lupus and psoriasis that binds to IL-6 receptors blocking FcR activation; in COVID-19 patients, Tocilizumab could reduce SARS-CoV-2-induced inflammatory responses [185]. Accordingly, several case reports have referred positive outcomes regarding Tocilizumab [113,186,187,188,189,190], but clinical impact of Tocilizumab on COVID-19 patients as an approved clinical approach has not been evaluated yet. In line, to further investigate the efficacy and safety of Tocilizumab in patients with COVID-19, a controlled clinical trial is now under way (ChiCTR2000029765) (Figure 8V). Overall, the combination of an immunomodulatory agent to reduce the cytokine storm with an antiviral agent may give physicians more time to provide supportive treatment for patients with COVID-19.At the time of writing this review, due to the lack of a specific available therapy, plasma from convalescent patients containing specific antibodies has been proposed as a principal treatment [190,191], for patients in rapid disease progression, severe or critical conditions (Figure 8VI). In a recent retrospective study, one dose (200 mL) of convalescent plasma (CP) collected from 10 severe adult cases has been reported to be tolerated; thus, increasing or maintaining high level of neutralizing antibodies broke down the viral load in seven days, improve clinical symptoms and paraclinical criteria within three days and lung lesions were found to be differently absorbed on radiological examination within seven days [192]. Therefore, being CP a promising rescue option for severe COVID-19, several clinical trials (ChiCTR2000030010, ChiCTR2000030179, and ChiCTR2000030381) are in progress to investigate the efficacy and safeness of CP direct infusion in COVID-19 patients [191]. In addition, combined therapy with mAbs and Remdesivir seems to be an ideal therapeutic option for COVID-19 [193]. Pharmaceuticals companies are now focused on searching for specific and effective mAbs against COVID-19. Taking into account that technologies capable of making fully human antibodies such as human single-chain antibody variable fragments (Hu-scFvs) or humanized-nanobodies (single-domain antibodies, sdAb) able to overpass virus-infected cell membranes (trans bodies) and to interact or interpose with biologic processes required for virus replication are already available [194].A large number of clinical trials regarding cell-based therapies have been started in China during COVID-19 outbreak. Among these, mesenchymal stromal cells (MSCs)-based therapy displayed strong safety profile and possible efficacy in patients with ARDS, according to COVID-19-related clinical studies listed on the WHO’s International Clinical Trials Registry Platform (WHO ICTRP) and National Institutes of Health’s clinical trials.gov databases [195]. Nevertheless, further investigations are required to better understand if these therapies could be effective in treating respiratory virus-induced complications. MSCs have been largely employed in basic research and clinical trials [196,197,198], and their safeness and effectiveness have been extensively documented especially in immune-mediated inflammatory disorders, such as graft-versus-host disease (GVHD) [199] and systemic lupus erythematosus (SLE) [200]. MSCs immunomodulatory and differentiation abilities [201] as well as their competency to produce several cytokine types or to directly interact with immune cells have been already described [202]. Indeed, they are activated by pathogen-associated molecules (PAMPs) such as single or double-stranded RNAs [203,204], priming the immune response during infections. Two clinical investigations of systemic MSC administration in patients with either COVID-19 or avian influenza A (H7N9) have been recently published [205,206]. The first one, a single-center MSC transplantation pilot study, was aimed at exploring MSCs therapeutic potentiality in patients with COVID-19 pneumonia and conducted at the You’an Hospital in Beijing, China, from 23 Jan 2020 to 16 Feb 2020 (ChiCTR2000029990). Seven patients with COVID-19 pneumonia, SARS-CoV-2 RNA positive, with different degrees of severity, including one critically ill requiring ICU care were enrolled and monitored for 14 days after MSC injection. A significant improvement of pulmonary function and symptoms were observed two days after MSC transplantation characterized by an increase of peripheral lymphocytes and of the anti-inflammatory IL-10 levels and a decrease of the C-reactive protein and TNF-α amounts [205]. Moreover, an increment of the CD14 + CD11c + CD11bmid regulatory dendritic cell (DC) population and a decrease of cytokine-secreting immune cells such as CXCR3 + CD4 + T cells, CXCR3 + CD8 + T cells, and CXCR3 + NK were detected within 3–6 days in the treated patients compared to the placebo control group [205]. MSCs play a role in attenuating cytokine storm, most importantly, because these cells do not express ACE2 and TMPRSS2 viral receptors are insusceptible of SARS-CoV-2 infection. These observations are in agreement with the knowledge that MSCs induce the maturation of dendritic cells into a novel Jagged-2-dependent regulatory dendritic cell population [207], shifting the Th1/Th2 balance towards Th2. Thus, from these preliminary results, it seems evident that MSCs intravenous transplantation could represent a secure and effective treatment in patients with COVID-19 pneumonia, especially those critical. Indeed, it inhibits the over activation of the immune system and promotes endogenous repair by preventing pulmonary fibrosis and improving both pulmonary microenvironment and lung function [205].More than 15 potential vaccine candidates for COVID-19 are under development around the world, including inactivated, recombinant subunits, nucleic-acid-based, adenoviral vector, and recombinant influenza viral vector vaccines [208]. Moreover, taking into consideration the strong homologies existing among the various coronavirus strains, it was thought that vaccines acting on other coronaviruses, such the avian live IBV vaccine (strain H) directed towards the chicken CoV IBV, could be a valuable alternative therapeutic strategy [209].The Coalition for Epidemic Preparedness Innovations (CEPI) recently announced that three programs aimed to develop COVID-19 vaccines, by utilizing established vaccine platforms, have started [210]. In addition, CEPI already financed the company Moderna, Inc. to compare mRNA therapeutics and vaccines, allowing the release of the first batch of mRNA-1273 in February 2020, which is an mRNA vaccine against SARS-CoV-2 ready for phase I study in the United States. Available online: https://investors.modernatx.com/news-releases/news-release-details/moderna-ships-mrna-vaccine-against-novel-coronavirus-mrna-1273 (accessed on 24 February 2020).More recently, scientists from the University of Pittsburgh have announced a potential vaccine against SARS-CoV-2, delivered throughout a fingertip-sized patch, capable of producing SARS-CoV-2 specific IgG antibodies, sufficient for virus neutralization in mice. This vaccine, called PittCoVacc (acronym of Pittsburgh coronavirus vaccine), is a trimeric recombinant SARS-CoV-2-S1 subunit vaccine delivered intracutaneously by microneedle arrays (MNAs) [211]. Delivering vaccine components to a defined skin microenvironment improves safety by reducing systemic exposure, allowing to reach high vaccine concentrations with a relatively low dose of antigen [212,213]. Furthermore, the skin delivery strategy promotes strong and long-lasting antigen-specific antibody responses due to both the high immunogenicity [214,215,216,217,218] and the redundant immunoregulatory circuits of the skin [217,219,220]. Given the urgent need for COVID-19 vaccines, MNAs strategy seems to be a promising immunization approach against coronavirus infection including SARS, MERS and other emerging infectious diseases.On April 24, the Oxford ChAdOx1 nCov-19 vaccine was the first in Europe to start human trial stage, with 1110 healthy volunteers enrolled for the tests. Oxford scientists have already employed ChAdOx1 in the past to dispense vaccines against Ebola, Chikungunya, Rift Valley fever and, above all, MERS. ChAdOx1, a chimpanzee-derived adenovirus vector, has been employed to deliver the full-length MERS spike gene and shown to induce large amounts of neutralizing antibodies against MERS in a mouse model [221,222]. Therefore, the modified ChAdOx1 vaccine, carrying the SARS-CoV-2 spike gene is under human trial stage. On April 30, the University of Oxford has announced a collaboration with the UK-based global biopharmaceutical company AstraZeneca for further development, large-scale production and potential delivery of the COVID-19 vaccine candidate. Available online: https://www.ovg.ox.ac.uk/news/landmark-partnership-announced-for-development-of-covid-19-vaccine (accessed on 30 April 2020). Since ChAdOx technology is already available and formerly tested in humans for other vaccines, Phase III will consist in administering vaccine to volunteers following them into their regular environments to ensure that these subjects actually become immune to the disease up to three years. If trials succeed, Oxford researchers have proposed to complete testing throughout ring vaccination, namely delivering vaccine to members of the first circle of contacts of COVID-19 positive people and then to evaluate if the virus spreads to the second circle, as was previously done during the 2018 Ebola epidemic in the Democratic Republic of the Congo.Overall, a joint effort headed to apply both already consolidate and innovative approaches, such as AI to facilitate drug discovery, will be required to develop a broad-spectrum antiviral drugs and vaccines towards existing and potential future coronavirus infections to prevent another highly pathogenic virus epidemic. Moreover, continuous collaboration in basic and clinical studies will improve the discovery of new antiviral drugs with therapeutic potentials, decrease the time for drug release on the market and make them affordable for all countries. Furthermore, vaccine delivery strategies and cell-based therapies benefit from the significant progresses made by recombinant DNA technologies combined with emerging biotechnology and bioengineering methodologies. Thus, these approaches can speed up the development and set up of new vaccines and clinical therapies to fight against novel pathogens to protect public health all over the world.This study represents a holistic picture of the current investigations in response to the outbreak of COVID-19. The current pandemic is obviously an international public health problem and it remains a challenging task to fight the SARS-CoV-2 of unknown origin and mysterious biologic features. Lesson from the previous two pandemics, MERS and SARS outbreaks, provide valuable insights about how to manage the current pandemic and provide a reference for future studies to combat disease progression. Despite SARS-CoV-2 rapid transmission, the scale up country readiness, speedy response teams and the capacity of all laboratories are reducing the spread of the virus as well as its mortality rate. As the pandemic is still ongoing and expanding, further studies on all aspects of the disease are needed to better understand the infection, beneficial treatments and development of vaccines. Nevertheless, this pandemic, together with the previous ones, have taught us in the harshest possible way that the entire scientific community must be vigilant and ready to advice appropriate containment and screening measures to avoid the spread of any future emerging pathogen.H.H. and M.S. conceived, planned the project and organized the manuscript; H.H., F.P.S., H.S., A.D.L. and M.S. performed a literature search and prepared the draft of the manuscript; H.H., F.P.S., H.S., A.D.L., L.M., R.R., D.S., C.A. and M.S. participated in data curation and writing—revising the manuscript. H.H., L.M., R.R., D.S., C.A. and M.S. discussed the conception of the review and contributed to the final manuscript preparation. H.H., D.S., C.A. and M.S. supervised the manuscript. All authors have read and agreed to the published version of the manuscript.This research received no external funding.We would like to acknowledge Majidi Nezhad for providing the meteorological data and climate analyses and Gaia Scoarughi and Adeleh Salehi for drawing figures.The authors declare no conflict of interest.The following abbreviations are used in this manuscript: 16 nonstructural polyproteins (nsps 1–16); 2019 novel coronavirus (2019-nCoV); 3-chymotrypsin-like protease (3CLpro); 3C-like protease (3CL); Acute respiratory distress syndrome (ARDS); adeno-associated virus (AAV); angiotensin receptor blockers (ARBs); angiotensin-converting enzyme 2 (ACE2); AP-2-associated protein kinase 1 (AAK1); artificial intelligence (AI); avian infectious bronchitis virus (IBV); basic reproduction number (R0); blood oxygen saturation (SpO2); bronchoalveolar lavage (BAL); centers for disease control and prevention (CDC); Chinese Center for Disease Control and Prevention (China CDC); chronic obstructive pulmonary disease (COPD); Coalition for Epidemic Preparedness Innovations (CEPI); computed tomography (CT); computer-aided design (CAD); convalescent plasma (CP); coronavirus (CoV); cytotoxic T lymphocyte (CTL); dendritic cell (DC); dipeptidyl peptidase 4 (DPP4, also known as CD26); envelope glycoprotein (E); enzyme-linked immunosorbent assay (ELISA); European Center for Medium-Range Weather Forecasts (ECMWF); European cities (ECs); feline infectious peritonitis (FIP); graft versus-host disease (GVHD); hemagglutinin-esterase glycoprotein (HE); heptad repeat 1 (HR1) and heptad repeat 2 (HR2); high-resolution CT (HRCT); HR2-derived peptides (HR2P); EK1 (a modified OC43-HR2P peptide); human coronaviruses (HCoVs); human CoVs (HCoVs); human single-chain antibody variable fragments (Hu-scFvs); humanized-nanobodies (single-domain antibodies, sdAb); intensive care unit (ICU); Istituto Superiore di Sanità (ISS); Janus kinase (JAK); membrane glycoprotein (M); mesenchymal stromal cells (MSCs); microneedle arrays (MNAs); Middle East Respiratory syndrome coronavirus (MERS-CoV); monoclonal antibodies (mAbs); naso-pharyngeal (NP); National Health Commission of the People’s Republic of China (NHC); nonstructural protein 14 (nsp14); Northern Hemisphere (NH); nucleocapsid phosphoprotein (N); open reading frames (ORFs); oro-pharyngeal (OP); papain-like protease (PLpro); partial pressure of arterial oxygen to fraction of inspired oxygen ratio (PaO2/FiO2); pathogen-associated molecules (PAMPs); Pittsburgh coronavirus vaccine (PittCoVacc); polypeptides (pps); porcine epidemic diarrhea virus (PEDV); real time quantitative polymerase chain reaction (RT-qPCR); receptor-binding domain (RBD); recombinant human ACE2 (rhACE2); replication–transcription complex (RTC); RNA-dependent RNA polymerase (RdRp); RNase P gene (RP); severe acute respiratory syndrome coronavirus (SARS-CoV); single-stranded positive-sense RNA (+ssRNA); Southern Hemisphere (SH); spike glycoprotein (S); subgenomic RNAs (sgRNAs); swine acute diarrhea syndrome coronavirus (SADS-CoV); swine enteric Alphacoronaviruses (SeACoVs); systemic lupus erythematosus (SLE); transmissible gastroenteritis virus (TGEV); virus-like particles (VLP); WHO’s International Clinical Trials Registry Platform (WHO ICTRP); World Health Organization (WHO).Classic subgroup clusters of coronaviruses within the family Coronaviridae, subfamily Orthocoronavirinae and the respective genera: Alphacoronavirus, Betacoronavirus, Gammacoronavirus and Deltacoronavirus.Origin and evolution of (A) SARS-CoV, (B) MERS-CoV and (C) SARS-CoV-2 in the various hosts. Initially all viruses existed in diverse bat species as CoV-related viruses (SARSr-CoV, MERSr-CoV and SARSr-CoV-2); sequential mutations and recombinations allow them to adapt to intermediate hosts and finally humans [15].Typical coronavirus virion structure and proteins. The coronavirus genome encodes a (S) spike glycoprotein, an (E) envelope glycoprotein, a (M) membrane glycoprotein, a (N) nucleocapsid phosphoprotein and a (HE) hemagglutinin-esterase glycoprotein.Graphic genome structures of SARS-CoV-2, SARS-CoV and MERS-CoV. Each coronavirus (CoV) genome is schematically represented in the order of 5′-ORF1a-ORF1b-S-E-M-N-3′. The coronavirus genomes encode two replicase polypeptides pp1a and pp1ab translated from ORF1a and ORF1b; four structural genes encoding for four structural proteins including (S) spike, (M) membrane, (E) envelope and (N) nucleocapsid proteins. The single-stranded RNA genomes of SARS-CoV-2 (~29.8 kb), SARS-CoV (~29.7 kb) and MERS-CoV (~30.1 kb) harbor two large genes, the ORF1a (red) and 1b (blue) genes encoding accessory genes (nsps 1–16, shades of red and blue). Encoded nonstructural proteins: 16 nsps (nsp1-nsp16) in SARS-CoV-2, SARS-CoV and MERS-CoV. Along with structural proteins (S, E, M and N), the 3′-terminus of the SARS-CoV-2 and SARS-CoV genomes contain eight accessory proteins (3a, 3b, p6, 7a, 7b, 8b, 9b and orf14 and 3a, 3b, p6, 7a, 7b, 8a, 8b and 9b, respectively) while MERS-CoV genome contains only five (3, 4a, 4b, 5 and 8b). The genes encoding accessory proteins are unique in different coronaviruses in terms of number, genomic organization, sequence and functions (data extracted from [35,49,57]).Monthly temperature (degree C) reanalysis maps using ECMWF dataset of all the world. The temperatures at 2-m height, obtained from ERA-interim datasets (https://climatereanalyzer.org/), have been processed to extract monthly means maps for the period November 2019 to February 2020. ERA-interim is a global reanalysis of recorded climate observations over the past 3.5 decades. It is presented as a gridded data set at approximately 0.7 degrees spatial resolution and 37 atmospheric levels. ERA-interim is produced by the European Center for Medium-Range Weather Forecasts (ECMWF) (https://climatereanalyzer.org/).Organ involvement confirmed by clinical features or bioptic sampling in COVID-19 patients (A). Table describing main observed disorders (B).Median times, in days, from the onset of symptoms to death, to hospitalization, from hospitalization to death with and without intensive care unit (ICU)-admittance (Report based on available data on July 9th, 2020 collected from Istituto Superiore di Sanità, ISS).Schematic representation of SARS-CoV-2 infection and virus-induced human immune system response. Proposed drugs directed both towards specific SARS-CoV-2 molecular targets and biologic processes are highlighted: inhibitors of SARS-CoV-2 fusion/entry targeting ACE2 receptor, spike protein, TMPRSS2 or HR1 and HR2 epitopes and clathrin-mediated endocytosis (I); molecules against SARS-CoV-2 main protease (II); molecules against viral genome replication (III); CRISPR technologies targeting SARS-CoV-2 RNA genome (IV); modulators of SARS-CoV-2 induced inflammatory response (V) and human neutralizing antibodies (VI). ACE2, angiotensin-converting enzyme 2; TMPRSS2, type 2 transmembrane serine proteases; RdRp, RNA-dependent RNA polymerase; HR1, heptad repeat 1; HR2, heptad repeat 2; HR2P, heptad repeat 2-derived peptides; EK1, a modified OC43-HR2P peptide. Adapted from [223].Host factors(s) involved in SARS-CoV, MERS-CoV and the SARS-CoV-2 replications [4,15].Comparison of main features among SARS-CoV, MERS-CoV and SARS-CoV-2 [24,76,86,88,93,94,95].Average humidity and temperature in 10 different cities in Europe and North Africa between November 2019 to March 2020. The first five cities represent significant communities where transmission of COVID-19 was reported, whereas the second 5 cities are expected to be less exposed to COVID-19 due to different weather conditions.Clinical manifestations of COVID-19.

Med-MDPI/ijerph_5/ijerph-17-16-05649.txt

@@ -0,0 +1 @@
+The aim of this study was to analyze the health and wellness status perception in amateur half-marathon runners according to sex, age, being injured or not during the two months prior to the race, and having the support or not of qualified staff for race preparation. Six hundred and twenty-four amateur level half-marathon athletes (515 men and 107 women; 41.5 ± 10.1 years) participated in the study. One week before competing in a half-marathon, participants answered the Hooper Index and the SF-36 questionnaire. Women stated higher stress before competing in the race (p < 0.01) compared to men and the group of runners of <40 years stated greater fatigue (p < 0.05) compared to the group of >40 years. Women showed a better quality of life in physical and emotional role dimensions (p < 0.05), and the group of >40 years showed a better quality of life in the emotional role dimension (p < 0.05). The group that had suffered an injury (InjuryYes) declared greater muscle soreness (MusclSore; p < 0.01), and the group that had qualified staff (QualifStaffYes) declared a higher level of stress (p < 0.05) and fatigue (p < 0.01). The Injury No (InjuryNo) group showed a better quality of life in the physical function dimension (p < 0.01). The group that did not have qualified staff (QualifStaffNo) showed a better quality of life in the dimensions of body pain, general health, vitality, social function (p < 0.05), and mental health (p < 0.01), while the QualifStaffYes group showed better results in the dimensions of physical function and emotional role (p < 0.05). Sex, age, being injured or not during the two months prior to the race, and having the support or not of qualified staff for the race preparation can influence the health and wellness status perception.Road running has evolved as an activity of increasing popularity [1], and almost every weekend races of this type take place worldwide. Specifically, the half-marathon is one of the most popular long-distance races as demonstrated by the 173 annual races homologated by the Royal Spanish Athletics Federation [2] and runners participating [3,4]. This type of races involves runners of different sex [5,6] and age [4] who are or are not injured [7], who train or do not train with qualified staff, who are highly trained and looking to improve their performance, or who are amateur runners (unprofessional) with a low level of training that simply aim to finish the race [4,8]. Despite the high participation of runners in this type of races [3] and that long-distance events are one of the most strenuous activities [9], the organizers of these events do not usually request any health check on runners as a requirement to participate [10]. It would be interesting to know, if the participants have adequate health perceptions to be able to make the efforts that are needed during the race, since if their perception is not good, runners, coaches, and event organizers could consider specific action strategies.It may be difficult to screen the health of a large group of participants due to the costs, availability, accessibility, and time to objectively measure health, especially in the moments before the race. There are some validated self-administered questionnaires for this goal such as the SF-36 [11] or the Hooper Index [12]. The SF-36 measures the perception of the health of the participants, and the index proposed by Hooper et al. [12] measures the well-being ratings of athletes relative to stress, fatigue, muscle soreness, and sleep quality. These questionnaires have been used in other sport modalities, such as soccer [13,14,15], basketball [16], handball [17] or trail running [18], but we have not found any studies that analyze the health or the wellness status perception in amateur (non-professional) half-marathon runners before the competition. It would be interesting to know the health and wellness status perception of amateur half-marathon runners because, in many cases, they do not monitor and control their training and competition, and to make the relevant physical effort during this type of competition, they may need to have an adequate initial health status.Therefore, in addition to ascertaining the perception of health and well-being in this group, it would be interesting to know if there are differences in the runners’ perception of health and wellness status according to their generic characteristics (sex, age (<40 and >40 years), whether they have been injured before or not, and whether they train with qualified coaches or not). This information could provide relevant data for coaches and runners about the perception of health and wellness status in half-marathon runners and the differences in these variables in the different population groups that take part in the races. Agrawal and D’Silva [19] obtained differences in the health and well-being perception according to sex; however, Batmyagmar et al. [20] did not find these differences based on sex in marathon runners. Valovich et al. [21] observed that the athletes who were injured declared lower values in the physical dimension compared to the non-injured athletes. Knowing whether athletes who have suffered an injury have a worse well-being and health perception or not could be interesting for studying the need to implement health-specific programs in injured athletes. Although there are some scientific studies in the literature, contradictory results have been found and, therefore, more studies may be necessary in this regard.The aim of the present study was to analyze the wellness (i.e., stress, fatigue, muscle soreness, and sleep quality) and health status perception (i.e., physical function, physical role, body pain, general health, vitality, social function, emotional role, and mental health) in amateur half-marathon runners according to sex, age, having been injured or not during the previous two months, and having the support or not of qualified staff for race preparation prior to the competition. The hypothesis of the study was that the wellness and the health status perception before competing in a half-marathon could be different depending on sex, age, having been injured or not during the previous two months and having the support or not of qualified staff.This study involved 624 amateur runners (41.5 ± 10.1 years) of whom 515 were men, 107 women, and two who were considered non-dichotomous and whose data were not considered for the analysis by sex. The runners were divided into two groups according to age: of all the runners participating in the study, 229 were under forty years old (<40 years; 30.7 ± 5.7 years) and 395 were over forty years old (>40 years; 47.7 ± 6.1 years). It was decided to take this cut-off age (<40 years old and >40 years old) into consideration according to the athletic racing regulations of the country where the study was conducted and the category classification used by the event organizers. One hundred and twelve participants had had an injury in the two months previous to participation in the race, and 512 had had no injury. One hundred and sixty-seven participants trained under the supervision of qualified staff and 457 did not. The inclusion criteria in the study were to have an official federative license, to be over 18 years old, and to have prepared to run in one of the following half-marathon races held in the Basque Country (Spain) in 2019: Behobia–San Sebastián, Zurich Marathon of San Sebastián (half-marathon modality), and/or Vitoria–Gasteiz Half-Marathon. Approval from the organizers of the indicated events was obtained before starting the study. Participant recruitment was carried out by the race organizers. Each organizer sent the questionnaire to all those who registered for the race and all the responses were collected from those runners who agreed to answer. There were no exclusion criteria; all of the responses to the questionnaire were accepted. All the participants were informed of the objectives of the study as well as the research procedure, and they voluntarily participated in it. The study followed the guidelines established in the Declaration of Helsinki (2013) and was approved by the Ethics Committee for Research with Human Beings (CEISH: M10_2019_243) of the University of the Basque Country (UPV/EHU).The study was conducted from November 2019 to January 2020. The generic data from the runners were collected one week before participating in a half-marathon (Behobia–San Sebastián, Zurich Marathon in San Sebastián or Half-Marathon in Vitoria–Gasteiz). All study participants answered the parameters included in the Hooper et al. [12] study and the SF-36 questionnaire [11] on health-related quality of life anonymously. The participants provided their answers electronically.The generic data on all the runners were obtained with a questionnaire composed of 2 sections and 12 items: (1) 2 items (items 1–2) with short responses, referring to generic data, had to be completed by the runners (sex and age) and (2) 10 items (items 3–12) with a dichotomous response (yes or no) and multiple choice that referred to sport data (i.e., number of races contested during the year [N° races], having suffered an injury or not in the two months before the race that would have prevented the athletes from carrying out their usual or planned training for at least one week [InjuryYes and InjuryNo], preparation time for the race [RTPrep], average days/km per week of training during the previous six months [AverTrainDay and AverTrainKm], having qualified staff like a bachelor or graduate in Physical Activity and Sports Sciences, in Nutrition and Dietetics or any other type of qualified staff [QualifStaffYes] or not [QualifStaffNo] for the preparation of the race and type of qualification of the staff [QualifType]).The Hooper Index, previously validated [12] and used with amateur athletes in the Spanish language [13,16,17], consisted of four items and was passed to all the participants with the purpose of assessing the perception of the level of stress (Stress), the state of fatigue (Fatigue), sleep quality (SleepQual), and muscle soreness (MusclSore) in the two weeks prior to participation in the race. The responses to the items were on a Likert-type scale (1 = very, very poor and 7 = very, very good) [21].The SF-36 Health Questionnaire [11] is composed of 36 indicators that assess both positive and negative states of health. The 36 items in the instrument cover the following dimensions: physical function, physical role, body pain, general health, vitality, social function, emotional role and mental health. For the present investigation, a previously validated version was used in the two official languages of the autonomous community where the tests were held [22,23]. The responses to the items were on a Likert-type scale (1 to 5 with different responses depending on the question: 1 = excellent and 5 = poor, 1 = yes, it limits me a lot and 5 = no, it does not limit me at all, and 1 to 3 [items intense efforts [IntEff]—bathing or dressing by themself [BathDress]: 1 = Yes, it limits me very much and 3 = no, no limits for me). The responses given to each item were used as well as the sum of the items that make up each of the 8 dimensions analyzed in the questionnaire (physical function, physical role, body pain, general health, vitality, social function, emotional role and mental health). The dimensions were scored on a 0 (poor) to 100 (good health) [24].The results are presented as mean ± standard deviation (SD) or as frequencies or percentages of the total, for each of the answers provided by the participants in each item or question. Normal distribution and homogeneity of variances were tested using the Kolmogorov–Smirnov and Levene tests, and non-parametric statistical techniques were used. The internal consistency between the different items of the SF-36 questionnaire was evaluated using Cronbach’s alpha. The Mann–Whitney U Test was used to analyze the differences existing in the different dimensions of the Hooper Index and SF-36 questionnaire between the groups (men and women, runners <40 and >40 years, InjuryYes and InjuryNo, QualifSstaffYes, and QuialifStaffNo). In addition, the percentage of the mean difference (Dif.%) was calculated through the formula: Dif. (%) = [(Mean 2 − mean 1)/mean 1] × 100). The effect size (ES) was calculated according to the method proposed by Cohen [25]. Effect sizes less than 0.2, between 0.2 and 0.49, between 0.5 and 0.79, or greater than 0.8 were considered trivial, small, moderate, or large, respectively. Statistical analysis was performed with the statistical software IBM Statistical Package for Social Sciences Statistics, version 23.0 (IBM Inc, Armonk, NY, USA). The level of significance was set at p < 0.05.Table 1 shows the descriptive results of all the half-marathon amateur runners relative to the general data (generic data and sport data). The main profile of the amateur half-marathon runner was that of a man (82.5%), >40 years (63.3%), to have competed in 3.07 ± 0.92 races per year, to not have been injured during the two months prior to participation in the race (82.1%), and to not use qualified personnel to prepare for the event (83.2%). Participants in the study trained an average of approximately three days per week and had completed 36.17 ± 20.57 km of training per week for the six months prior to participating in the race.The results for internal consistency showed adequate values for the set of items included in the SF-36 questionnaire (Cronbach’s Alpha = 0.71, n = 35 items), highlighting the total physical function (Cronbach’s Alpha = 0.91, n = 10 items), total physical role (Cronbach’s Alpha = 0.91, n = 4 items), and total emotional role (Cronbach’s Alpha = 0.94, n = 3 items) subscales. Table 2 shows the responses of the amateur half-marathon runners, as well as those obtained by the runners according to the sex and age of the participants (<40 and >40 years) on the Hooper Index and on the physical dimension of the SF-36 questionnaire. Regarding the results on the Hooper Index, women had higher stress than men before the race (Dif. = 12.5%, p < 0.01, small ES) and the group of runners of <40 years declared greater fatigue than the group of runners of >40 years (Dif. = −5.1%, p < 0.05, small ES). With respect to the variables on the physical dimension of the SF-36 questionnaire, women had lower scores in doing less than they would have liked because of physical health (PHDoLess) (Dif. = −3.9%, p < 0.05, small ES) compared to men. In addition, women showed a better quality of life in the dimensions of Physical role compared to men (Dif. = 7.1%, p < 0.05, small ES). According to the age of the runners, the group of <40 years declared better values in Health and health during the previous year (HealthPrevYear) (Dif. = 13.0% to 15.2%, p < 0.01, small ES), but showed worse results in reduced time spent at work or in daily activities because of physical health (PHRedTsDayAct), PHDoLess, difficulty doing a job or daily activities (PHDifficJob) (Dif. = −4.1% to 5.4%, p < 0.05 or p < 0.01, small ES) than the group of >40 years. The group of >40 years declared a greater limitation in performing intense efforts (IntEff) (Dif. = −1.8%, p < 0.05, trivial ES) compared to the group of <40 years. Table 3 shows the responses of the amateur half-marathon runners, as well as those obtained by the runners according to the sex and age of the participants (<40 and >40 years) on the Emotional dimension of the SF-36 questionnaire. Regarding the results on the variables of the Emotional dimension of the SF-36 questionnaire, women reduced time spent at work or in daily activities because of emotional problems (EPRedTsDayAct), doing less than they would have liked because of emotional problems (EPDoLess), doing things less carefully than they would have liked (EPLessCareful) (Dif. = −5.0%, p < 0.01, small ES), and they had lower scores in Nervous, LowMorale, discouraged and depressed (DiscourDepress), and Exhausted (Dif. = −3.7% to −6.3%, p < 0.05 or p < 0.01, small ES) compared to men. In addition, women showed a better quality of life in the dimensions of Emotional role compared to men (Dif. = 7.7%, p < 0.05, small ES). According to the age of the runners, the group of <40 years showed worse results in EPDoLess and, EPLessCareful, Nervous, LowMorale, Exhausted, Tired (Dif. = 3.3% to 9.7%, p < 0.05 or p < 0.01, small ES) than the group of >40 years. In addition, the group of >40 years showed a better quality of life in the Emotional role dimension compared to the group of <40 years (Dif. = 7.3%, p < 0.05, small ES).Table 4 shows the results obtained by the amateur half-marathon runners depending on whether they had suffered an injury (InjuryYes) or not (InjuryNo) in the 2 months prior to participating in the race, as well as depending on whether they had qualified staff (QualifStaffYes) or not (QualifStaffNo) for the preparation of the race. Regarding the Hooper Index, the group of Injury runners declared greater MusclSore than those who had not suffered injury (Dif. = −20.7%, p < 0.01, moderate ES) and the group of runners QualifStaffYes declared a higher level of Stress and Fatigue compared to the group QualifStaffNo (Dif. = −7.5% to −7.8%, p < 0.05 or < 0.01, small ES). With respect to the Physical dimension variables of the SF-36 questionnaire, the group of runners InjuryYes showed worse results in Health, HealthPrevYear, IntEff, bending or kneeling (BendKneel), and PHRedTsDayAct (Dif. = −9.7% to 5.1%, p < 0.05 or p < 0.01, small ES) compared to runners InjuryNo. In addition, the Injury group did not show a better quality of life in the Physical Function dimension compared to the InjuryYes group (Dif. = 1.0%, p < 0.01, trivial ES).The group of runners QualifStaffYes showed worse results in PHRedTsDayAct, PHDifficJob (Dif. = 2.8% to 3.2%, p < 0.05, small ES) compared to the QualifStaffNo group. By contrast, the QualifStaffNo group of runners declared greater difficulties for BendKneel because of physical health (PHBendKneel) (Dif. = −1.8%, p < 0.05, small ES) compared to the QualifStaffYes group. The QualifStaffNo group showed a better quality of life in the dimensions of Body Pain and General Health compared to the QualifStaffYes group (Dif. = 3.7% to 4.9%, p < 0.05, small ES), while the QualifStaffYes group did show better results in the Physical Function dimension (Dif. = −0.4%, p < 0.05, trivial ES).Table 5 shows the results obtained by the amateur half-marathon runners depending on whether they had suffered an injury (InjuryYes) or not (InjuryNo) in the 2 months prior to participating in the race, as well as depending on whether they had qualified staff (QualifStaffYes) or not (QualifStaffNo) for the preparation of the race. Regarding the Emotional dimension variables of the SF-36 questionnaire, the group of runners QualifStaffYes showed worse results in EPRedTsDayAct, EPDoLess, EPLessCareful, Nervous, Exhausted and Tired (Dif. = 3.0% to 5.9%, p < 0.05 or p < 0.01, small ES) compared to the QualifStaffNo group. By contrast, the QualifStaffNo group of runners declared greater difficulties for reduced social activity because of physical health and/or emotional problems (PHEPActivSocial; Dif. = 3.0%, p < 0.05, small ES) compared to the QualifStaffYes group. The QualifStaffNo group showed a better quality of life in the dimensions of Vitality, Social Function, and Mental Health compared to the QualifStaffYes group (Dif. = 3.3% to 4.6%, p < 0.05 or p < 0.01, small ES), while the QualifStaffYes group did show better results in the Emotional Role dimension (Dif. = −8.5%, p < 0.05, small ES).The aim of the present study was to analyze the wellness (i.e., stress, fatigue, muscle soreness and sleep quality) and the health status perception (i.e., physical function, physical role, body pain, general health, vitality, social function, emotional role and mental health) in amateur half-marathon runners according to sex, age, being injured or not during the previous two months before the race, and having the support or not of qualified staff for the race preparation. The results obtained in the present study show that sex, age, having suffered an injury and the fact of training with qualified professionals can influence the health and wellness status perception in amateur runners at the moment before competing in a half-marathon. The novelty of this study was that this is the first investigation to analyze individual health perception before the race of amateur half-marathon runners. The Hooper Index (i.e., the amount of stress, fatigue, muscle soreness, and sleep quality) has been reported to be associated with athletes’ training loads [14,26,27]. In addition, the validity of the use of SF-36 health data has been demonstrated valid for measuring health indicators like physical function, physical role, body pain, general health, vitality, social function, emotional role, and mental health [11]. Mainly, the wellness index and the health measures have been focused on populations with disparate characteristics such as professional athletes, people with pathologies or youngsters [20,28,29,30]. However, scarce literature is available about amateur half-marathon runners. Considering the growth of the participation rates of different sex and age amateur runners in these races, and the scant control of health indicators at an amateur level in comparison with professionals, it would be interesting to know the health self-perception of the athletes not only as a general description but also distinguishing by sex and age [31]. Knowledge about the perception of the health of the amateur runners would allow us to propose specific actions, such as training with qualified staff as well as recommendations to avoid injuries, if necessary. In this respect, the main results of our study showed that women declared higher stress levels than men. Likewise, women perceived greater difficulties due to the fact of their physical state (i.e., PHDoLess, EPRedTsDayAct, EPDoLess, EPLessCareful; Dif. = −1.8% to 13%, p < 0.05 or p < 0.01, small ES) and greater difficulties based on emotional components (i.e., LowMorale, DiscourDepress, and Exhausted; Dif. = 4.4% to 9.7%, p < 0.05 or p < 0.01, small ES) the days before the competition in comparison to men. Nevertheless, women showed better values on the physical and emotional dimensions than men. Although Batmyagmar et al. [20] did not observe differences on the self-reported health measures depending on sex in elderly (i.e., over 60 years old) marathon runners, Agrawal and D’Silva [19] reported better values in Indian men versus women both on physical and mental dimensions. In this line of thought, other authors observed that women obtained worse health indicators than men [32]. The contradictory results found in the literature may be due to the differences in the participants’ ages and characteristics in the different studies (young versus older, athletes versus non-athletes), so more studies in this regard may be necessary. On the other hand, these results could be explained by social and biopsychosocial factors which could affect women’s self-esteem. As such, previous research has highlighted that the lower self-esteem in women could be possibly due to the work pathways across the life course as most women are responsible for caring for children, carrying out the home tasks, having difficulties to work outside the home with continuing employment being uneven and stalled [33,34] which can cause them to feel less valued and withdrawn from social life and lead to a worse perception of physical and mental health.Regarding age, the half-marathon runners who were under 40 years old declared higher Fatigue than those athletes above 40 years old during the period prior to competition. In addition, athletes who were under 40 years old perceived greater difficulties due to the fact of their physical and emotional state (i.e., PHRedTsDayAct, PHDoLess, PhDifficJob, EPRedTsDayAct, EPDoLess and EPLessCareful) and showed higher levels in Nervous, LowMorale, Exhausted and Tired in comparison with athletes over 40 years old. Moreover, the athletes over 40 years old showed better quality of life values within the emotional dimension compared to athletes who were under 40 years old. Contrary to these findings, Agrawal and D’Silva [19] observed that people of the age group between 35 and 44 years old presented a better quality of life both in physical and mental dimensions, measured by the SF-36 health questionnaire, in comparison to older (over 45 years old) age groups. Surprisingly, our findings about the perception of Fatigue, as well as Physical and Emotional dimensions showed significantly lower values for half-marathon runners who were under 40 years old. These results could be explained by Waskiewicz et al. [4] as they reported associations between general motivation categories and the age of the athletes. As such, these authors observed that age was positively correlated with health orientation achievement, whereas personal goal achievement, competition, psychological coping, life meaning, and self-esteem were all negatively correlated with age [4]. Thus, it seems that while older marathon runners focus their objectives on health, the younger ones prioritize the objectives related to personal and competitive achievements and psychological aspects. This could cause higher pressure and, consequently, higher Fatigue perception and significantly lower self-reported health measures in the athletes who were under 40 years old. These results may highlight the need to analyze how runners of under 40 years prepare for competition and what factors influence having greater fatigue and a poor perception of health.Acute and long-term sport injuries negatively affect athletes who suffer them, potentially lowering their quality of life and individual running performance due to the involvement of opposed emotions and inactive time periods [35]. In this respect, Valovich et al. [21] observed that those athletes who were injured at the time of completing the SF-36 declared lower values for physical functioning and limitations due to physical health problems, body pain, social function, and the physical dimension compared to the non-injured athletes. The results of our study showed that injured athletes presented higher pain levels and worse health self-perception values in comparison to non-injured ones. Likewise, injured athletes declared a feeling of worsening health compared to the last season, as well as limitations to perform IntEff and BendKneel. Additionally, injured athletes denoted a reduction on TsDayAct and lower values in the Physical function dimension compared to non-injured athletes. One of the possible causes may be that injuries normally cause tissues to tear, resulting in a less compliant area [36], and consequently increasing the pain sensation [21]. Furthermore, injuries can negatively affect race preparation, leading to psychological consequences (e.g., frustration, depression or tension) [37] which impact the athlete’s health [38]. Thus, training sessions should include strategies focused on reducing pain as well as implementing psychological interventions during rehabilitation programs [39], for example, self-regulation techniques (thought stopping, somatic relaxation, breathing) and stress management [40] or mindfulness [41]. These strategies have been shown to have positive effects in reducing the injury rate and also help to control psychosocial factors that increase the risk of injury [42]. The implementation of these strategies could be crucial to optimizing the return to the competition process, improving the athletes’ daily lives and reducing the re-injury risk, since to have suffered a previous injury is considered as the main risk factor for the occurrence of a new injury [43].Although the quality of life based on health indicators has been analyzed for adolescent athletes and nonathletes [44], this is the first study differentiating the athletes depending on the qualified staff involved in the training process. Considering this, athletes who had the support of qualified staff (i.e., QualifStaffYes) during their preparation declared greater Stress and Fatigue levels compared to autonomous athletes (i.e., without support of qualified staff). Likewise, problems in Physical health and higher Nervous and Tired levels were reported by QualifStaffYes compared to QualifStaffNo. However, QualifStaffYes obtained better values in Physical function and Emotional role, and worse values in Body pain, General health, Vitality, Social function and Mental health dimension. On the other hand, QualifStaffNo presented greater limitations to BendKneel and in PHEPActivSocial compared to QualifStaffYes. Generally, athletes turn to qualified staff to optimize their preparation. However, this action seems to imply an increase in athletes’ stress levels [45], mainly due to the pressure that is felt by the athlete to comply with all scheduled workouts [46]. This stress can negatively impact on the athlete’s well-being due to the increased stress-related fatigue [47] as well as their general health, since psychological stress is closely related to a reduction of the immune function [48]. Additionally, the great amount of time required for high-intensity endurance training seems to negatively affect the time available for social contact [20], increasing athletes’ nervousness and fatigue. On the other hand, supervised training has been demonstrated to be safer and more effective than autonomous training in several populations [49,50]. This could explain why QualifStaffNo presented limitations such as BendKneel, since running implies a great demand on specific body structures (e.g., knee or Achilles tendon), which must be compensated for by specific training methods and recovery protocols. Those athletes QualifStaffNo would find it more difficult to apply the specific training methods and recovery protocols effectively. The results obtained in the present study suggest the importance of incorporating qualified staff such as bachelors or graduates in Physical Activity and Sports Sciences (PASS) or in Nutrition and dietetics during half-marathon preparation, mainly, to adjust the training periodization exercises and to reduce some negative emotional components [51].This study is not without limitations. The first is that no information about the training strategies performed by the athletes (i.e., training contents, methodologies, and periodization) was reported, because it was considered that an excessively long questionnaire would have reduced the athletes’ response rate. Additionally, training load values (internal and/or external variables) were not collected. This information would make it possible to establish cause–effect relationships between the training process followed by amateur half-marathon athletes and their health-related quality of life. In this sense, future research lines should consider recording training contents, methodologies, and periodization. The second limitation is that no information was collected about the results obtained in the competition. Despite this, our study presents several strengths, highlighting the large sample (i.e., 624 amateur half-marathon participants in three different competitions) and the quality of life based on health indicators differentiating between sex, age, previously injured and non-injured athletes and the qualified staff involved in the training process.The results obtained in the present study show that the health and wellness status perception is different depending on sex and age in amateur runners who compete in a half-marathon. In the same way, having suffered an injury and the fact of training with qualified professionals can influence the health and wellness status perception in amateur runners at the moment before competing in a half-marathon. Therefore, it would be interesting if both the organizers and the qualified personnel that collaborate in the preparation of amateur athletes took into account sex, age, and injury history in order to offer the best possible preparation to the athletes. The main findings of this study could be useful for the athletes, coaches and race organizers involved in the half-marathon context, to know how sex, age, having suffered an injury and training with qualified personnel affect the athletes’ well-being and health perception, in order to modify training strategies. For organizers of athletic competitions, these findings could provide them with the relevant data to know the athletes’ health self-perception and, consequently, to supply adequate recommendations based on health indicators before the race. It could be interesting for coaches to know the athlete’s well-being and health perception in order to be able to establish the objectives and the necessary training contents plan. This knowledge will allow trainers, on the one hand, to offer their professional services more appropriately, and on the other hand, to adjust the athletes’ training schedules and periodization. For runners, these data could be relevant to be aware of their individual initial state and health self-perception before the competition and, thereby, to request help, if necessary, to optimize their performance/health during races.Thus, future studies should include training strategies performed by the athletes (i.e., training contents, methodologies and periodization) and the training load values (i.e., internal and/or external training load) in order to be able to analyze the influence of these variables on the health and well-being perception. Also, sport performance should be correlated with the perception of health, training hours and training load values. Likewise, it would be interesting to carry out similar studies with runners of other distances (i.e., marathon or ultra-marathon) as well as in other sport modalities (e.g., cyclists, mountain runners) in order to compare them with the results obtained in the present study.All authors have read and agree to the published version of the manuscript. Conceptualization, E.R. and J.Y.; formal analysis, J.Y., J.R.-N. and E.R.; investigation, J.Y. and E.R.; methodology, J.Y. and E.R.; project administration, J.Y.; resources, J.Y.; supervision, J.Y. and E.R.; validation, J.Y. and E.R.; visualization, J.Y. and E.R.; writing—original draft, J.R.-N., I.A., D.C., J.R.-G., J.Y., and E.R.; writing—review and editing, J.Y., D.C., and E.R.Yes. The authors gratefully acknowledge the support of the Spanish government subproject Integration ways between qualitative and quantitative data, multiple case development, and synthesis review as the main axis for an innovative future in physical activity and sport research (PGC2018-098742-B-C31) (Ministerio de Ciencia, Innovación y Universidades, Programa Estatal de Generación de Conocimiento y Fortalecimiento Científico y Tecnológico del Sistema I + D + i), that is part of the coordinated project—New Approach to Research in Physical Activity and Sport from a Mixed Methods Perspective (NARPAS_MM) (SPGC201800X098742CV0).The authors gratefully acknowledge the “CD Fortuna”, “Fly Group 99 SL”, and “Federación Alavesa de Atletismo” (Behobia San Sebastian, Zurich Maraton de San Sebastian, and Media Maraton de Vitoria-Gasteiz, respectively) race organizers.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.Results of all the amateur half-marathon runners related to the generic and training data.<40 years = runners under 40 years old; >40 years = runners over 40 years old; N° Races = number of races contested per year; InjuryYes = having suffered an injury in the two months before the race; PrePart = previous participation in the race to be contested; PreMark = mark obtained in the previous edition; RTPrep = preparation time for the race; AverTrain = average days per week of training during the previous 6 months; QualifStaff = having qualified staff for the preparation of the race; QualifType = type of qualification of the staff for the preparation of the race; PASS = bachelor or graduate in physical activity and sports sciences; NutritDiet = bachelor or graduate in nutrition and dietetics; Others = other types of qualified staff.Results obtained for the Hooper Index and for the physical dimension of the SF-36 questionnaire obtained by the amateur half-marathon runners participating in the study and based on the sex and age of the participants (<40 and >40 years).Dif. (%) = average difference in percentage; ES = effect size; <40 years = runners under 40 years old; >40 years = runners over 40 years old; HealthPrevYear = health during the previous year; (difficulty for): IntEff = intense efforts; ModEff = moderate efforts; ShopBag = picking up or carrying a shopping bag; ClimbSevStair = climbing up several floors of stairs; Climb1floor = climbing up a single floor of stairs; BendKneel = bending or kneeling; WalkMore1km = walking a km or more; WalkSevm = walking several hundred meters; Walk100m = walking about 100 m; BathDress = bathing or dressing by themself; PH = physical health; RedTsDayAct = reduced time spent at work or in daily activities; DoLess = doing less than they would have liked; DifficJob = difficulty doing a job or daily activities; DifficJobMoreCost = difficulty doing a job or daily activities or costing more than normal; HealthWorse = health is going to get worse; HealthExcell = excellent health; Results are expressed as mean ± standard deviation. * p < 0.05 significant differences between means ** p < 0.01 significant differences between men and women or between <40 years and >40 years values.Results obtained for the Emotional dimension of the SF-36 questionnaire by the amateur half-marathon runners participating in the study and based on the sex and age of the participants (<40 and >40 years).Dif. (%) = average difference in percentage; ES = effect size; <40 years = runners under 40 years old; >40 years = runners over 40 years old; EP = emotional problems; LotEnergy = a lot of energy; PHEPActivSocial = reduced social activity because of physical health and/or emotional problems; RedTsDayAct = reduced time spent at work or in daily activities; DoLess = doing less than they would have liked; LessCareful = doing things less carefully than they would have liked; LowMorale = low morale; CalmRelax = calm and relaxed; DiscourDepress = discouraged and depressed. Results are expressed as mean ± standard deviation. * p < 0.05 significant differences between means ** p < 0.01 significant differences between men and women or between <40 years and >40 years values.Results obtained for the Hooper Index and for the Physical dimension of the SF-36 Questionnaire obtained by the amateur half-marathon runners participating in the study depending on whether they had suffered injury or not during the two months prior to participation in the race and depending on whether or not they had qualified staff for the preparation of the race.Dif. (%) = average difference in percentage; ES = effect size; Injury(Yes/No) = having suffered an injury or not in the two months before the race that would have prevented the athlete from carrying out their usual or planned training for at least one week; QualifStaff(Yes/No) = having qualified staff or not for the preparation of the race HealthPrevYear = health during the previous year; IntEff = intense efforts; ModEff = moderate efforts; ShopBag = picking up or carrying a shopping bag; ClimbSevStair = climbing up several floors of stairs; Climb1floor = climbing up a single floor of stairs; BendKneel = bending or kneeling; WalkMore1km = walking a km or more; WalkSevm = walking several hundred meters; Walk100m = walking about 100 m; BathDress = bathing or dressing by themself; PH = physical health; RedTsDayAct = reduced time spent at work or in daily activities; DoLess = doing less than they would have liked; DifficJob = difficulty doing a job or daily activities; DifficJobMoreCost = difficulty doing a job or daily activities or costing more than normal; HealthWorse = health is going to get worse; HealthExcell = excellent health. Results are expressed as mean ± standard deviation. * p < 0.05 significant differences among means ** p < 0.01 significant differences between InjuryYes and InjuryNo or between QualifStaffYes and QualifStaffNo values.Results obtained for the Emotional dimension of the SF-36 Questionnaire obtained by the amateur half-marathon runners participating in the study depending on whether they had suffered injury or not during the two months prior to participation in the race and depending on whether or not they had qualified staff for the preparation of the race.Dif. (%) = average difference in percentage; ES = effect size; QualifStaff = having qualified staff for the preparation of the race; EP = emotional problems; LotEnergy = a lot of energy; PHEPActivSocial = reduced social activity because of physical health and/or emotional problems; RedTsDayAct = reduced time spent at work or in daily activities; DoLess = doing less than they would have liked; LessCareful = doing things less carefully than they would have liked; LowMorale = low morale; CalmRelax = calm and relaxed; DiscourDepress = discouraged and depressed. Results are expressed as mean ± standard deviation. * p < 0.05 significant differences among means ** p < 0.01 significant differences between InjuryYes and InjuryNo or between QualifStaffYes and QualifStaffNo values.

Med-MDPI/ijerph_5/ijerph-17-16-05650.txt

@@ -0,0 +1 @@
+Health impacts of electronic cigarette (e-cigarette) vaping are associated with the harmful chemicals emitted from e-cigarettes such as carbonyls. However, the levels of various carbonyl compounds under real-world vaping conditions have been understudied. This study evaluated the levels of carbonyl compounds (e.g., formaldehyde, acetaldehyde, glyoxal, and diacetyl, etc.) under various device settings (i.e., power output), vaping topographies, and e-liquid compositions (i.e., base liquid, flavor types). The results showed that e-vapor carbonyl levels were the highest under higher power outputs. The propylene glycol (PG)-based e-liquids generated higher formaldehyde and acetaldehyde than vegetable glycerin (VG)-based e-liquids. In addition, fruit flavored e-liquids (i.e., strawberry and dragon fruit) generated higher formaldehyde emissions than mint/menthol and creamy/sweet flavored e-liquids. While single-top coils formed 3.5-fold more formaldehyde per puff than conventional cigarette smoking, bottom coils generated 10–10,000 times less formaldehyde per puff. In general, increases in puff volume and longer puff durations generated significantly higher amounts of formaldehyde. While e-cigarettes emitted much lower levels of carbonyl compounds compared to conventional cigarettes, the presence of several toxic carbonyl compounds in e-cigarette vapor may still pose potential health risks for users without smoking history, including youth. Therefore, the public health administrations need to consider the vaping conditions which generated higher carbonyls, such as higher power output with PG e-liquid, when developing e-cigarette product standards.Carbonyl compounds are the most abundant toxic chemicals emitted from electronic cigarettes (e-cigarettes) [1,2,3,4]. Among the carbonyls found in e-cigarette emissions (e-vapor), two of the major carbonyls (i.e., formaldehyde and acetaldehyde) are known human carcinogens [5]. Glyoxal, which is known to cause allergic reactions, was also found in e-vapor [6]. Several flavoring carbonyls in e-liquids (e.g., vanillin, cinnamaldehyde) have shown increased cell toxicity [7,8]. It is also worth mentioning that 2,3-butanedione (diacetyl) and 2,3-pentanedione (acetylpropionyl) were found in e-vapors with ‘buttery’ flavored e-liquids [9] which are known to cause bronchiolitis obliterans (aka ‘popcorn lung’) [10]. However, factors affecting carbonyl emissions from e-cigarettes are still not fully understood. Most current studies report that higher e-cigarette power outputs significantly increase formaldehyde and acetaldehyde emissions [11,12,13,14,15,16,17]. However, there is a 1000-fold difference across the literature in formaldehyde emission rate data from e-cigarettes at comparable e-cigarette device power outputs [12,13,14], suggesting that carbonyl emissions are not dependent on e-cigarette power output alone and that other factors should be considered.Among the potential other factors, flavoring agents used in e-liquids could be sources of potentially harmful carbonyl emissions. Thermal fragmentation of flavoring chemicals has been shown to form carbonyls under the burning temperatures of conventional cigarettes [18], but the contribution of flavoring chemicals to carbonyl emission has not been sufficiently studied for e-cigarettes [19,20]. In addition, some flavored e-liquids and e-vapors contain chemicals [e.g., diacetyl and acetylpropionyl] that can be toxic at sufficient levels [7,8,9]. However, levels of the potentially harmful chemicals in e-vapor resulting from flavoring additives were not fully accessed under e-cigarette vaping topography [9].The contribution of e-liquid base materials to carbonyl emissions is also largely unknown. It has been shown that thermal degradation of vegetable glycerin (VG) and propylene glycol (PG), the main base materials used in e-liquids, generates various carbonyl compounds [15,21]. However, the thermal degradation of VG and PG has not been studied across a wide range of e-cigarette coil temperatures [14,16,22] even though coil temperature is an important determinant of e-cigarette carbonyl emissions [13].E-cigarette vaping topography (i.e., puff volume and duration) could also potentially affect carbonyl formation. Vaping topography can affect carbonyl emissions by modifying e-cigarette heating coil temperatures [23]. However, most of the previous studies examining e-cigarette carbonyl emissions could not mimic actual e-cigarette user’s behavior [12,13,15,24,25]. Those studies generated e-vapors similar with the ‘Health Canada Intense (HCI) Regime’ (55 mL puff volume, 2 s puff duration, every 30 s) which was developed for conventional cigarette smokers and not e-cigarette vapors [26]. Therefore, real-world vaping topography should be used to study carbonyl emissions. In addition, most of the preceding studies only focused on formaldehyde and acetaldehyde emissions. However, other potentially harmful carbonyl compounds, such as acrolein and glyoxal, in e-vapor also need to be evaluated under real-world vaping conditions.To address these knowledge gaps, this study evaluated the impacts of real-world vaping conditions (i.e., real-world e-cigarette heating power, vaping topography, and e-liquid components) on the emission of six potentially harmful carbonyls (i.e., glyoxal, formaldehyde, acetaldehyde, acrolein, diacetyl, and acetylpropionyl) and thirteen additional carbonyl species described below.E-vapors were generated based on the protocols used in our previous studies [27,28,29]. In brief, vaping topographies of 23 current e-cigarette users measured in our lab. Vaping topographies (i.e., puff volume, duration, and interval) were measured using a CReSS Pocket device (Borgwaldt KC Incorporated, North Chesterfield, VA, USA) [30] with the Rutgers’ IRB approval (Pro20140000589). Demographic data of the study participants, observed vaping topographies, device settings, and e-liquid compositions from the users are summarized in Tables S1 and S2.The wide ranges of experimental conditions were used reflecting real-world vaping conditions (Table 1). Throughout the experiments, the median values of the observed power output (6.4-watt), puff volume (90 mL), and puff duration (3.8 s) were used. The selected vaping topography was consistent with the median value of the reported e-cigarette vaping topographies, which was 91 mL puff volume (ranging from 51 to 133 mL) and 3.8 s puff duration (ranging from 2.65 to 4.3 s) [8,31,32,33,34,35,36,37]. Square-shaped vaping pattern was used (Figure S1) instead of bell-shaped topography which is used for cigarette smoking.A variety of e-liquids were prepared to assess the impact of e-liquid compositions on carbonyl formation. Freshly prepared e-liquids in our lab were used in all experiments for quality control purposes because large variability in commercial products were reported in a previous study [38]. E-liquids used in our study were composed of vegetable glycerin (VG, USP grade, J.T. Baker, Phillipsburg, NJ, USA), propylene glycol (PG, USP grade, Sigma-Aldrich, St. Louis, MO, USA), (-)-nicotine (≥99.0%, Sigma-Aldrich), and flavoring agents (The Perfumer’s Apprentice, Scotts Valley, CA, USA). Three base materials, 100% VG, PG:VG mixture (v/v = 1:1), and 100% PG e-liquids, containing 12 mg/mL nicotine were tested to evaluate the impact of e-liquid base material on carbonyl formation. Then, eight flavored e-liquids (strawberry, dragon fruit, menthol, sweet cream, Bavarian, cinnamon, bubble gum, and graham cracker) were used to generate e-vapors to evaluate the impact of flavoring agents on carbonyl formation. These eight flavoring agents selected are the most frequently used ones in e-liquid recipes on the market [39]. The flavored e-liquids in the experiments were prepared by adding 10% of flavoring agents (1% for the cinnamon flavor) in VG by volume ratio. The chemical components of the flavoring agents were not fully disclosed but all of them were consisted of natural/artificial flavorings in PG as a main solvent (see more information in Supporting Information). All the e-liquids were prepared freshly before the day of experiments and sat overnight for stabilization. The e-liquid preparing procedures were consulted by local vape shop owners.The device used in this study was a versatile refillable tank type e-cigarette which contained replaceable Nichrome heating coils (dual-bottom coils with 0.8 Ω resistance) and was obtained from an e-cigarette retailer (The Council of Vapor, Walnut Creek, CA, USA). Two types of battery boxes, an Apollo Valiant battery (Apolo E-cigarette, Concord, CA, USA) and a Sigelei-100W battery (Sigelei US, Pomona, CA, USA), were used to provide a wide range of power outputs from 3 to 80 watts. To test the impact of device power output settings and vaping topographies on carbonyl emission, the average and the 95% of the observed device power outputs (14.7 watts and 31.3 watts) and 95% of observed puff volume (170 mL) from the 23 subjects were used for e-vapor generation. The variable device power outputs were achieved by changing the battery voltage with fixed coil resistance (0.8 Ω).The sampling and analytical protocols for carbonyl measurements were developed based on the U.S. EPA compendium method OA-11A [40]. In brief, the sampling system was composed of a LXe1 smoking machine (Borgwaldt KC Incorporated, Hamburg, Germany), a sampling chamber, a 2,4-dinitrophenylhydrazine (DNPH) cartridge (Waters, Milford, MA, USA), and a vacuum pump (Figure S2). E-vapors were generated using the smoking machine under the experimental conditions described above and specified in Table S3. The generated e-vapors were directly introduced into the 2 L sampling chamber. The inlet of the DNPH cartridge was connected with the sampling chamber, and 30 puffs of e-vapor were collected using the DNPH cartridge with a sampling flow rate of 200 mL/min. After sampling, both ends of the DNPH cartridge were sealed, and the cartridge was stored in an aluminum zip lock bag at 4 °C until analysis.The sampled DNPH cartridges were then eluted with 4 mL of high-performance liquid chromatography (HPLC) grade acetonitrile (ACN, Sigma-Aldrich), and 20 μL of the eluted DNPH-aldehyde derivatives were injected into the HPLC/UV system (a Perkin Elmer Series 200 HPLC and a Perkin Elmer 785a UV/vis detector, Perkin Elmer, MA, USA) which was equipped with a Waters Nova-Pak C18 column. The mobile phase was programmed as follows: after holding 100% of solvent A (H2O, HPLC grade, Sigma-Aldrich,/ACN/THF [tetrahydrofuran, Fisher Scientific, Hampton, NH, USA] = 6/3/1) for 4 min, the mobile phase was changed to 100% solvent B (ACN/H2O = 6:4) over 20 min, then 100% solvent B was held for 10 min. The mobile phase was set to a constant flow rate of 1 mL/min and the UV detector was set at an absorbance wavelength of 365 nm.Calibration curves for the nineteen carbonyls were prepared using purchased DNPH-aldehyde analytical standards (ResTek, Bellefonte, PA, USA) and five carbonyl standards prepared in our lab (Table S3). For the preparation of these five carbonyl standards, glyoxal (40%) and vanillin (99%) were obtained from Sigma-Aldrich. Cinnamaldehyde (≥98), diacetyl (99%), and acetylpropionyl (97%) were purchased from Alfa Aesar (Haverhill, MA, USA). To prepare the standards, known amount of the five carbonyls were spiked into the DNPH cartridge and eluted with ACN. Limits of detection (LOD) and limits of quantification (LOQ) were three and ten times the standard deviations of the standard with the lowest concentration (n = 7).For all the experimental conditions, mean and standard deviations were estimated and presented. Two-tailed Student’s t-tests were conducted using R 3.4.3 (R Foundation, Vienna, Austria) to compare the mean values across different e-cigarette vaping conditions.Figure 1 shows the impact of e-cigarette power output settings and variations in base material on the emission of formaldehyde and acetaldehyde, which are carcinogenic carbonyls found in e-vapor. Higher device power outputs increased formaldehyde emissions from all three base materials (Figure 1a). The amounts of formaldehyde generated at 31.3 watts were 39.3%, 111.0%, and 142.0% higher than the amounts of formaldehyde generated at 6.4 watts for VG, PG:VG, and PG based e-liquids, respectively. At 31.3 watts e-cigarette power output, PG:VG and PG based e-liquids generated 57.8% and 86.9% more formaldehyde than VG based e-liquids (p < 0.003).PG-based e-liquids generated significantly higher amounts of acetaldehyde than VG-based e-liquids (Figure 1b). PG:VG mixture (v:v = 1:1) and PG based e-liquids generated 2.7 and 8.5 times more acetaldehyde than VG-based e-liquids at 6.4 watts power output condition (p < 0.001). The increase in e-cigarette power output from 6.4 watts to 31.1 watts did not increase acetaldehyde formation for VG e-liquid, but significantly increased for PG:VG and PG based e-liquids by a factor of 2.8–4.7 (p < 0.001). In addition to formaldehyde and acetaldehyde, higher power output also generated other harmful carbonyls (Table S4). We observed 240 ± 13.7 ng/puff of glyoxal generated from VG-based e-liquids at 31.3 watts. Compared with 6.4 watts conditions, 31.3 watts generated twice the amount of acrolein, n-butylaldehyde and isovaleraldehyde (p < 0.001). Interestingly, PG and PG:VG based e-liquids under higher power outputs emitted higher amount of o-tolualdehyde, while highest n-hexaldehyde were observed under medium power output (14.7 watt) for the all e-liquids.Carbonyl compounds generated from flavored e-liquids are shown in Figure 2 and Table S5. The fruit-flavored e-liquids (i.e., strawberry and dragon fruit) generated 1.7–2.6 times higher amounts of formaldehyde than spicy (cinnamon), and creamy/sweet (Bavarian cream, sweet cream, bubble gum, and graham cracker) flavored e-liquids.The fruit and menthol flavored e-liquids generated 5–40% more formaldehyde than non-flavored VG e-liquid, while other flavored e-liquids generated less formaldehyde. In addition, acetaldehyde levels generated from the flavored e-liquids were below or similar to the quantification limit. Acrolein was generated at levels of approximately 20–30 ng/puff for 5 of the flavored e-liquids and more than 4 times of that level for Graham cracker flavor.Diacetyl and acetylpropionyl are the ‘butter’ flavoring chemicals and are known to increase lung airway injury (aka ‘popcorn lung’) [41]. Three flavored e-liquid samples (i.e., Bavarian cream, sweet cream and graham cracker flavors) emitted 21.1–86.4 ng/puff of diacetyl, but acetylpropionyl was not detected in any sample. Four flavored e-liquids (i.e., Bavarian cream, sweet cream, bubble gum, and graham cracker flavors) emitted 45.2–184.4 ng/puff of vanillin, and cinnamaldehyde (473.1 ± 234.9 ng/puff) was detected only from the cinnamon flavored e-liquid.Variations in puff volumes and puff durations significantly changed carbonyl levels in e-vapors (Table 2). An increase in a puff volume from 35 mL to 90 mL led to 15.6% and 23.8% higher amounts of formaldehyde formation for 2 s and 3.8 s puffs, respectively (p < 0.016), while the mean formaldehyde emissions at 90mL and 170 mL puff volumes with same puff durations were not significantly different. Longer puff durations generated significantly higher amounts of formaldehyde. Acetaldehyde levels under the flow rates faster than 45 mL/s (i.e., 90 mL, 2 s puff and 170 mL, 2 s puff) were below limit of quantification. o-Tolualdehyde levels were higher at 2 s puff durations than 3.8 s puff durations at each puff volume, while higher n-hexaldehyde emissions were observed at 3.8 s than 2 s puff durations.This study adds new evidence on the levels of carbonyls emitted from e-cigarette. In this study, a variety of carbonyls were measured under wide ranges of e-cigarette use patterns: e-liquid compositions, power outputs, and vaping topography. The formation of carbonyls using various combinations of base materials and device power outputs were explored. Thermal decomposition of VG and PG forms carbonyls during e-cigarette vaping [21], and increased coil temperatures accelerate the decomposition rates of e-liquid base materials [42]. PG formed higher levels of carbonyl compounds (e.g., formaldehyde, acetaldehyde, butylaldehyde, tolualdehyde) during e-cigarette vaping than VG, probably because PG has a lower thermal decomposition temperature than VG. The thermal decomposition of PG starts as low as 127 °C [43], while VG requires at least 200 °C for its thermal decomposition reaction to begin [44]. Kosmider et al. [14] also reported that PG-containing e-liquids generated significantly higher amounts of formaldehyde and acetaldehyde than VG based e-liquids. Glyoxal, which was shown to cause allergic reaction [6], was observed only with VG-based e-liquid under high power output setting. It has been reported that thermal oxidation of VG leads to the formation of glyoxal [15].Higher e-cigarette power outputs increased formaldehyde emissions for both top and bottom coils (Figure 3). The formaldehyde levels observed in this study were within interquartile ranges of literature values [4,13,15,45,46,47], but we observed wide varieties in reported formaldehyde levels. The reported e-vapor formaldehyde levels have been shown to range from 0.01 µg/puff to 342.2 µg/puff for top coil device and range from 0.02 µg/puff to 220.0 µg/puff for bottom coil e-cigarettes [4,13,14,15,19,22,45,46,47,48,49]. The wide range of e-cigarette carbonyl emission levels reported literatures might be a factor of the coil settings [13]. The top coils formed higher amounts of formaldehyde per puff (23.35 ± 59.68 µg/puff) with conventional cigarette smoking (12.32 ± 9.65 µg/puff) due to the limited e-liquid supply to the heating coil. A top coil is located on the top of the atomizer with long wicks dropping down into the e-liquid tank (Figure S3).A long wick cannot supply enough e-liquid to the coil, and the limited e-liquid supply can easily dry up the heating coil, leading to a rapid coil temperature increase. The dramatic increase of coil temperature is known as ‘dry puff’ or ‘dry hit’, which results in significantly increased amounts of carbonyl formation [48]. In contrast, a bottom coil is located at the bottom of the atomizer, with a short wick contacting the e-liquid (Figure S3). Bottom-coils, commonly used in the current generations of e-cigarettes, generally provide consistent hits without ‘dry puffs’. Consequently, a bottom coil generated 10–10,000 times less formaldehyde per puff than conventional cigarettes due to stable e-liquid supply rates and coil temperatures. Gillman et al. [13] stated that e-cigarette devices with steady e-liquid supplies to the coil generated the lowest amounts of formaldehyde.In addition to the device construction, variations in aldehyde emissions from flavored e-liquids might be affected by the differences in boil points, evaporation rates, and thermal decomposition rates [13]. But, the reasons for the differential carbonyl formation patterns across different flavoring agents are cannot be explored completely because flavor manufacturers usually do not disclose the ingredients in their products [53]. Based on partially revealed information from one vendor (The Perfumer’s Apprentice), the flavoring agents consist of PG, water, ethyl alcohol, and natural/artificial flavoring chemicals. Pyrolysis of flavoring chemicals was known to be the major source of carbonyls in e-vapor [19,20]. In addition, PG in flavoring agents might also contribute to aldehyde formation. However, current knowledges on carbonyl formation from flavored e-liquids are not fully understood. Further studies of thermal degradation of flavoring chemicals are warranted to better understand the contribution of flavoring agents to carbonyl formation.Diacetyl concentrations observed in our samples are comparable to those reported in previous studies [9,37]. Even though many e-liquid manufacturers use acetoin as a safe alternative of diacetyl and acetylpropionyl, diacetyl could be found in e-vapor due to the use of natural flavors containing diacetyl and acetoin-to-diacetyl conversion during storage [10,54]. Based on the diacetyl concentrations we measured, assuming 200 puffs/day of e-cigarette vaping on average, daily diacetyl exposure levels of butter-like flavored e-cigarette users (4.22–17.28 g/day) were much lower than the reported threshold level. In a rodent in vivo study, diacetyl exposures (100 ppm) to 6 h per day for 12 weeks (equivalent to 6.82 mg/day for 12 weeks assuming 0.2 mL mean tidal volume and 250 breaths/minute) caused nasal injury and peribronchial lymphocytic inflammation [55]. However, considering uncertainties in animal-to-human extrapolation and extreme e-cigarette users (e.g., cloud chasers using sub-ohm e-cigarettes with intensive use, >1000 puffs/day), potential health impact of chronic low-level diacetyl exposures should be further accessed by the regulatory authorities. In addition to the diacetyl, other flavoring chemicals (e.g., vanillin and cinnamaldehyde, etc.) could promote adverse health outcomes. An in vitro study demonstrated that vanilla and cinnamon flavored e-liquids had three and ten-fold lower no-observable-adverse-effect-level (NOAEL) doses (0.1–0.01% dose) than VG only e-liquid (0.3% in culture media), respectively [7]. The cytotoxicity of cinnamaldehyde (IC50 = 0.037–0.04 mM) was approximately 100 times higher than that of vanillin (IC50 = 2.5–4 mM) [8]. The impact of flavoring chemicals on human health need to be further studied using real-world relevant doses, such as presented in this study, because the flavoring chemicals have been identified as one of the most concerning chemicals found in e-cigarette emissions [56,57].Increased puff volumes with a fixed puff durations were shown not only to increase the amounts of e-vapor passing through the heating coil but also to decrease its temperature due to increased flow rates [23]. Puff volumes and puff durations determine the volume of air and its flow rate passing through the e-cigarette heating coils. The significant differences in carbonyl emissions measured between 35 mL and 90 mL puff volumes observed in our study might be due to the increased e-vapor masses. However, the carbonyl composition might also be affected by coil temperature changes. The formation of diverse carbonyl species under different air flow regimes might indicate that the changes in coil temperature could affect the thermal degradations of the e-liquid components.Our study makes an important contribution to the literature by using vaping topographies based on real-world user vaping patterns. As noted earlier, the puff volumes used in most previous e-cigarette vapor emissions studies were based on regular cigarette smoking topographies and were usually much lower than that of e-cigarette users. The short puff durations used in previous studies (≤2 s) might be insufficient to heat up the heating coil to evaporate e-liquid [23] and thus may have underestimated potential exposures to toxic carbonyl emissions from e-cigarette vapor during real-world usage.Carbonyl exposure distributions for e-cigarette and conventional cigarette were estimated using the Monte Carlo method (Figure 4). Input parameters were the observed e-cigarette emission data in this study (Table 2 and Tables S4 and S5), reported cigarette smoke carbonyl levels (Table S6) [50], and daily e-cigarette and cigarette use patterns [31,58]. The distribution of carbonyl exposures associated with recent generation e-cigarettes with bottom coil setting were compared to the exposures from conventional cigarette smoking (Figure 4 and Table S7). Daily average acetaldehyde, diacetyl, and acrolein exposures from e-cigarette were approximately 100, 125 and 21 times lower than conventional cigarette, respectively, with little to no overlap of the exposure populations. However, e-cigarette users could be exposed to 2- and 4-fold higher formaldehyde and glyoxal in a day than cigarette smokers, respectively, and near complete overlap of the distributions. Given the daily exposure estimates, e-cigarette users should aware that e-cigarette might be less effective harm reduction product when they employ vaping conditions that resulted in high carbonyl formation (e.g., top-coil device, high power output, PG e-liquid, and large flavoring additives, etc.). In addition, e-cigarette vaping is still expected to pose potential health risks due to the non-threshold characteristics of carcinogenic carbonyls (i.e., formaldehyde and acetaldehyde) and should not be considered harmless. It is also worth mentioning that, as noticed above, e-cigarette users are prone to be exposed to more glyoxal than cigarette users. Glyoxal has been identified as an occupational allergen among health care workers who use glyoxal containing disinfectants [6]. An in vitro study showed that glyoxal was shown to deplete glutathione, increase the production of reactive oxygen species (ROS), and induced cell damage to isolated rat hepatocytes [59]. Higher device power outputs could increase glyoxal exposures and via the mechanisms listed above induce airway oxidative stress. Since there is no such study of health impacts of e-cigarette glyoxal exposure, it needs to be further evaluated for better harm reduction.Moreover, carbonyl compounds in e-vapor were shown to form secondary harmful chemicals. Autoxidation of acetoin, which is a safer alternative of ‘butter’ flavoring chemicals (i.e., diacetyl and acetylpropionyl) could form diacetyl during e-liquid storage [54]. Further, acrylamide is neurotoxic and has potency to cause cancers in the reproductive and endocrine systems [60]. The reaction between acrolein, which is known to present in e-cigarette vapor, and amino acid or ammonia could form acrylamide [61], but the formation of acrylamide has not been studied in e-vapor. In addition to acetoin and acrolein, other precursor chemicals may present in e-vapors. Future research needs to access the formation of secondary air toxics induced by e-vapor.Even though we thoughtfully identified large numbers of carbonyl compounds induced by various e-cigarette vaping conditions, this study still has several limitations. First, the DNPH cartridges were designed for the gas phase carbonyl sampling rather than particle phase carbonyls [40]. Carbonyl collection efficiencies for the e-vapors using the DNPH cartridges might be lower than the labeled efficiencies for the gas phase carbonyls because carbonyls in e-vapors are reported to be present in both gas and particle phases [15]. Second, our analytical method might have underestimated unsaturated aldehydes and ketones. Unsaturated carbonyls, such as acrolein, crotonaldehyde, and cinnamaldehyde, and DNPH adducts could further react with additional DNPH to form side products [62]. Further studies also need to test additional carbonyl sampling methods such as N-methyl-4-hydrazino-7-nitrobenzofurazan, 4-(2-aminooxyethyl)-morpholin-4-ium chloride or DNPH-hydroquinone methods.E-vapor carbonyl levels vary under different vaping conditions and product materials. Higher carbonyl levels were found for PG e-liquids, higher power outputs, and top coil settings. PG-based e-liquids under 31.3-watt generated approximately 2.6, 11.2, and 200-fold higher formaldehyde, acetaldehyde, and acrolein than VG e-liquid under 6.4-watt. Despite the fact the FDA has announced a guideline that regulates e-cigarette flavorings for e-cigarette devices targeted to minors or marketed to target minors [63], our results are still useful because there is still huge flavored e-liquid market deemed to not target youth. It is worth mentioning that the flavored e-liquids that are not being sold after the guideline implementation (e.g., JUUL pods except for tobacco and menthol flavor) could always come back onto the market when the vendor meets the standards set out in the premarket authorization [63]. Our findings suggest that upper limits of e-cigarette power output need to be provided for different e-liquid compositions and coil types to minimize carbonyl emission. In addition, other potentially harmful carbonyls, diacetyl, vanillin, and cinnamaldehyde, were identified from the flavored e-liquids. Furthermore, flavored e-liquids changed the profile of carbonyl formation. However, the impact of flavoring chemicals could not be satisfactorily explored due to the limited availability of product content information. Future studies, therefore, will need to evaluate the impact of flavoring chemicals on the formation of carbonyls formation and the inhalation toxicity of flavoring chemicals by themselves.The following are available online at https://www.mdpi.com/1660-4601/17/16/5650/s1, Table S1. Summary of the study participants, Table S2. E-cigarette vaping patterns from the study subjects (N = 23), Figure S1. E-cigarette vaping topography observed from 23 e-cigarette users, Figure S2. Scheme of the carbonyl sampling system, Table S3. Retention times, calibration parameters, LODs, and LOQs for the selected carbonyls, Table S4. Impact of power outputs and base materials on carbonyl levels in e-vapor (mean ± standard deviation, n = 5), Table S5. Impact of flavoring agents on carbonyl levels in e-vapor (mean ± standard deviation, n = 5), Figure S3. Example of the top and bottom coil settings, Table S6. Carbonyls emitted from the e-cigarette and conventional cigarette. E-cigarette carbonyl levels were measured in this study and carbonyl emissions from cigarette were adopted from Fujioka and Shibamoto (2006), Table S7. Estimated daily carbonyl exposures for e-cigarette and cigarette users.Conceptualization, Y.S., Q.M., C.D.; methodology, Y.S., Q.M., C.W.; resources, Q.M. and C.D.; formal analysis, Y.S., Q.M.; investigation, Y.S., Q.M., C.W.; visualization, Y.S.; writing—original draft preparation, Y.S.; writing—review and editing, Y.S., Q.M., C.W., C.D., O.W., S.S.; supervision, Q.M.; project administration, Q.M.; funding acquisition, Q.M. All authors have read and agreed to the published version of the manuscript.The authors wish to thank Cancer Institute of New Jersey, New Jersey Health Foundation, and National Institute of Environmental Health Sciences (P30 ES005022) for funding this study, and all study participants for their support of the study. Contributions by OW were supported in part by K01 CA189301 from the National Cancer Institute and the FDA Center for Tobacco products. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding organizations.The authors declare no conflict of interest.(a) Formaldehyde and (b) acetaldehyde levels (µg/puff) generated by combinations of base materials (100% VG, PG:VG 1:1 (v/v), and 100% PG) and device power outputs (6.4, 14.7, and 31.3 watt for low, medium, and high power, respectively). 90 mL puff volume, 3.8 s puff duration, and 24 s intervals was used as vaping topography, and 12 mg/mL nicotine was added into all e-liquids.Carbonyl levels (ng/puff) generated by different flavored e-liquids (ST: strawberry, DF: dragon fruit, MT: menthol, CM: cinnamon, BC: Bavarian cream, SC: sweet cream, BG: bubble gum, and GC: graham cracker, 10% by volume, 1% for cinnamon flavor in VG-base) and non-flavored VG e-liquid (VG). Other vaping parameters are 6.4 watts power output, 90 mL puff volume, 3.8 s puff duration, and 24 s puff interval.Formaldehyde levels (log-scale) from e-cigarettes with different coil structure (top- and bottom coil) and power outputs (low and high levels determined by lower and higher power outputs [5 watts for top coil and 18 watts for bottom coil] used in the literatures). Red dots indicate our results, and black dots are the results obtained from literatures [4,13,14,15,19,22,45,46,47,48,49]. Formaldehyde levels in cigarette smoke from literatures [50,51,52] are shown for comparison purpose.Estimated daily carbonyl exposures (µg/day) for (A) e-cigarette vapers and (B) conventional cigarette smokers.E-vapor generation conditions used in this study 1.1 24 s puff intervals were used for all conditions; 2 Strawberry, Dragonfruit, Menthol, Cinnamon, Bubblegum, Bavarian cream, Sweet cream, and Graham cracker; 3 1% for cinnamon flavor (v/v).Impact of vaping topography on carbonyl levels in e-vapor (mean ± standard deviation, ng/puff).† 6.4 W power output, 1.5 mm air hole, VG based e-liquid containing 12 mg/mL nicotine, and 24 s puff interval were used; †† ND indicates non-detected; ††† <LOD indicates the measurement which is below the detection limit; †††† <LOQ indicates the measurement which is below the quantification limit.

Med-MDPI/ijerph_5/ijerph-17-16-05651.txt

@@ -0,0 +1 @@
+Estimation of the intensity of physical activity (PA) based on absolute accelerometer cut points (Cp) likely over- or underestimates intensity for a specific individual. The purpose of this study was to investigate the relationship between absolute moderate intensity Cp and the first ventilatory threshold (VT1). A group of 24 pregnant and 15 nonpregnant women who performed a submaximal incremental walking test with measures of ventilatory parameters and accelerations from three different accelerometers on the wrist (ActiGraph wGT3X-BT, GENEActiv, Axivity AX3) and one on the hip (Actigraph wGT3X-BT) were analyzed. Cp were determined corresponding to 3 metabolic equivalents of task (MET), using the conventional MET definition (Cp3.5) (3.5 mL/kg×min) and individual resting metabolic rate (Cpind). The ventilatory equivalent (VE/VO2) was used to determine VT1. Accelerations at VT1 were significantly higher (p < 0.01) compared to Cp3.5 and Cpind in both groups. Cp3.5 and Cpind were significantly different in nonpregnant (p < 0.01) but not in pregnant women. Walking speed at VT1 (5.7 ± 0.5/6.2 ± 0.8 km/h) was significantly lower (p < 0.01) in pregnant compared to nonpregnant women and correspondent to 3.8 ± 0.7/4.9 ± 1.4 conventional METs. Intensity at absolute Cp was lower compared to the intensity at VT1 independent of the device or placement in pregnant and nonpregnant women. Therefore, we recommend individually tailored cut points such as the VT1 to better assess the effect of the intensity of PA.Accelerometers are often used as a measure of free-living physical activity (PA) to estimate frequency, duration, and intensity [1]. In order to estimate the intensity of PA, calibration of accelerometers based on physiological responses to various activities is required. Calibration studies use oxygen consumption as the criterion measure and specific statistical methods, such as regression models or machine learning-based modeling, to determine the corresponding accelerometer cut point at a certain intensity threshold [2,3]. Usually, these intensity thresholds are used to define activity classes based on the average energy expenditure, expressed in multiples of the conventional metabolic equivalent of task (MET) ratio. The 1 MET value is defined as the energy expended by a subject at rest, which equals an oxygen consumption of 3.5 mL/kg×min for a 70 kg person [4]. Common classifications of PA intensity classes are light (< 3 METs), moderate (3–5.9 METs), and vigorous (≥ 6 METs) [5,6,7]. However, using the conventional 1 MET to classify PA intensity classes already causes misclassifications, compared to approaches where the individual resting metabolic rate would have been used [8,9]. Moreover, it must be considered that accelerometer cut points, which are based on the same absolute approach, are independent of individual performance capacity which leads to different physiological and metabolic strain (internal load) at the same absolute intensity [10]. To counteract that, relative accelerometer cut points based on the individual maximum performance capacity can be used. Several studies showed that the estimation of the duration of moderate-to-vigorous physical activity (MVPA) was shorter when relative comparisons to absolute cut points were applied, independent of body mass index (BMI). Therefore, absolute cut points overestimate MVPA compared to relative cut points [2,11]. When applying relative cut points, activity counts at a fixed relative intensity (e.g., 60% heart rate reserve (HRR)) increase with increasing fitness level [12]. On the contrary, low-fit persons had significantly higher percentage of maximum oxygen uptake (%VO2max) at absolute cut points (e.g., 2020 cpm) compared to fit persons [11]. The use of absolute accelerometer cut points is therefore likely to over- or underestimate PA intensity and volume in a certain PA intensity class for a specific individual. Individualized activity measurements (i.e., relative accelerometer cut points), which take into account the performance capacity of each individual, will allow us to draw more valid conclusions about the internal loads due to different intensities of PA.Relative accelerometer cut points are usually derived using fixed percentages of an individual’s maximum heart rate (HRmax), maximum oxygen uptake (VO2max), or HRR [13]. Such an individualized approach improves validity, but will still lead to some inaccuracy in the prescription of exercise intensity, because exercising at a calculated and fixed relative intensity such as 85% HRmax causes different metabolic and cardiorespiratory responses across individuals [14,15]. To overcome these problems, exercise can be prescribed based on submaximal markers such as the first and second ventilatory or lactate thresholds (VT1 and VT2; LTP1 and LTP2) which rely on the detection of physiological thresholds dependent on exercise intensity [16,17]. Hence, less variability in interindividual metabolic responses is expected when being active at a certain intensity relative to these thresholds. Indeed, Moser et al. [18] recently showed consistent metabolic responses during continuous cycle ergometer exercise five percent above and below LTP1 and LTP2. Furthermore, lactate thresholds as well as their physiological equivalents VT1 and VT2 allow conclusions about the fitness level and adaptations to exercise, and are very sensitive in reflecting differences in endurance performance [16,17].Two studies by Gil-Rey et al. [19,20] estimated the intensity of PA from individually tailored accelerometer cut points derived from lactate thresholds in postmenopausal women. Individually tailored cut points revealed similar time for MVPA in high- and low-fit groups. In contrast, MVPA was overestimated in low-fit and more strongly in the high-fit group when absolute accelerometer cut points at moderate intensity (3–5.9 METs) were applied. They showed that individually tailored rather than traditional absolute accelerometer cut points estimate an individual’s activity level (i.e., time spent in different intensities of PA) more accurately. Thus, using individually tailored cut points from physiological thresholds may provoke greater adaptions to exercise and reduced interindividual variability of metabolic responses, as well as less overestimation of PA intensity. To date, it is unknown how well absolute accelerometer cut points are related to a physiological threshold such as the VT1 in pregnant as well as in young nonpregnant women. We hypothesized that the two absolute accelerometer cut points, derived from the 3 MET moderate intensity definition using either the conventional 3.5 mL/kg×min or the individual resting metabolic rate, will be lower compared to the individual threshold derived from VT1.Therefore, the aim of this study was to compare accelerometer cut points from different devices and placements, derived from the 3-MET absolute moderate intensity definition with the first ventilatory threshold (VT1) determined in a short submaximal incremental walking test (IWT) in a sample of pregnant and younger nonpregnant women.In total, a group of 30 pregnant (second and third trimester) and 17 nonpregnant women were tested at the University of Graz, Austria and the University of the Witwatersrand, Johannesburg, South Africa (22 pregnant, 9 nonpregnant). Subjects in Graz were active pregnant women and a mix of highly active and well-trained nonpregnant women (mainly students). In South Africa, subjects were sedentary and low-active pregnant women and a mix of sedentary, low-active, and well-trained nonpregnant women. Before any study procedures were undertaken, the participants completed informed consent forms and were familiarized with the testing protocol. The study protocol was approved by the local ethics committees (SA clearance certificate no. M160532; GZ. 39/42/63 ex 2015/16).Participants performed one IWT while wearing a portable gas analyzer and four different accelerometers. Before the IWT, resting metabolic rate was measured during 15 min of supine resting. IWT was conducted on a 400 m outdoor running track and started at 3 km/h. Walking speed was paced by audio signals given in 10 m intervals and increased by 0.5 km/h every 50 m up to the maximum individual walking speed. Maximum walking speed was defined as the speed where participants were unable to walk the given pacer speed.Gas exchange data were continuously measured breath by breath by a portable gas analyzer in Graz (CORTEX METAMAX 3B, Cortex Biophysik GmbH, Germany) and Johannesburg (OXYCON Mobile, CareFusion GmbH, Hoechberg, Germany). Calibration of volume, O2, and CO2 gas sensors was performed prior to every test according to the manufacturer’s guidelines. For activity measurements, all participants were equipped with four accelerometers in total. Three different types of accelerometers were attached on the nondominant wrist, placed in a random order: ActiGraph wGT3X-BT (ActiGraph, Pensacola, FL), GENEActive (Activeinsights, Kimbolton, UK), Axivity AX3 (Axivity Ltd., Newcastle upon Tyne, UK). In addition, one accelerometer (ActiGraph wGT3X-BT) was placed on the left hip. Prior to the measurements, all accelerometers were initialized with a data sampling frequency of 100 Hz and a sampling range of ± 8 g.Gas exchange data were exported into Microsoft Excel files (Microsoft Corporation, Redmond, WA, USA) in 15 s epochs using the manufacturer’s software. Oxygen uptake (VO2) was converted to MET3.5 by using the conventional conversion factor (1 MET = 3.5 mL/kg×min) and to METind by using the individual resting metabolic equivalent (1 METind) defined as mean oxygen uptake from the last 10 min of the 15 min supine resting position. Based on the raw triaxial accelerations, the vector magnitude (expressed in milligravity (mg) units) was calculated using the Euclidian norm minus one (ENMO = √(ax2+ay2+az2)-1g) [21]. Therefore, the raw files from all devices were imported into R statistical software V3.1.2 (R Foundation for Statistical Computing, Vienna, Austria) by which the metric ENMO was calculated in 15 s epoch using the package GGIR V1.2-0. Processed files were then exported into Microsoft Excel. To determine the absolute accelerometer cut points at moderate intensity (3 METs), we performed an individual linear regression analysis between the ENMO and the oxygen uptake during the IWT, based on MET3.5 (Cp3.5) and METind (Cpind). Individual physiological threshold was defined as the first ventilatory threshold (VT1), using the ventilatory equivalent (VE/VO2) to determine VT1 as the minimum of VE/VO2 without an increase in VE/VCO2. The evaluation was carried out by two independent examiners using computer-supported linear regression analysis to increase objectivity. In case of disagreement, the results were discussed with a further examiner.Data analysis was performed using GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA). The Shapiro–Wilk test was used for confirmation of normality. For normally distributed data, independent t tests were used to assess differences between pregnant and nonpregnant women. If data were not normally distributed, Mann–Whitney U tests were applied. To determine the effect of different cut points (Cp3.5, Cpind, and VT1) and devices/placement (GENEActiv, Axivity, ActiGraph wrist and hip) on acceleration (ENMO), we applied a two-way repeated measures ANOVA (two within subject factors) with post hoc Tukey multiple comparison test in pregnant and nonpregnant women. If sphericity was violated, the Geisser–Greenhouse correction was used. Spearman correlation coefficient was applied to evaluate the relationship between walking speed and accelerations at VT1. Data are presented as means ± standard deviation (M ± SD). Statistical significance was set at p < 0.05.We analyzed 39 data sets of healthy pregnant (n = 24; 27.7 ± 4.6 yrs) and nonpregnant (n = 15; 24.3 ± 2.2 yrs) women. In total, eight tests were excluded from the analysis because of incomplete data sets (six in the pregnant and two in the nonpregnant group). The mean age, weight, and BMI were significantly higher in pregnant women (gestational age: 26 ± 7 weeks), but maximum walking speed (vmax), VO2max, speed at VT1 (vVT1), and METind were significantly lower compared to nonpregnant women. Absolute oxygen uptake at VT1 (VO2VT1) was not different (p = 0.07) between pregnant (0.9 ± 0.2 L/min) and nonpregnant (1.0 ± 0.2 L/min) women, but calculated conventional MET values at VT1 were significantly lower in pregnant compared to nonpregnant women (3.8 ± 0.7 vs. 4.9 ± 1.4 METs, p < 0.01). Determined 1 METind was significantly higher compared to the conventional 1 MET value (3.5 mL/kg×min) in pregnant and nonpregnant women (Table 1). Bland–Altman interobserver comparison of VT1 determination (VO2VT1 bias: 0.01 ± 0.08 L/min) revealed a high level of agreement for the analysis.In general, accelerometer cut points (ENMO) showed higher values and higher interindividual variability in pregnant compared to nonpregnant women (e.g., SD = 136 vs. SD = 50 for Cp3.5). In both groups, interindividual variability was less for the hip worn device (Table 2). Two-way repeated measures ANOVA in pregnant woman showed a significant effect for the determination of cut points (F (1.76, 141.30) = 51.27, p < 0.01) and a significant effect for devices/placement (F (3, 80) = 3.16, p < 0.05) and no significant effect of interaction (F (6, 160) = 0.35, p = 0.91). Post hoc multiple comparison revealed no difference between Cp3.5 and Cpind, but ENMO was significantly lower for Cp3.5 and Cpind compared to the ENMO at VT1 in pregnant women. Comparison of devices in pregnant women showed significantly higher ENMOs for wrist-worn GENEActiv compared to hip-worn ActiGraph (p < 0.05) and for wrist-worn ActiGraph compared to hip-worn ActiGraph and wrist-worn Axivity (p < 0.05). In nonpregnant women, there was a significant effect for the determination of cut points (F (1.28, 69.36) = 56.19, p < 0.01) and no significant effect for devices/placement and interaction (F (3, 54) = 0.42, p = 0.74; F (6, 108) = 0.38, p = 0.88). ENMO was significantly different for Cp3.5 and Cpind compared to the ENMO at VT1 and between Cp3.5 and Cpind (p < 0.01).Figure 1 shows the oxygen uptake expressed in MET3.5 and METind as well as the ENMO of all devices at comparable walking speeds of the IWT for pregnant and nonpregnant women. Walking speed and accelerations at VT1 were not significantly correlated in pregnant (rpreg) but significantly correlated in nonpregnant (rnon) women for GENEActiv (rpreg = 0.13/rnon = 0.57) and Axivity (rpreg = 0.30/rnon = 0.69). Both groups showed no significant correlation between vVT1 and ENMO at VT1 for wrist-worn ActiGraph (rpreg = 0.10/rnon = 0.41), but for hip-worn ActiGraph, values were significantly correlated (rpreg = 0.62/rnon = 0.69). Comparing ENMOs within all wrist-worn devices, GENEActiv and Axivity were similar in their measurements while wrist-worn ActiGraph showed higher mean accelerations with increasing speed. This difference was stronger in pregnant compared to nonpregnant women and at higher speeds above VT1. Walking speed at VT1 corresponded to 3.8 ± 0.7 and 4.9 ± 1.4 conventional METs in pregnant and nonpregnant women, respectively. Values ranged between 2.48 and 7.73 conventional METs and were lower at VT1 compared to the 3-MET absolute moderate intensity definition in three cases in pregnant and in one case in nonpregnant women.Accelerometer cut points at absolute moderate intensity definition (3 METs) were significantly lower compared to the intensity at VT1 in a short maximal incremental walking test. The underestimation of intensity compared to VT1 was independent of the accelerometer device or placement and the applied 1 MET value in pregnant and nonpregnant women. Walking speed at VT1 was 5.7 ± 0.5 and 6.2 ± 0.8 km/h, which corresponded to an oxygen uptake of 3.8 ± 0.7 and 4.9 ± 1.4 conventional METs in pregnant and nonpregnant women, respectively. Whether during pregnancy or not, a certain duration of moderate-intensity physical activity (MPA), vigorous-intensity physical activity (VPA), or a combination of them (MVPA) is recommended in order to gain specific health benefits. Application of PA recommendations using fixed absolute intensities (e.g., MPA: 3–6 METs) [22] may lead to insufficient health benefits in our group of pregnant and nonpregnant women. In our sample, activity according to fixed absolute moderate intensity may be not intense enough to provoke larger adaptions of the cardiorespiratory system since 3 METs were lower compared to the intensity at VT1 in all women except four. On the contrary, in individuals with lower fitness level, overloading or discouragement due to unattainable recommendations could be the result of recommendations based on absolute intensities [10]. Individually tailored metabolic or physiological accelerometer cut points were already shown to reduce this methodological error and to provide more meaningful results [19,20]. To determine individualized accelerometer cut points, a three-phase model [16] can be applied which allows one to detect the transition from phase 1 to phase 2 of energy supply, independent of the individual performance level, by using an individual threshold like VT1. Phase 1 is characterized by a metabolically inter- and intramuscular balanced situation. Activities within this phase can be maintained for several hours without becoming fatigued [23]. The metabolic situation in phase 2 is systemically balanced but activity duration is limited [16]. Deliberate activity in phase 1 or 2 will therefore cause specific adaptions on a local and systemic level and the exact definition of these phases enables precise prescription and interpretation of intensity [14,24]. Therefore, optimized health benefits are suspected when recommendations are based on individual metabolic thresholds (i.e., VT1 = lower limit for MPA), which are standard in performance development in structured training processes [25]. Furthermore, a more accurate assessment of the intensity of PA would enable better associations with health outcomes, dose–response relationships, and behavior surveillance [19,26].The accelerations at absolute cut points (Cp3.5, Cpind) were significantly lower compared to VT1. Mean values for Cp3.5 were higher in pregnant compared to nonpregnant women (e.g., Axivity: 165 ± 118 vs. 95 ± 43 mg). This difference of accelerations between groups was smaller for Cpind and VT1 values as well as for the hip-worn device (Table 2). Such interindividual variability has already been shown for intensity cut points relative to heart rate reserve in a group with heterogeneous cardiorespiratory fitness. A high-fit group had higher accelerometer counts at the same relative intensity compared to a low-fit group [12]. Clear influence on varying accelerometer cut points was also shown for age and overweight/obesity [27,28]. However, age and weight affect cardiorespiratory fitness, which tends to decrease with both age [29] and obesity [30]. In our study, pregnant women had significantly higher body mass and maximum performance capacity was significantly lower compared to nonpregnant women. However, accelerometer cut points were higher (for wrist-worn devices) within pregnant women but showed no correlation with the walking speed at VT1. Therefore, accelerometer values from wrist-worn devices could not be attributed to intensity in pregnant women. As walking economics were shown to change in pregnancy [31], higher accelerations in pregnant woman might more likely show differences in walking style. In nonpregnant women, high interindividual variability at the single cut points can be explained by differences in performance capacity, due to walking speeds at VT1 (higher speed implicates a higher performance capacity) being significantly correlated to cut points, except for the wrist-worn ActiGraph. Accelerations in wrist-worn devices generally varied between the constant speed increments of IWT and were in some cases generally higher from the start of IWT compared to average values. In contrast, the hip-worn device showed less variability and significant correlations between walking speed and accelerations at VT1 (rpreg = 0.62/rnon = 0.69) in pregnant and nonpregnant women, which is in line with a study by Ozemek et al. [12]. The hip-worn device seems to be less affected by the walking style and may therefore provide more meaningful results, especially in pregnant women. Mean ENMO accelerometer values in nonpregnant women for Cp3.5 were in line with recent findings from the literature, presenting similar values for wrist-worn GENEActive (93.2 mg) and ActiGraph (100.6 mg) and hip-worn ActiGraph (69.1 mg) devices [5].Mean speed at VT1 in our study (pregnant: 5.7 km·h−1 and nonpregnant: 6.2 km·h−1) is comparable to other studies, which determined a walking speed of ≈ 5 km·h−1 in older healthy men and women (56 ± 16 yrs) [32] and of 5.1 and 5.5 km/h in postmenopausal women [19,20]. Determination of VT1 in a walking test seems to be applicable in a wide range of populations. Furthermore, walking tests are highly practicable due to the short duration of the test (average duration 6.5 min) and the fact that VT1 can also be assessed using heart rate variability measurement of a simple heart rate monitor [32]. This enables testing several subjects at once in a short period of time requiring only heart rate monitors and pacing.In both groups, 1 METind was significantly higher compared to the conventional 1 MET value (3.5 mL/kg×min). Higher resting metabolic rates in pregnant women compared to the conventional 1 MET value are different compared to the literature, where no difference to the conventional 1 MET value was found [9]. This might be due to no acclimatization period and the relatively short measurement period of the resting state in our study. Higher resting metabolic rates of our sample of mainly active and well-trained women are in line with the recent literature, where energy expenditure of active healthy women was found to be underestimated by the conventional 1 MET [33]. Determination of the absolute accelerometer cut points at moderate intensity (3 METs) by the METind resulted in higher values compared to the conventional equivalent (MET3.5), but in significantly different values compared to VT1. Therefore, using the METind can partly compensate for the error made by using the fixed 1 MET definition, but not for the differences in performance capacity.However, this study is not without limitations. The protocol of the IWT, with 50 m speed increases, allows a time-efficient determination of the first threshold (average duration 6.5 min). Although small increases of 0.5 km/h per increment favor a fast adaption, 50 m is a relatively short distance to adjust to the pacing speed. However, with increasing speed, walking time of the single increments decreases, which might have increased variance at higher speeds. For the definition of accelerations at a constant speed, increments with longer and equal duration would provide more precise results. Prior to IWT, individual resting metabolic rates were defined from a 15 min supine rest position without any guidance regarding the fasting state. This is not according to the general practice, as usually subjects have to be overnight fasted, run through an acclimating period and a longer measurement period [9,33], which would not have been feasible in pregnant women. Because of this, the determined resting metabolic equivalent needs to be considered with caution. Nevertheless, our resting metabolic equivalent determination was sufficient to get performance parameters related to the subject’s actual metabolic status, considering their actual weight, age, and performance capacity. Furthermore, one could criticize that the tests were conducted by two research groups in different countries. Between groups, we used different gas analyzer devices, possibly influencing the results, although appropriate calibration was performed and devices were shown to provide reliable measurements with adequate validity for field-based measurements [34]. Furthermore, the majority of women in South Africa were black Africans, compared to Caucasian in the Austrian population, which was not considered in the analysis even though African Americans were shown to have lower accelerometer cut points compared to Caucasians in a maximal graded exercise treadmill test [2]. Assessing differences in thresholds, the role of race was considered negligible due to individual analysis. Nevertheless, this approach might have increased interindividual variability of accelerometer data. However, the generalizability of these findings is limited due to the small number of subjects. Future research should take these limitations into consideration and validate the findings in a larger, more representative sample.Intensity at absolute 3 MET accelerometer cut points was lower compared to the intensity at VT1 independent of the device or placement in pregnant and nonpregnant women. The application of the individual resting metabolic equivalent results in an approximation to the first ventilatory threshold but does not provide an alternative for individually tailored activity cut points. Using absolute accelerometer cut points, which are independent of the individual performance capacity, can lead to different physiological and metabolic strain at the same absolute intensity, possibly causing under- or overloading for a particular person. Therefore, individual thresholds based on physiological parameters, such as the VT1, are recommended to quantify the intensity of PA. A short incremental walking test can be a time-efficient method to define these thresholds and, thus, can be used for tailoring accelerometer cut points to individual differences in performance capacity.Conceptualization, M.N.M.v.P., P.D., and P.H.; Methodology, P.D., A.M., M.N.M.v.P., and P.H.; Software, M.C.S., J.J., and P.B.; Validation, P.B.; Formal analysis, M.C.S., J.J., and P.B.; investigation, P.D., A.M., E.D.W., G.M., and P.B.; Resources, P.H.; Data curation, G.M. and P.B.; writing—original draft preparation, P.B.; Writing—Review and editing, M.N.M.v.P., P.D., E.D.W., M.C.S., J.J., G.M., and P.H. All authors have read and agreed to the published version of the manuscript.This research received no external funding.“Open Access Funding by the University of Graz”.The authors declare no conflict of interest.Mean ± SD oxygen uptake (VO2) expressed in MET3.5 (1 MET = 3.5 mL/kg×min) and METind (1 MET = 3.7 ± 0.5/4.2 ± 0.6 mL/kg×min in pregnant/nonpregnant women) as well as the Euclidian norm minus one (ENMO=√(ax2+ay2+az2)-1g) from GENEActiv, Axivity, ActiGraph wrist and hip for each single load step of the incremental walking test (IWT). Arrows mark cut point values determined from MET3.5 (Cp3.5), METind (Cpind), and the first ventilatory threshold (VT1).Comparison of anthropometrics, performance of the incremental walking test (IWT), and the individual resting metabolic equivalent between pregnant and nonpregnant women.BMI = Body Mass Index, VO2max = maximum oxygen uptake, VO2VT1 = oxygen uptake at the first ventilatory threshold, vmax maximum walking speed, vVT1 = walking speed at the first ventilatory threshold, METind = calculated individual resting metabolic equivalent, METs = rates of energy expenditure (1 MET is equivalent to 3.5 mL·kg−1·min−1, results are shown as M ± SD. 1 significantly higher than the comparison group (p < 0.05). 2 significantly higher compared to conversion 1 MET = 3.5 mL/kg×min (p < 0.05).Metric Euclidian norm minus one (ENMO) of the determined accelerometer cut points of the different devices for pregnant and nonpregnant women.Cp3.5 = absolute accelerometer cut point at moderate intensity (3 MET) calculated using the conventional 1 MET = 3.5 m/kg×min value, Cpind = absolute accelerometer cut point at moderate intensity (3 MET) calculated using the individual resting metabolic equivalent, VT1 = first ventilatory threshold, mg = milligravity. 1 significantly different compared to Cp3.5 (p < 0.05). 2 significantly different compared to VT1 (p < 0.0001).

Med-MDPI/ijerph_5/ijerph-17-16-05652.txt

@@ -0,0 +1,18 @@
+The current significant human resource and workforce shortages of registered nurses (RNs) are impacting urban, suburban, and rural hospitals and healthcare facilities all over the globe, regardless of the entities’ economic and financial backgrounds. The purpose of this research study is to understand why non-traditional, returning, evening, and adult (NTREA) students decided to enrol at the Associate Degree in Nursing programme during their mid-adulthood? 40 s-career nursing students who are pursuing their nursing programme were invited to individual interview sessions and focus group activities on sharing and expressing the motivations in the New England region in the United States. Based on the theoretical framework of Social Cognitive Career Theory, the researcher concluded that family consideration and higher social status were two of the major themes. The study provided a blueprint for human resource professionals, health and social caring leaders, government agencies, policymakers, and researchers to reform their current nursing curriculum and health workforce policy to attract potential second-career nursing joining the nursing profession.The current significant human resource and workforce shortages of registered nurses (RNs) are impacting urban, suburban, and rural hospitals and healthcare facilities all over the globe, regardless of the entities’ economic and financial backgrounds [1]. In recent decades, the United States has invested a significant amount of financial aid and support both for secondary school graduates and for second-career changers and other non-traditional, returning, evening, and adult students (NTREAs) [2] who want to pursue career development in the field of nursing practice. Investments from the federal government, state governments, and county agencies have contributed to a potential increase in the number of non-traditional career pathways, and career development approaches in the nursing profession, as a large number of second-career individuals and career changers decide on nursing for their career development [3]. Second-career nursing professionals can bring to the medical field significant contributions in terms of industry experience and professional practices from their first careers [4], and much of that knowledge and those practices can be integrated into nursing management, administrative tasks, and interpersonal communication [5]. Therefore, the question facing the country is: In addition to fresh secondary school graduates, who will work in the health and medical care system? The answer to this question involves exploring new sources for nursing students and registered nurses and knowing the influences that motivate and encourage Americans to select nursing as a lifelong professional investment. This study focused on why non-traditional, returning, evening, and adult students [2] and second-career changers decide to choose nursing for their career pathway during the middle years of their adulthood. The profession of nursing focuses on the issues of health, social care, counselling, and education [5,6]. The natures of the health, medical, and nursing professions usually do not have significant overlaps with business organisations, but increasingly, many Americans have decided to leave their original career pathways in business and other professions and to become registered nurses. Nursing education programmes and professional training for potential registered nurses have one of the fastest-growing enrolments in colleges and universities. Many nursing students have asserted that becoming health, medical, and social care providers allowed them to satisfy their goals and internal egos [7]. Because many adults had had extensive professional work experience in other fields before they entered nursing programmes, most understand the relationships and issues associated with a second-career change to nursing [8]. Although the United States is facing a significant nursing shortage, many associate degree programmes in nursing require interviews, exam scores, and placement exams due to a large number of applicants and interested students [9]. Because many of the existing nursing programmes do not have an open-enrolment policy, many non-traditional, returning, evening, and adult students and second-career changers cannot pass the programme admission requirements [2,10]. Privacy safeguards prevent most colleges and universities from publishing students’ background information, but it is not hard to imagine that secondary school graduates and youths with an extensive general education background and recent training for university entry exams, such as the Scholastic Aptitude Test and American College Testing exams, take most of the enrolment slots in nursing programmes. Therefore, due to the difficulties of changing to a second career and the issues associated with general education and university entrance exam requirements, the non-traditional, returning, evening, and adult students and second-career changers have lower enrolments in nursing programmes than traditional-age students do [2,10]. Second-career nurses are looking for opportunities to make a difference in people’s lives and thus are willing to be full-time students and to agree to sacrifice their immediate earnings for their educational voyage [8,11,12]. In addition to people with no prior college or university educational background, many individuals with at least a bachelor’s degree in a field other than nursing have decided to enter nursing education programmes because of their career motivations, interests, goals, and educational achievements [13]. Unlike the case with some professional medical specialists, such as physical therapists, it is not uncommon for other members of the medical public, such as general physicians and nurses, to have access to and observe in clinical environments. Thus, many understand that the health, medical, and nursing professions are positive and supportive of their life goals and potential lifelong career investment. Unsurprisingly, many second-career nurses advocate that nursing is a life-long investment and career development commitment rather than a job [3,14]. Another previous study [8] explored the transition to nursing practice experiences of the nursing students (i.e., first- and second-career changing nursing students) in the United States. The mixed research results (i.e., 15 qualitative data information and 122 quantitative data information) indicated that stressors and coping, the prevalence of burnout and presenteeism, and difficulty describing nursing’s role were three of their reasons and sources of stress and burnout as first- and second- career-changing nursing students. In fact, due to the changing working environment, many individuals experienced different levels of problems [15,16,17,18,19]. Topics on career development, and particularly on second-career development during mid-adulthood, have not yet received adequate research attention and exploration [20]. However, without a doubt, second-career nurses must be one of the most important sources for filling the significant gaps in the current health, social care, and nursing professions internationally. That is true, particularly for nursing professionals with extensive professional skills and experience in the industry. Due to the rapid development of society’s social, cultural, and economic backgrounds, mid-life career changes are not uncommon [21,22,23,24]. However, ensuring individuals’ career and financial security after they complete a nursing programme is vital because most second-career people have family and financial responsibilities. Because of the demand for health, social care, and nursing professionals continues to increase, studies about second-career nurses are needed [25]. The purpose of this study was to use a phenomenological analysis [26] to explore the motivations for and reasons underlying the second-career changes by people in mid-adulthood who had decided to enrol in an Associate Degree in Nursing programme at a community college in the New England region in the United States. Because of the significant international human resource shortage of registered nurses, the results of this study will contribute to the field and should provide a blueprint for human resource professionals, healthcare administrators, university department heads, policymakers, educational researchers, nursing professionals, and government agencies in their efforts to reform and polish their current planning [9]. Encouraging people to change their career to a new one (i.e., in the health, social care, and nursing professions) is always a potential solution for filling the human resource gaps [27]. Therefore, expanding the understanding of career decisions and career beliefs always helps guide the management of educational and training programmes. Although some relevant quantitative information and data for different backgrounds and geographic regions already existed, this study will add important contemporary information about the backgrounds and interests of a unique group of participants undergoing mid-adulthood career changes to nursing. In short, one research question guided this research study, which was, Why did NTREA students decide to enroll in the Associate Degree in Nursing programme during their mid-adulthood?Based on the abovementioned elements and backgrounds, based on the lens of the theoretical framework (i.e., the Social Cognitive Career Theory), and the perspectives from the participants, the outcomes of this study tend to provide recommendations about human resources and workforce management in the field of health, social care and nursing profession. Nursing school leaders, government leaders, policymakers, human resource planners, and scholars to reform and polish the current education and human resource planning for the next decades. Second, many of the current colleges and universities academic programmes and courses are designed for full-time and day-time students with long-term commitments. However, NTREA students usually cannot enrol at any of the full-time programmes due to various responsibilities. Colleges and universities leaders should take this opportunity to reform their part-time and evening programmes for NTREA students. The current research study was guided by the Social Cognitive Career Theory (SCCT) [13,21,23,28,29,30]. The SCCT was developed based on the recommendations and elements from the ideas of Social Cognitive Theory by Bandura [31,32,33,34,35]. According to the guidelines of SCCT, the researchers advocated a learning model and cognitive behaviours [36] prompting people to understand, make sense, and establish their ideas. Figure 1 outlines the relationships. Three key points have been outlined, as the following, The formation and elaboration of occupational interests, goals, and motivation;The intention of selecting academic and occupational directions and choices;The performance, evaluation, and persistence in academic and occupational directions and intentions.The SCCT highlighted the impacts and relationship between cultural, social, and economic elements and influences on people’s self-knowledge, self-understanding, sense-making process, and opportunity outcomes and results [16,37,38,39]. The process and element can be interconnected. For example, individuals’ career decision can be influenced by one or more than one element. To illustrate for this research study, it is sometimes observed that NTREA [2] nursing students decided to study a college degree and professional licensure programmes due to financial consideration. However, as social impacts may influence individuals’ financial incomes and consideration, individuals may change their mind during their educational voyage. As a result, based on the guidelines of SCCT [13], the researcher would like to use this theory to understand why NTREA students during their mid-adulthood decided to study their Associate Degree in Nursing in the New England region in the United States.The researcher employed the qualitative research method [40,41,42] as the structure of this study. Unlike the quantitative research studies with surveys and questionnaires, qualitative researchers tend to understand the sharing, lived stories, personal background, feedback, and sense-making process of the individuals [43]. It is worth noting that qualitative data information is rich and in-depth as qualitative researchers need to access the participants and collect some Why and How [41] interview questions about participants’ background. Such studies usually gather in-depth background and sharing for the research question. The New England region in the United States was the research background of this study (i.e., Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, and Connecticut). First, unlike other small nations and countries, the United States is one of the largest nations internationally. Therefore, different groups and populations may have a different understanding, concepts, and values. As a qualitative research study, the researcher needed to select a targeted region for the research activity. In order to understand the holistic situations and issues of an American region, the researcher invited participants from each of the State within the New England region in the Northeast. Figure 2 showed the map of the New England Region. Forty (40) nursing students who are currently enrolled at an Associate Degree in Nursing were invited. All agreed to participate in this study. All the participants were NTREA students [2] and second career changers with previous working experiences after secondary school graduation. In other words, these participants were not traditional-age students who completed their secondary school qualification directly from secondary schools. All were American citizens or permanent residents and residents of one of the six states in the New England region in the United States. The researcher invited these participants with the purposive sampling strategy based on the personal network and connection. Please see Appendix A for the participants’ demography. The participants needed to meet the following criteria, which were:
+          
+            (1)
+            Never completed a postsecondary qualification;
+          
+          
+            (2)
+            Never work in the nursing profession;
+          
+          
+            (3)
+            Left school for at least ten years before this nursing programme;
+          
+          
+            (4)
+            Be a resident of one of the six states in the New England region for at least two years.
+          
+        Never completed a postsecondary qualification;Never work in the nursing profession;Left school for at least ten years before this nursing programme;Be a resident of one of the six states in the New England region for at least two years.The researcher, as an experienced education, public health, nursing, social caring, and psychology researcher in the North American and the Asian-Pacific regions, the researcher has established some networks within the fields. With the purposive sampling strategy, the researcher invited these participants based on personal network and referral from others. In order to invite these participants into the study, the researcher sent each potential participant an email invitation with the interview questions, nature of the study, risks, and the content form or agreement. As a result, 40 participants agreed on participation. After all of the participants agreed on the study, the researcher started to arrange the interactions and data collection procedure. Two types of tools were employed, including individual interview sections and focus group activities [44]. Please see Appendix B for the individual interview sessions questions, and Appendix C for the focus group questions. The individual interviews were conducted to explore participants’ understanding, lived stories, in-depth sharing, and sense-making process about their learning experience and career decision under a private sharing environment. The focus group activities were conducted to explore some similar background, social and cultural understanding, personal sharing, expectations, goals, interests, educational understanding, achievements, difficulties, and interests. Both research tools were useful in regard to listening to the voices and sharing by engaging the participants individually and collectively. As for the procedure of the data collection, please refer the Figure 3. All participants have the rights to participate or withdraw from the study at any time of the process. The general inductive approach was employed for the qualitative data collection and data analysis process. In order to seek useful data information for the research question, first, the researcher developed a set of interview protocol and interview questions. According to Seidman [45], privacy and confidentiality are important for lived stories sharing. Therefore, each individual interview section was hosted in a private room at public libraries, college classrooms, and community centres. No additional and third parties involved. Each interview lasted from 64 min to 109 min. Second, after each participant completed the individual interview sections, all participants were invited into eight focus groups activities via social media (i.e., WhatsApp Group Chat). Due to the location (i.e., six states within the New England region), not all participants could come to a targeted location for the focus group activities. Therefore, the online-based focus group activities overcame the issue. Five participants were invited to each focus group activity. In total, eight focus group activities were hosted. Each focus group activity lasted from 81 min to 112 min. After the researcher analysed the data information, the information was sent to the related participant for the member checking process. Due to the location issue, the researcher sent the related written transcripts to each participant via email. In addition, the researcher also invited each for a member checking interview for the purpose of confirmation. As for this member checking interview session via social media, the interview session lasted from 20 to 34 min. As a result, all participants agreed to their data information and approved their parts. The researcher first transcribed all the voice records and messages into written transcripts. After the researcher transcribed the oral and verbal data information from voices messages to written transcripts, (i.e., 40 interview sessions and eight sessions of focus group activities), the researcher collected more than 500 pages of written transcripts filled with rich, in-depth life stories and sharing from 40 s-career nursing students who were currently enrolled in one of the Associate Degree in Nursing programmes in the New England region in the United States. Based on the written transcripts, the researcher re-read the data information multiple times in order to figure out the connections and ideas between the participants. It is worth noting that the researcher merged the interview sessions and focus group activities sharing for each participant. Therefore, no particular categories for interview sessions and focus group activities. Some qualitative researchers [2,21,23,25,27,46,47,48,49,50] advocated that large-size data information should be narrowed down to meaningful themes and subthemes. Therefore, the researcher employed the general inductive approach [51] for data analysis. First, based on the recommendations from the general inductive approach [51], the researcher employed the open-coding technique for the first-level themes categories [41]. After the open-coding technique, the researcher categorised 28 themes and 32 subthemes. However, based on the recommendations, the themes and subthemes should be further narrowed down. Open-coding technique [41,42,52] is one of the useful tools for qualitative researchers to narrow down large-size data information into first-level themes and subthemes. Unlike quantitative research studies with statistics, the key to qualitative research studies is to narrow down the data information into meaningful themes. Therefore, the open-coding technique was appropriate. Second, however, after the first-level themes and subthemes were merged. The numbers of themes and subthemes (i.e., first-level themes) were large for reporting. In other words, it is not appropriate to report the findings with more than ten themes and ten subthemes. Therefore, in order to meet the standardised requirement for a qualitative research study, the researcher continued to employ the axial-coding technique for the second-level themes categories [17,21,22,37,39]. After the axial-coding technique, the researcher merged two themes and four subthemes for reporting [41]. For the data analysis procedure, please refer to Figure 4. All signed and unsigned content forms and agreements, personal contacts, audio-recording, written transcripts, computer, and related materials were locked in a password-protected cabinet. The researcher was the only person for any access. After the study was completed, the researcher immediately destroyed and deleted all the related materials for personal privacy. Due to the content forms and agreements between the researcher and the participants, the college, detailed address, age, skin colour, and race information would not be outlined. Each would be given a pseudonym [41]. All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the University Ethics Committee (2019–2020/Fall/Winter). Based on the research question of this study, the researcher categorised the data information into the following themes and subthemes for reporting. As a study for the nursing education and nursing workforce management, the researcher aims to understand why did NTREA students decide to enrol at the Associate Degree in Nursing programme during their mid-adulthood. In fact, a second career-changing decision during the mid-adulthood is not an easy step, particularly for individuals joining the vocational-oriented programme (i.e., nursing) which required a large part of internship and placement. Through this qualitative study, the researcher was able to conduct an inductive analysis of the data and establish themes and subthemes, answer the research question, and discuss the findings. Analysis of the data information yielded two themes (family considerations and personal goals/improved social status) and four subthemes. The findings discussed below are supported by verbal quotations from the transcripts. Table 1 refers to the themes and subthemes of this study. In contrast to fresh secondary school students, university graduates, and youths without family and financial considerations, most of this study’s participants had family responsibilities as parents and/or caregivers of their own parents [53]. Therefore, more than half of their sharing and life stories concerned how the career change to nursing might provide different positive outcomes for their family members. All participants expressed that their strongest intention was based on their goals for their family members and not on their own educational achievement and interests [13]. Although a few indicated that they wanted to join the nursing profession because of their childhood experience, most highlighted that their goals for their family, financial resources, work schedule, and positive influences on their children were their priority. The following sections elaborate on the participants’ rich and in-depth sharing and subthemes. The four subthemes included quoted sharing and opinions from all participants. In other words, all participants shared similar subtheme categories. The subthemes money and financial resources were shared more than 120 times in both interviews and focus group sessions. Therefore, second-career nursing students tended to be influenced by financial resources as their primary goal [13,23]. From their perspective as parents, all desired to make more money for their school-age children––for example, one believed that the salary from her nursing position could provide better education for her children, saying: I want my children to have a better education and life…after-school programmes with piano playing and sports cost money…working as a housewife doesn’t help…I needed to make extra earning[s] to support my kids…my family cannot just rely on my husband’s [re]sources as all family [desires]and programme fees are not cheap at all…Another participant also shared a similar idea, saying: Although I agreed on the idea about man should be the one who makes money for the family, I still want to take some responsibilities for money-making…I want to offer a better future for my children…although YMCA offers some cheap courses or after-school programmes, we still need money and tuition fees for those courses…I need to be a useful person to make money for my family…Although many indicated that the government did not require such after-school activities, all wished to provide a meaningful childhood to their kids. The researcher captured two participants sharing, said:…I lived in a poor urban community when I was young…I couldn’t experience any after-school sports life and activities due to my family’s financial problems…I wanted to learn piano but I couldn’t as my father did not have much extra money…I cannot allow this happen to my kids again 40 years later.Poor is not illegal…but without money, we cannot do anything…I wish my children can be useful people for our community and country…so I need to make money to support my children for additional learning…or life experiences…even reading clubs and Sunday schools…we still need to pay a fee for the community supporting fee…so I need to give them a meaningful childhood…Based on the sharing, many indicated that financial considerations and resources took important roles in their decision-making process. For example, more than three-fourths indicated that they would need to live with food stamps if they could not upgrade their skills, saying: The economy is very hard for high school graduates and unskilled people, like me…both of us [husband] are working as labour workers with the minimum wage…this is not a way to continue…at least I don’t want [my]children to [repeat]this … I need to go to school and be a useful person…this nursing programme [has] allowed me to upgrade myself…In addition to the participants with parental responsibilities, several single participants indicated that they sought a stable financial income and earning support for their living standard and spending needs for their parents [54,55,56,57]. Although the single participants did not have significant spending for their children, many still needed financial resources for food, rental fees, and other costs for their aged parents. People of colour and minorities mostly shared such expressions. Many indicated that working for a basic income could not satisfy their needs as a person in society, saying: I am not smart, but I can still work…I can work for many labour jobs…I want to seek a position…I spend a certain amount of time and receive the balance amount of money and salary. As I am working in [in-home] social caregiving, I want to work as a nurse for a better salary with similar responsibilitiesIt is true that without enough money, but just for my basic needs…I cannot be promoted or gained my personal enhancements…I want to improve myself as a useful person in the community…And I think being a nurse is a good way to go…so, after I asked my families recommendations…I want to study the nursing programme for a better life or changeIn short, many participants, particularly parents, had decided to go back to school, with their children and the financial health of their family as one of their strongest motivations for studying nursing for their career development. “I can take the overnight shift schedule, as I can spend the day-time with my children” (P#30, Supermarket Associate). It is worth noting that the idea of an “on-shift schedule” was mentioned 123 times, according to the study’s transcripts. Single and married participants had different types of responsibilities, from family work to personal development, that influenced their decision-making processes and career development endeavours during their transition to nursing during their mid-adulthood. Because of the social bias about traditional working hours (i.e., from 9 AM to 5 PM), the general public often neglects to appreciate the difficulties and flexibilities of working professionals who take the on-shift duties. Nursing and medical professionals, for example, usually need to work for the whole year without a holiday. However, the participants of this study asserted that they wanted to take an on-shift working schedule so they could manage their daily activities with their children and others [1]. For example, a participant reported that the on-shift schedule allowed him to do some banking activities and governmental work during regular office hours, saying: The traditional office hour positions do not allow me to go to the bank and government works…also, I cannot even go to interviews due to the schedule conflicts…the on-shift schedule allows me to manage my plan and do something I need to do besides jobs…Many parents indicated that they could spend much more time with their children if they took the overnight shift at the hospital. One shared an expectation about time management, saying: I can work at night once my sons [go to] sleep at night at 9 PM…I work at 11 PM in the hospital for the night shift…my husband takes care of my sons’ morning tasks, and I take care of the after-school activities. We can make sure there is an adult at home…Another parent also shared a similar experience in favour of the children’s advantages and the time arrangement, saying: I can send my children to the school on my way to the hospital…if I cannot do that, I can send them to the school bus if I work for the morning shift…I will be home by 4 PM and they finish their after-school programmes at 3:45 PM…[the] overnight shift could be easier, but at least I can work on it…In particular, most parents indicated that the overnight shift allowed them to handle their family responsibilities during the day-time and to work at night. Although a previous study explored that some nursing professionals disliked the on-shift schedule due to its time management conflicts, many participants in this study expressed positive feedback [58]. As a result, they did not need to stay at home as a housewife with limited incomes and earnings. In addition to the advantages of a balanced schedule for family and children’s concerns, a large group also maintained that the traditional office hours were not suitable for their outcome expectations [13,59]. Many indicated that in addition to the office hours, many positions in the service management and hospitality industry required an on-shift schedule due to the industry’s non-stop business operations. Such operations and management also happen in most of the health and social care professions. One participant in this study said: I enjoy the on-shift working timetable as we can enjoy different assignments and patients every day. Unlike my current job with the same tasks, [with the] different shift[s] at the hospital…morning, afternoon, and overnight…the tasks are different from day to day…and I also work with different co-workers and people…I like this type of working environment…In short, nursing is a challenging career that can require medical professionals to handle different patients and accidents daily, and the participants not only found that requirement to be satisfying, they also understood the operation and appreciated the on-shift scheduling. In addition to nursing’s financial considerations and scheduling management opportunities, many participants stated that the professional image and career development of nursing professionals increased their mental position with their children and family members. Without a doubt, family considerations were always placed as the priority of these groups of participants. One said: If I d[id] not have my children, I w[ould] not come back to school. I c[a]me back to school because I have to provide a better future for my kids and my family…I want to be a positive model for my children to learn…from my hardworking image…I need to tell my kids your mom is strong…and you should be strong too…Questions about how the expectation of presenting positive images for their children and family members influenced their career decisions were asked in both the individual and focus-group activity sections, so all shared their opinions on this issue. Several significant opinions are listed below [60,61,62,63]. For example, a participant believed that her own positive modelling could be a learning experience for her son, saying:I always believed modelling is a way of learning. I learned to be a strong woman from my mom. My mom worked very hard to take care of my sibling and my grandparents during her mid-age. I absorbed this idea from my mom…I need[ed] to be a strong mom after I married. My son is in high school now. I want to build up a strong image [for] my son…you[r] mom can be a very strong student and mom as well…Additional, similar sharing about becoming a positive image and learning model for one’s children were captured [60,61,62,63]. One participant believed that children should have a positive learning model during their young age in order to motivate their learning and development. That participant indicated that the parents should be such a learning model, saying: Both of my parents used their hands to raise my sibling and I up when they were young. I used to work as a housewife for a decade…I think I am a bit lazy if I compare to my mom’s experience. I told my children that their grandma always worked from 9AM to 9PM many years ago…but I don’t work…although it is a different situation and social-economic system…I need to be a positive image and model [for]my childrenBoth positive and negative images and behaviours always influenced the behaviours and development of children. In addition to the statements from parents with young children, three interesting sharing were captured from parents with 12th-grade students who were planning to go to university the next year. Those participants believed that their own modelling (i.e., going to community college as a student) could encourage their children’s mental development and social cognitive learning. For example, a participant shared that she intended to be a positive model at university for her children, saying: …I go to college during mid-age…my children will go to college soon too…so we go to college at the same time. I want to encourage my children to…pursue their dream and seek a college degree…not like me…just a high school graduate for more than three decadesAnother similar life story about attending college with her children was shared by a third participant. Interestingly, that participant was enrolled at the same community college with her son that semester as classmates in the same general education course. Both encouraged each other and shared materials and group project ideas. The mother said: I can study with my son…this is very rewarding…as a mom, I d[id]n’t know how can I enter the society and community of my children…[but] this college and education and degree [work] engage[s] us…we learn, we grow, and we are getting closer to each otherIn short, based on those findings, the researcher found that the ideas of modelling from both the SCCT [13,59] and Social Cognitive Theory [60,61,62,63] strongly influenced the career development and career decisions of these groups of study participants, because they tended to become learning models. To illustrate such benefits to both young children and traditional-age university students, the participants shared significant stories about their efforts to motivate their family members for a better future. With the supports of the previous literature and studies, the researcher could conclude the connections and relationships between the SCCT and the decide-making process of the participants. Unlike fresh secondary school graduates and other youths, second-career nurses usually have already acquired decades of working and life experience before joining the nursing education programme at the community colleges. Therefore, they have numerous responsibilities, personal backgrounds, considerations, comparisons, and concerns, from education to social biases and discrimination [64,65,66]. Based on the sharing from the study’s participants, many indicated that seeking a college degree would help them to achieve a higher social status that they then could promote as valued skilled professionals for their communities and country [67]. Although the United States federal government and state governments sometimes provide financial support for unemployed, unskilled, and low-income workers, many such individuals have asserted that they need to use their hands and labour to seek adequate resources to support their family. Such modelling of hard work can encourage children to seek and achieve a better future [22,30,68,69,70]. For example, one participant told that she intended to become a skilled worker for her community because she wished to invest her energy back into her hometown, saying: …many of my neighbours are college graduates, they always contribute their energy and spare time to the community…such as community centres, libraries, weekend reading groups and so on…but I am useless. Therefore, I need to be a skilled worker and contribute my energy back to my hometown…Many stated that they wanted to set their long-term career development pathway in the direction of health and social care professions, particularly the nursing profession, after completing their degree programme. For example, a participant admitted that during the previous decade she had spent almost all of her savings towards her nursing degree programme, saying: I am spending all my savings as I have to work as a full-time student without [an] additional job commitment. I have to focus on my study…I cannot work anymore. But I think it is worth it. I want to become a nurse, and this is my lifelong investment…I believe no pain, no gain…The statement “long-term and lifelong investment” was shared many times by the participants, with many advocating that their college degree could help them to achieve their goals as registered nurses in the future. For example, several participants shared that a college education was expensive but worth the investment. The knowledge, skills, abilities, and leadership gained from their group work, placements, vocational training, communication interactions, and professional knowledge were hard to obtain without the guidelines from a college education. The researcher captured two statements, stated: …college is a place for us to train up our brain…nursing education, as a vocational training programme…allowed us to learn the skills from placement, professors, and experienced nurses at both [the] hospital and classroom environment. Even if I don’t work as a nurse or retire from the hospital in the future, I still gain some valuable skills and knowledge as a lifelong investment…I used to plan to do some community services and Sunday school volunteering…but I think I should train up myself…as a useful person in the community…Sunday school is good…but nursing education further allowed me to upgrade myself as a nurse or useful medical professionals…for my country…It is worth noting that although operational skills (e.g., office tasks) and home economics (e.g., housework) are vital in many societies, most people have the desire to further improve themselves by receiving additional skills from college. For example, the researcher captured an interesting sharing from a participant with nearly two decades of experience as a security guard at a hospital, who said: I help the medical professionals to mark down the schedule…the ambulance operation…I help to move the patients…I guide the patients to different levels and floors…but I am the assistant as I am unskilled. I want to invest my soul, my skills, and my career development for the life-long changing. I believe working in a hospital is very meaningful to our community. I came to this nursing programme because I want to help my community for life-long investment…not only [for] me but my townIn conclusion, in accord with the guidelines of the SCCT [13,59], and the research question of this study, this study’s results demonstrate that financial considerations and personal goals play important roles in individuals’ decisions to switch their career pathway and commit to the nursing profession during mid-adulthood. The result of this study might reflect on two previous studies about how personal goals and financial consideration may influence their decision-making process. Although the participants were from different parts of the region and had various backgrounds and responsibilities, most believed that without a stable financial income, they could not continue their standard of living [21,39]. Also, specifically among parents, many maintained that gaining a favourable image (e.g., going back to school during mid-adulthood and becoming a registered nurse) could serve as a positive model for their family members. Such thinking echoed both the SCCT [13,59] and the Social Cognitive Theory’s [31,32,34,35,60,71] exposition and described how such elements or goals could impact individuals’ decision-making processes. The current study investigated why did NTREA students decide to enrol at the Associate Degree in nursing Programme during their mid-adulthood. By employing the lens of the Social Cognitive Career Theory (SCCT) as the theoretical framework, the researcher categorised family consideration and higher social status for the future as two of the main themes, reasons and motivation for their decision. According to the SCCT [13], individuals’ career decisions can be influenced by their personal goals, interests, and educational achievements. In accord with that theme, all of this study’s participants indicated that they always put their family members and children as their priority (i.e., as their personal goals) [21,22,72,73]. Reflecting to a previous study [20] about second-career engineers, the career change decisions reported in this study were without a doubt made due to the participants’ family issues, with one saying, “Many of my life decisions were made due to my family members, my spouse, my children, and my parents…mid-aged parents usually do not have many choices…I would save my choices to my children and family members” (P#26, Housewife). With the reflection of a previous study, the results indicated that parents from low-income families tended to spend their resources on their children for after-school programmes, even if they had limited financial resources [74]. With the reflection of in this study, all parents echoed the findings of the earlier research and indicated that the additional earnings from a nursing position could increase the educational level and life standard of their children and allow their children to access additional achievements and learning by, for example, after-school programmes.Although the study’s participants were raised in different families and various geographic areas in the New England region, similar opinions about family responsibilities and the connections to financial resources ran through their reports. The ideas might reflect on a previous study. The results indicated that financial considerations could play significant positions and roles for individuals in their choices for their career development and their educational direction [75]. For example, other studies have explored the relationship between job loss and employment, based on the SCCT [13,21]. In this study, the participants echoed the previous studies and expressed that their family needed additional resources for housing mortgage and bills and that their previous careers and positions couldn’t support them. Previous studies [21,22,23,29,30] indicated that the expectation of an outcome plays an important role in how individuals select their career development and how individuals undergo their decision-making processes. In this study, most participants believed that the promise of additional financial resources informed their motivation and decision-making processes for transitioning to nursing from their previous career pathway. The participants’ behaviours reflected and echoed the SCCT [59,69,76,77] factors and how individuals should be treated.In terms of working schedule management, many indicated that the nursing working schedule allowed them to manage their family responsibilities as parent and professional staff. With the reflection of a previous study [58], the sharing from the participants echoed the ideas about working schedule and working hours are one of their biggest concerns. Most indicated that keeping non-traditional office hours and schedules allowed them to manage their daily activities around their children’s responsibilities [21]. Many indicated that in addition to the office hours, many positions in the service management and hospitality industry required an on-shift schedule due to the industry’s non-stop business operations. Such operations and management also happen in most of the health and social care professions. A previous study [3] showed that nursing faculty members liked to take on unusual responsibilities, and indeed nursing and medical professions always provide new challenges and tasks with their daily operations, as one said, “I don’t want to work as an office worker, nursing professional…fit my personal schedule and family responsibilities…” (P#9, Housewife). With the reflection of many previous studies [21,37,70], almost all shared their appreciation that the schedule flexibilities would allow them to expand their interests in their potential nursing positions for both family and personal reasons. Those revelations reflected the SCCT [13,69,76,77] elements about outcome expectations for personal satisfaction in light of the workplace challenges and flexibilities of their daily tasks and activities.In the sharing about the model of their children and family members, a previous study indicated that many nurses tended to establish a positive image and model [60,61,62,63] as learning tools [78]. Indeed, only a few of this study’s participants were single people without children. Many considered that their career development and activities should be based on the development of their children and planning of family members. With the reflection of a previous study, family members and parents’ images and behaviours always influenced the behaviours and developments of children [79]. Furthermore, the image of strong and hardworking parents did not influence just a single participant’s decision, but the choices of a large number of people.In terms of a nursing career as their long-term career promotion and development, almost all participants indicated that career promotion is one of their significant considerations. Many previous studies have indicated that often, second-career nursing students in their mid-adulthood find it difficult to rearrange their time management, spend their financial savings, and balance their responsibilities because family and financial issues are vital, particularly for parents with additional responsibilities [8,20,22,80]. Therefore, enrolling in a nursing education programme must be planned and discussed for a variety of reasons and purposes. The results from this study echoed most of those previous studies, in that most participants maintained that their achievements in their nursing programme allowed them to register and become a registered nurse, which is one of the most important elements. With the reflection of a previous study, the results of this study indicated that many participants had the desire to become skilled professionals and receive the appropriate training [81] to gain both general skills (e.g., communication, interpersonal behaviours, foreign language, time management, and mathematics) and specific skills (e.g., social caring, patient management, bookkeeping, data collection and analysis, and hospital management) from their college degree programme in order to upgrade themselves as skilled professionals [82].Although this research study provided the linkage between the SCCT and the behaviours of mid-aged NTREA nursing students in the New England region in the United States, it has several limitations. First, a qualitative research methodology [41] with the application of the general inductive approach [51] can be limited to be collecting data from different individuals. Second, due to the limited numbers of participants, the current research study could only invite mid-aged 40 NTREA nursing students. Future research can expand the population to a larger-size research study in order to cover a wider understanding. Third, due to the size of the United States, the researcher could only target the New England region as the mean. However, the understanding, feedback and career perspectives of people, groups, individuals, and populations in different parts of the United States could be different. Therefore, future research studies can cover different parts of the nation for wider coverage. Fourth, the current research study only covered American citizens and permanent residents in the United States. However, as the United States always welcomed international students for different types of academic and vocational programmes, international students’ voices may be covered for future research studies. Fifth, due to the nature of this study, this study tended to focus on the career decision and career perspective of a small group of mid-aged NTREA nursing students in the New England region. However, the experiences, sense of belonging, and learning strategies are not examined. Therefore, future researchers should seek understanding, feedback, and perspectives about these elements in order to expand the overall performances and holistic pictures for a better environment. Sixth, the current study tended to focus on the problems and reasons why NTREA students decided to enrol the nursing education programme at one of the community colleges in the United States. Future research studies should expand the scopes to the barriers and problems they encounter during the programme and after graduation. This research study has contributed to research practices devoted to nursing education, career perspective and human resource trends in two directions. First, the shortages of health, medical, nursing, social caring, and counselling professionals can be a critical and significant element internationally. This research study and its related outcomes served as the blueprint for nursing school leaders, government leaders, policymakers, human resource planners, and scholars to reform and polish the current education and human resource planning for the next decades. Second, many of the current colleges and universities academic programmes and courses are designed for full-time and day-time students with long-term commitments. However, NTREA students usually cannot enrol at any of the full-time programmes due to various responsibilities. Colleges and universities leaders should take this opportunity to reform their part-time and evening programmes for NTREA students. This research is supported by Woosong University Academic Research Funding 2020.The authors declare no conflict of interest.Participants’ Demography.Individual Interview Session QuestionsWhy do you want to study the nursing programme at a community college?What made you want to exercise this career switching from current position to potential registered nurse? Why and How?Why would a community college become a tool for your education? Why not other alternative routes?How would you describe your motivations and reasons for this nursing programme?For the enrolment of this nursing programme, have you ever considered nursing as your career pathways? If so, may you please tell me more?As a non-traditional, returning, evening and adult (NTREA) nursing student, how would this status change and describe your educational experience?Would the NTREA student status positively or negatively impact and influence your educational experience, educational decision, career decision, and personal experience?Would your occupational interests, goals and motivations positively or negatively influence your educational experience, educational decision, career decision, and personal experience as NTREA nursing student?Any follow-up questions.Focus Group QuestionsAs NTREA nursing student at a community college, how would everyone describe your experiences? Please share, and we can see there are there any core similarities and differences.What are the main motivations and reasons for everyone to join the nursing programme and the nursing profession after the degree? May you please describe your understanding?Would your personal goals, educational background and previous experiences influence your decision?Do you think will this nursing programme help you to achieve your personal goals and expectations? Why and how?Do you think would your background as NTREA students positively or negatively influence any of your experiences? Why and how? For any directions.Any follow-up questions.The relationships of the elements of the Social Cognitive Career Theory.The New England region.The procedure of the data collection.The procedure of the data analysis.Themes and subthemes.

Med-MDPI/ijerph_5/ijerph-17-16-05653.txt

@@ -0,0 +1,640 @@
+Contemporary social and health care services exhibit a significant movement toward increasing client involvement in their own care and in the development of services. This major cultural change represents a marked shift in the client’s role from a passive patient to an active empowered agent. We draw on interaction-oriented focus group research and conversation analysis to study workshop conversations in which social and health care clients and professionals discussed “client involvement”. Our analysis focuses on the participants’ mutually congruent or discrepant views on the topic. The professionals and clients both saw client involvement as an ideal that should be promoted. Although both participant groups considered the clients’ experience of being heard a prerequisite of client involvement, the clients deviated from the professionals in that they also highlighted the need for actual decision-making power. However, when the professionals invoked the clients’ responsibility for their own treatment, the clients were not eager to agree with their view. In addition, in analyzing problems of client involvement during the clients’ and professionals’ meta-talk about client involvement, the paper also shows how the “client involvement” rhetoric itself may, paradoxically, sometimes serve to hinder here-and-now client involvement.Cultural change is an opportunity. A culture refers to “a shared set of ideas, norms, and behaviours common to a group of people inhabiting a geographic location” [1], and cultural change makes it possible for people to remove those social and cultural deficits that have led to a repression of certain parts of the population [2]. Cultural change is nonetheless always a complex, multifaceted phenomenon. It is inherently threatening and psychologically stressful in that it introduces more variation to the basic assumptions that underlie people’s actions [3]. Cultural change is associated with various contradictions, such as those between values and practices [4,5], which in turn may be caused by society changing more rapidly than specific organizations and institutions [6]. Allowing and creating space for negotiation has thus been promoted as a significant way to deal with the challenging situations of cultural change [5].One major cultural change that has recently taken place in the context of social and health care services is related to the involvement of clients in their own care, as well as in the planning, development, and evaluation of services. The notion of client involvement entails the client’s right to be informed about issues relating to them, the opportunity to express their opinion when decisions about their own care are made, and participation in the planning, evaluation, and provision of services [7]. As any other cultural change, this shift is accompanied by enthusiasm, resistance, diversification of opinion, and a need for negotiation [8].Over the past decades, two contemporaneous trends have led toward increasing client involvement: the first concerns involving clients in their own care in social and health care services, and the second involving clients in planning and developing services. Traditionally, professionals have made decisions on the basis of their medical knowledge, relying on what they deem best for the client without really involving them in the decision-making process [9]. In this “paternalistic care philosophy” the role of the client has mainly been restricted to expressing their agreement with the professional’s decision [10]. The “consumerist movement” sought to increase clients’ opportunities to decide what services and treatments were most suitable for them [11]. The role of a professional became limited to providing the kind of medical information that a client would not have access to without the professional’s specialized expertise [10]. More recently, client-centeredness has become a key guiding paradigm in social and health care services. Its core idea is to elicit and understand clients’ needs, concerns, and expectations in order to reach a shared understanding of the problem and its treatment [12]. Even today, the client-centered care philosophy emphasizes an equal, collaborative partnership between a professional and a client, thus representing a marked shift from the traditional asymmetric doctor–patient relationship that involved a passive patient and a dominant clinician [13].Along with gaining power to influence their own care, clients are increasingly encouraged to contribute to the planning, evaluation, and development of the services they use [14]. This is part of the larger development of involving citizens in public policymaking, discussed broadly in, for instance, the fields of service management [15,16,17] and public administration [18,19,20]. Typically, client input has been elicited in a fairly restricted manner, inviting their reactions to specific services in a form of structured feedback survey or a client questionnaire [21]. Nowadays, clients are given a more active role in quality improvement when they are invited to co-develop services in collaboration with professionals. The main idea in co-development is the creation of value through interaction between service providers and users [22]. This trend has created a more equal relationship between clients and professionals as it strives for a genuine dialogue between participants [23].In addition to leading to more accessible and acceptable provision of services [14], client involvement has been seen as a normative good that is valuable in itself [24]. It is argued to improve democracy and social inclusion by placing clients at the heart of service delivery [24]. From the client’s viewpoint, an equal collaborative partnership between clients and professionals, the maintenance of trust, participation in knowledge production, and shared decision-making are crucial building blocks of involvement [9,25,26]. Clients seem to wish greater involvement in service delivery but they also want professionals to recognize this wish as optional and as varying according to the context, time, and individual situation [9,26]. The professionals, in turn, have been noted as valuing client involvement as such but to be reluctant to adopt it as a guiding clinical practice [27]. In their view, client involvement may be ineffective and too time consuming, and they are concerned that attention will be directed away from “actual client-work” [24,27,28]. In addition, some studies have reported that professionals feel intimidated by the new power relations: greater client empowerment may be experienced as threatening professional boundaries and competencies [27,29].As noted above, client involvement in their own care and development of services has been studied from multiple perspectives, considering both clients’ and professionals’ views [9,24,26,27]. What has been investigated less is the mutually congruent or discrepant views that the professionals and clients may have on the topic. This would be important to study as the prior research has pointed to the direction that the professionals and clients’ expectations on client involvement may vary [30,31,32]. In this paper, our aim is to investigate the implicit expectations that professionals and clients express when they talk about client involvement. Our aims are:To assess the degree to which social and health care professionals and clients share or differ in their expectations of client involvement.To analyze in detail the content of the social and health care professionals’ and clients expectations, paying spcific attention to where the two participant groups differ.To assess the degree to which social and health care professionals and clients share or differ in their expectations of client involvement.To analyze in detail the content of the social and health care professionals’ and clients expectations, paying spcific attention to where the two participant groups differ.We assume that the results can increase the overall understanding of the role of clients in social and health care services and thus ultimately help us evaluate the potential of client involvement in developing and providing good quality services.In this paper, we use a combination of methods deriving, on the one hand, from interaction-oriented focus group research [33,34,35,36,37,38,39], and on the other hand, from conversation analysis [40,41,42,43]. This means that we operate at the intersection of the substance of conversation and its interactional dynamics, linking our analysis to both the content of the group members’ utterances and the patterns of interaction that they create, see, e.g., [44].Previous research on focus groups has shown that, in addition to analyzing the content of the group members’ talk, the researcher may also observe how members of the group interact with one another and use these observations as part of the analysis [33,45,46]. Such observations can help the researcher, for example, “to explore the arguments people use against each other, identify the factors which influence individuals to change their minds and document how facts and stories operate in practice—what ideological work they do” [47] (p. 117). As Morgan [45] (p. 718) has pointed out, there is an “inherent connection between the substantive content of ‘what’ a person says and the interactive dynamics of ‘how’ he or she says those things.” The consideration of this connection is elementary when the topic of “client participation” is discussed in a conversation between professionals and the very clients whose participation is at stake at the level of conversation.In practice, our analysis we examined those segments of interaction where “client participation” was topicalized and discussed. The investigation was guided by the following three questions:What are the views that immediately mobilize an assertion of consensus among the participants [47] (p. 109)?What are the views that are preceded and followed by explanations and accounts, which demonstrate a need to justify one’s views in front of the other participants (see e.g., [48])?Are some views received with explicit expressions of resistance and moral contempt or implicit expressions of opposition through, for example, silence [47] (p. 110); [49] (p. 172); [50]? What are these views substantially about?What are the views that immediately mobilize an assertion of consensus among the participants [47] (p. 109)?What are the views that are preceded and followed by explanations and accounts, which demonstrate a need to justify one’s views in front of the other participants (see e.g., [48])?Are some views received with explicit expressions of resistance and moral contempt or implicit expressions of opposition through, for example, silence [47] (p. 110); [49] (p. 172); [50]? What are these views substantially about?On the basis of these considerations, each segment was analyzed with reference to its level of congruency/discrepancy and the contents and implicit expectations that the participants in each case oriented to (for more details, see Section 2.5).Our data consist of interaction among social and health care professionals and clients in co-development workshops. These workshops were part of the “Social and health care professionals as experts on client involvement” project of the Finnish Institute of Occupational Health. The project involves municipal social and health care organizations and aims to promote work practices that enhance clients’ involvement in their own care, as well as in planning and developing services. As part of the project, six regionally comprehensive client-involvement workshops were held in five different social and health care organizations to develop their organizational work practices. The aim of the workshops was to create a shared view of client involvement, identify what needs to be improved, invent small experiments to change work practices, and evaluate these experiments. The workshops were based on expansive learning theory [51], the change-management workshop method [52], and service design. In this study, the data came from the first two workshop processes conducted in two large, municipal social and health care organizations. In the first organization, the workshop process targeted client involvement among clients with mental health problems and substance abuse. In this organization, the workshop meetings were audio-taped. In the second organization, the process focused on first-contact services for elderly clients. This process was video-recorded using one camera located in the corner of the room, and was also audio-taped.The data thus consisted of audio and video recordings of four three-hour workshops (12 h of interaction). The workshop meetings were organized around group discussion assignments on client involvement. These assignments involved, for instance, defining “client involvement,” creating a map of how client involvement has developed in the organization, and assessing stories of smooth and challenging customer journeys. The workshop participants were divided into small groups of four to five, sitting at round tables. The facilitators initiated the discussions on the assignments by giving instructions. The small groups discussed the assignment freely and made notes, after which each group shared the main point of their discussions with the whole group. Thus, the workshop discussions were relatively loosely structured, and the participants were able to choose how and how much to contribute to the discussions.The data consist of 35 different participants. Each of the four workshops had approximately 15 participants: eight to 12 professionals, two to four clients, and two to three facilitators. As the workshops were primary organized to develop organizational work practices, the participants were recruited within the organizations without any research-based inclusion or exclusion criteria. The 25 professionals participating in the workshops were chosen by the managers of the organization (in collaboration with the professionals) to represent different occupational groups working with the client group in question. These occupational groups included nurses (n = 7), service advisors (n = 8), social workers (n = 2), physiotherapists (n = 2), development specialists (n = 3), and department managers (n = 3). Most of the professionals (n = 24) were females and only one was a male (a nurse). We do not have the information of their ages and levels of experience but in general they represented the whole spectrum, from young to more experienced professionals. The six clients participating in the workshops had either an ongoing treatment at the organization, or had previously been treated there. Three of the clients were male and three were female. We do not have specific information of their ages, diagnoses or other backgrounds apart from information they told in the workshops. The clients were recruited by the professionals and many of them had already participated in the development of services in some way or another. Some of the clients had also acquired training in expertise of experience by a third sector organization and thus gained a more official role in the organizational development activities. The four facilitators were all females, with a background in social, educational, and health sciences, and had a vast experience in facilitating the organizational development processes. They worked at the Finnish Institute of Occupational Health and the National Institute of Health and Welfare.The study was conducted in accordance with the Declaration of Helsinki, and permission to collect the data was obtained from the health care districts and the Finnish Institute of Occupational Health’s Ethics Committee (23 November 2018, project 3517803). Informed, written consent was obtained from all participants before they participated in the study, and they were advised that they could withdraw their consent at any point during the data collection. All names and other details that could enable identification of the participants have been altered in the text and data excerpts.Our interactional data from the co-development workshops were analyzed with methods of conversation analysis [40,41,42,43] and interaction-oriented focus group research [33,34,35,36,37,38,39]. We began our analytical process by watching and listening several times to the recordings, making notes on the segments during which the topic of “client involvement” was discussed. Although the workshop assignments revolved around this very topic, there was a lot of discussion on other related topics as well, such as multiprofessional collaboration. This study, however, is based only on the collection of those segments of interaction where client involvement was the participants’ main topic (n = 108). In order to warrant a more detailed analysis of these segments, they were transcribed using the conventions of conversation analysis, which necessitates focus not only on talk, but also on the ways in which the participants’ turns are received by the co-participants on a moment-to-moment basis, whether turn transitions are accompanied by overlap or silence, and whether the participants engage in salient nonverbal behavior in terms of gaze direction, gestures and facial expressions, see [41] (pp. 265–269) and the Appendix A. Thereafter, we started to work with the data-segment collection in a data-driven way, probing the categories and patterns identified in a single data segment against every new segment of data. Later, we tested our intersubjective grasp of these patterns with three analysts’ (E.W., S.K., and L-L.U.) independent coding of pieces of data, which led to several further specifications into categories that we had jointly agreed upon and that we could reliably identify in the data. By focusing on the participants’ ways of receiving and responding to each other’s views in the group, we classified each segment as exhibiting either congruence or discrepancy, and then compared the specific contents of each segment, paying specific attention to the implicit expectations that the participants oriented to between the two participant groups. In so doing, we also identified tensions between the participants’ views on client involvement and the clients’ opportunities to influence the co-development workshop discussion in the here and now. The data extracts presented in this paper are drawn from across our entire data set on the basis of their capacity to demonstrate the between-group differences on which our analysis focuses.In the following, we present the results of our qualitative analysis in three sections, each of which focuses on one specific topic in our participants’ talk about client involvement, which arose inductively from our analysis of the empirical data. First, we examine how the members of our workshops discussed the ideal of promoting client involvement, demonstrating the high level of consensus that existed among our participants with regard to the topic. Second, we consider an issue that the participants oriented to as more conflicting: is it enough that clients are heard or should they also have actual power to influence the decisions made regarding their services? Finally, we examine the differences in how clients and professionals oriented to the complex interwovenness of cooperation, authority, and responsibility.Our analysis thus centers around the question of whether the participants express mutually congruent or discrepant viewpoints when discussing these three topics. To give the reader an overall grasp of the prevalence of these patterns across our entire data set, Table 1 summarizes the numbers of instances of congruence and discrepancy with reference to the three topics. The four columns of the table show these numbers both (1) within the groups of professionals only and (2) within the groups consisting of both professionals and clients.In our workshop data, both the clients and the professionals agreed that, on a general level, client involvement should be promoted by giving the clients more opportunities to influence the development of their services. This topic was often discussed as a future ideal, the main obstacle for its current realization being the health care system. The “system” was seen as a common enemy that neither the clients nor the professionals had power to influence. This will be demonstrated in Extract 1, in which one of the professionals (P1) suggests that client involvement necessitates trust between clients and professionals, but that the professionals have not been given time to build trusting relations with clients, as their employer demands them to have a high number of appointments per day. The excerpt is taken from a small group discussion during the first workshop, in which the participants—consisting of both professionals and clients—are given the task to discuss how they understand the concept of client involvement.Extract 1.
+          
+            
+              01 P1:
+              mikä tohon asiakkaan osallistamiseen
+              
+                what is needed to make the client involved
+              
+            
+            
+              02  
+              tarvitaan niin tota jos on ikääntyneist kyse ni se
+              
+                I mean erm if they are elderly people then
+              
+            
+            
+              03  
+              et saa sen semmosen luottamuksen siihen ni se vaatii
+              
+                gaining that trust requires sufficient
+              
+            
+            
+              04  
+              sen ajan et jos se niinku työnantaja sit taas sanoo
+              
+                time but if the employer then says that
+              
+            
+            
+              05  
+              et pitää olla käyntejä niin ja niin paljo ni sit
+              
+                you should have so and so many appointments
+              
+            
+            
+              06  
+              toisaalta et se et saa sen luottamuksen asiakkaaseen
+              
+                and on the other hand that trust is needed to
+              
+            
+            
+              07  
+              ja saa hänet osallistuu ni tota se tarvitsee kyl sen
+              
+                get her involved then it really requires a
+              
+            
+            
+              08  
+              tietyn ajan.=
+              certain amount of time.=
+            
+            
+              09 P2:
+              =kyllä.
+              =yes.
+            
+            
+              10 C1:
+              ja tähän liittyy myös se et tuota sitte tää
+              
+                and this is also connected to the point that the
+              
+            
+            
+              11  
+              palveluntuottaja tai kotipalvelussa henkilö ei vaihdu
+              
+                service provider or homecare personnel doesn’t change
+              
+            
+            
+              12  
+              tai että lääkäri ei vaihdu tai et (.) sais niinku
+              
+                or that a doctor doesn’t change or that (.) so that
+              
+            
+            
+              13  
+              tniitten määrättyjen tuttujen [ihmisten kanssa asioida.
+              you could deal with the [same familiar people.
+            
+            
+              14 P2:
+              [samat työntekijät.
+              [the same employees.
+            
+          
+        mikä tohon asiakkaan osallistamiseen
+                what is needed to make the client involved
+              tarvitaan niin tota jos on ikääntyneist kyse ni se
+                I mean erm if they are elderly people then
+              et saa sen semmosen luottamuksen siihen ni se vaatii
+                gaining that trust requires sufficient
+              sen ajan et jos se niinku työnantaja sit taas sanoo
+                time but if the employer then says that
+              et pitää olla käyntejä niin ja niin paljo ni sit
+                you should have so and so many appointments
+              toisaalta et se et saa sen luottamuksen asiakkaaseen
+                and on the other hand that trust is needed to
+              ja saa hänet osallistuu ni tota se tarvitsee kyl sen
+                get her involved then it really requires a
+              tietyn ajan.=certain amount of time.==kyllä.=yes.ja tähän liittyy myös se et tuota sitte tää
+                and this is also connected to the point that the
+              palveluntuottaja tai kotipalvelussa henkilö ei vaihdu
+                service provider or homecare personnel doesn’t change
+              tai että lääkäri ei vaihdu tai et (.) sais niinku
+                or that a doctor doesn’t change or that (.) so that
+              tniitten määrättyjen tuttujen [ihmisten kanssa asioida.you could deal with the [same familiar people.[samat työntekijät.[the same employees.In the first lines (1–7), a professional (P1) elaborates on her view on client involvement. She suggests that, in order to get the client involved, a certain amount of trust between the client and the professional is needed. P1 also states that building a trusting relationship necessitates time, which she does not necessarily have. Immediately after her turn, another professional (P2) shows agreement with her view (line 9). Then, a client (C1) takes a turn, which she constructs as a straight continuation of P1′s turn (note the turn-initial connector ja “and” in line 10). In C1′s view, the question of trust is further associated with the issue of constantly changing service providers. P2 responds in overlap, suggesting candidate words for her co-participant’s turn completion (line 14). This type of anticipatory co-completion has shown to demonstrate understanding [53] and strong agreement between participants [54].In sum, both the professionals and clients agreed on the line of action that described obstacles and concerns about the realization of client involvement. They perceived these obstacles as being related to organizational factors that they had no power to influence, such as excessive workload (lines 4–8) and the permanency of the staff (lines 10–14). Thus, when considering client involvement as a future ideal whose realization was out of their hands, the views of the professional and client members of the workshop were mutually congruent.Listening to the client is commonly considered a critical component of all aspects of social and health care services, e.g., [55]. Understanding the client’s situation and lifeworld relies on the professional’s capacity to listen to the client’s experiences and to respond to what they hear [56]. Being heard during consultations is also something clients seem to desire more than anything else [57,58]. In our workshop discussions, both the clients and the professionals considered the professional listening to the client’s questions and concerns a prerequisite of client involvement. This topic was associated with a relatively high level of apparent consensus. Yet, content-wise, the professionals and clients emphasized slightly different ideas, which points to a subtle discrepancy between the views of these two participant groups. Below, we first show an example how this topic was discussed among the professionals only and then an example of the discussion among both professionals and clients.When conceptualizing client involvement, the professionals stressed the client’s need to be heard and understood. This pattern is exemplified by Extract 2, which starts by one of the professionals (P1) initiating talk about the question “what is client involvement?” asked by the workshop facilitators, and inviting other small-group members to talk about it.Extract 2.
+          
+            
+              01 P1:
+              mutta se että miten niinku (0.5) sitte se
+              
+                but then how about like (0.5)
+              
+            
+            
+              02  
+              asiakasosallisuus niin,
+              client involvement then,
+            
+            
+              03 P2:
+              yks on ihan se et miten tulee kuulluks.
+              
+                one thing is how you are heard.
+              
+            
+            
+              04 P1:
+              niin ja ymmärretyks sen [tarpeen kanssa,
+              yes and understood in terms of that [need,
+            
+            
+              05 P3:
+                          [mm.
+            
+            
+              06 P2:
+                          [nii,
+                          [yes,
+            
+            
+              07 P1:
+              että ku ensin on se tarve.
+              
+                when first there is that need.
+              
+            
+            
+              08 P2:
+              mmm,
+            
+            
+              09 P3:
+              nni on
+              
+                that’s right.
+              
+            
+          
+        mutta se että miten niinku (0.5) sitte se
+                but then how about like (0.5)
+              asiakasosallisuus niin,client involvement then,yks on ihan se et miten tulee kuulluks.
+                one thing is how you are heard.
+              niin ja ymmärretyks sen [tarpeen kanssa,yes and understood in terms of that [need,            [mm.            [nii,            [yes,että ku ensin on se tarve.
+                when first there is that need.
+              mmm,nni on
+                that’s right.
+              In lines 1–2, one of the professionals (P1) refers to the assignment (What is client involvement?) that the small group is supposed to discuss. She leaves the sentence unfinished (note the turn-final particle nii “then”), thus encouraging the other group members to present their views. As a response, another professional (P2) states that one aspect of client involvement is that the client is heard. P1 immediately responds with the particle nii “yes,” claiming agreement with the position presented by P2 [59]. P1 also extends P2′s turn by adding another element, the client’s need to “be understood,” to the basic idea. These types of extensions that grammatically complete the previous sentence have shown to display strong mutual engagement and shared understanding of the matter at hand [60]. At this point, both P3 (line 5) and P2 (line 6) produce minimal responses, thus demonstrating their agreement with P1′s view. P1 continues by highlighting that it is this client’s need that the professionals should understand (line 7). Again, both P2 and P3 display agreement with the view (lines 8–9). Thus, there seems to be a strong consensus among the professionals that listening to the client and providing them the experience of being heard is what essentially constitutes client involvement.The clients, however, raised the possibility that being heard is not the same thing as having actual power to influence the decisions about social and health care services. In this way, the clients invoked the question of an equal (or unequal) relationship between the professional and the client. This is what happens in Extract 3, in which the workshop participants are writing their views on post-it notes and choosing pictures that symbolize client involvement.Extract 3.
+          
+            
+              01 P1:
+              voisko lapset kuvata sitä asiaa
+              
+                could children illustrate a situation in which
+              
+            
+            
+              02  
+              et tulee kuulluks ja nähdyks.
+              
+                person is heard and seen.
+              
+            
+            
+              03 C1:
+              onks se nyt jos tulee kuulluks ni onks se
+              
+                is it then if someone is heard is it then
+              
+            
+            
+              04  
+              sama asia ku vaikuttaminen
+              
+                the same thing as influencing
+              
+            
+            
+              05  
+              et saa vaikuttaa jollain tavalla.
+              
+                that one can influence somehow.
+              
+            
+            
+              06 P2:
+              joo kyl se niinku joo-o,
+              
+                yeah I think uh yeah,
+              
+            
+            
+              07 P1:
+              mun mielest kyllä mut sä voit käyttää myös sitä sanaa.
+              
+                think yes but you can use that word too.
+              
+            
+            
+              08 C1:
+              se osallisuus no nii (.) kohdatuksi samanarvoisena.
+              
+                that involvement yes (.) to be considered equal.
+              
+            
+            
+              09 P1:
+              kaks viiva kolme kuvakorttii (.) meil on koht
+              
+                two to three pictures (.) we’ve used
+              
+            
+            
+              10  
+              kaikki otettu käyttöön.
+              
+                almost all of them.
+              
+            
+          
+        voisko lapset kuvata sitä asiaa
+                could children illustrate a situation in which
+              et tulee kuulluks ja nähdyks.
+                person is heard and seen.
+              onks se nyt jos tulee kuulluks ni onks se
+                is it then if someone is heard is it then
+              sama asia ku vaikuttaminen
+                the same thing as influencing
+              et saa vaikuttaa jollain tavalla.
+                that one can influence somehow.
+              joo kyl se niinku joo-o,
+                yeah I think uh yeah,
+              mun mielest kyllä mut sä voit käyttää myös sitä sanaa.
+                think yes but you can use that word too.
+              se osallisuus no nii (.) kohdatuksi samanarvoisena.
+                that involvement yes (.) to be considered equal.
+              kaks viiva kolme kuvakorttii (.) meil on koht
+                two to three pictures (.) we’ve used
+              kaikki otettu käyttöön.
+                almost all of them.
+              At the beginning of the extract, one of the professionals (P1) proposes a picture with a child on it and suggests that a child could illustrate the experience of being heard (lines 1–2). At that point, a client (C1) takes a turn but, instead of confirming P1′s proposal, he goes back to the professionals’ initial perception that highlighted the importance of being heard and questions if being heard is the same thing as being able to influence things (lines 3–5). The client’s challenge to the professional’s view is implicit in that it is presented in the form of a question, but—importantly—the client still raises the possibility that these two aspects of client participation may not always go hand in hand, which calls into question the emphasis on the professional view. In response to the client’s question, P2 produces a hesitant answer, which action-wise serves as a confirmation that “being heard” and “influencing” could essentially be perceived as the same thing. After this, P1 takes an even stronger position, claiming that, in her view, these two aspects of client participation are the same (line 7). She also concludes by stating that the client can also use the word “influence” (which most likely refers to the participants’ task of writing down their views on a post-it note). By designing her turn as a permission-like “commissive” (see the modal verb voida “can,” sä voit käyttää “you can use”; [61]) P1 positions herself as someone who has the right to direct the client’s actions in a workshop. After P1′s “permission,” the client suggests that involvement could mean that the client is considered equal (line 9). The professionals do not respond to this client’s suggestion but continue with the agenda of the workshop task.As demonstrated in Extract 3, the clients displayed an orientation to the expectation of what their role should be, not only that they would be heard with respect to their medical conditions and troubles, but that they would be considered equal to the professionals. Indeed, the notion of being heard is inherently asymmetrical in that it applies only to the clients, portraying them in a somewhat passive position in that their involvement is dependent on the professionals’ ability to understand their situation. What was at stake for the clients, then, was the real power to influence decisions about their services.As pointed out at the beginning of this paper, client involvement is often conceptualized with reference to an equal collaborative partnership between clients and professionals [9,25]. By shifting the distribution of power from professionals to clients, the latter are seen to be empowered with greater influence over the decisions that affect them [62]. The basic assumption is that when participating in making decisions about their own treatment, clients will take more responsibility for their situations and cope better [63]. The importance of responsibility as a result of empowerment was also acknowledged by the members of our workshops. However, while the professionals emphasized the responsibilities of the clients, they nonetheless defended their own right to decide on the suitable treatment for the client. The clients, on the other hand, resisted not only the professionals’ sole decision-making authority, but also their handing over the responsibilities to them. Again, we first show an example of the discussion among the professionals only and then among the professionals and clients.The professionals stressed the importance of the clients’ ability to cooperate in matters concerning their own care. The professional view did not really present this cooperative relationship as one between equals. Rather, the professionals expressed their frustration with situations in which clients do not understand what is best for them. According to the professionals, the clients should—paradoxically—accept their inability to understand what is best for them and give the decision-making power to the professionals. This orientation is visible in Extract 4, which starts by one of the professionals (P1) describing the challenges associated with a client refusing to adhere to a treatment recommendation by the professional.Extract 4.
+          
+            
+              01 P1:
+              on aika haasteellisii tilanteita et kun potilaat ei
+              
+                it’s quite challenging when the patients refuse
+              
+            
+            
+              02  
+              suostu menee tutkimuksiin eikä suostu ottaa lääkkeitä
+              
+                to go to examinations or won’t take their medication
+              
+            
+            
+              03  
+              eikä suostu tekee mitään ku ne ei ymmärrä sitä omaa
+              
+                or do anything because they don’t understand their own
+              
+            
+            
+              04  
+              tilannettaan niin asiathan ei kauheesti etene.
+              
+                situation so things won’t really progress.
+              
+            
+            
+              05 P2:
+              mmm nii.
+              
+                mmm yeah.
+              
+            
+            
+              06 P1:
+              eikä lääkärikään kauheest siinä voi auttaa jos potilas
+              
+                and the doctors can’t help much if the patient
+              
+            
+            
+              07  
+              ei oo yhteistyökykyinen tai kukaan terveydenhuollon
+              
+                is incapable of cooperating or no professional
+              
+            
+            
+              08  
+              ihminen oikein pysty auttamaan jos ei oo ja se tilanne
+              
+                can really help if so and that situation
+              
+            
+            
+              09 ��
+              ei parane välttämättä sitte yhtään [myöskään (.)
+              
+                won’t necessary get any better [either (.)
+              
+            
+            
+              10 P2:
+                          [mmmm.
+            
+            
+              11 P1:
+              pitäiskö nyt laittaa sit lappu,
+              
+                should we make a note then,
+              
+            
+          
+        on aika haasteellisii tilanteita et kun potilaat ei
+                it’s quite challenging when the patients refuse
+              suostu menee tutkimuksiin eikä suostu ottaa lääkkeitä
+                to go to examinations or won’t take their medication
+              eikä suostu tekee mitään ku ne ei ymmärrä sitä omaa
+                or do anything because they don’t understand their own
+              tilannettaan niin asiathan ei kauheesti etene.
+                situation so things won’t really progress.
+              mmm nii.
+                mmm yeah.
+              eikä lääkärikään kauheest siinä voi auttaa jos potilas
+                and the doctors can’t help much if the patient
+              ei oo yhteistyökykyinen tai kukaan terveydenhuollon
+                is incapable of cooperating or no professional
+              ihminen oikein pysty auttamaan jos ei oo ja se tilanne
+                can really help if so and that situation
+              ei parane välttämättä sitte yhtään [myöskään (.)
+                won’t necessary get any better [either (.)
+                          [mmmm.pitäiskö nyt laittaa sit lappu,
+                should we make a note then,
+              In lines 1–4, P1 produces a three-part list to describe challenging care-work situations. She first mentions clients refusing to go to referred examinations, secondly refusing to take their medication, and thirdly refusing to do anything, this final “extreme case formulation” serving as a way for the professional to legitimize her claim [64]. She also explains that such problems arise when clients do not understand their own situations (lines 3–4). In this way, the professional implies that the clients actually hinder the progress of their own care (line 4). In line 5, another professional (P2) shows agreement with the view, the particle nii “yeah” indicating that she is familiar with this type of situation [59]. P1 continues, stating further that the professional cannot help the client if they are “incapable of cooperating” (lines 6–8). What she seems to be suggesting is that the clients’ cooperation should realize in that they give the professionals the power to decide what is best for them. At this point, P2 minimally agrees, and P1 suggests that they write it down on the post-it note (line 11), thus treating her co-participant’s display of agreement as sufficient [65].Hence, although they emphasized the importance of cooperation, the professionals still portrayed the client’s role as quite passive. In order to receive adequate treatment and for care to progress, the clients were mainly expected to adhere to the professionals’ recommendations. It was thus suggested that the professionals had the ultimate authority to promote what they consider to be the best for the clients.Interestingly, however, the professionals also highlighted the need for the clients to take responsibility for their own care. This perspective to the issue is demonstrated in Extract 5, in which the participants discuss and write down their conceptualizations of client involvement. At the beginning of the extract, one of the client members of the workshop highlights the need to make a person become involved (line 1). This is, however, met with a lack of substantial agreement. Instead, the professional participants of the group turn the discussion toward the topic of “responsibility” as one aspect of client involvement (line 4).Extract 5.
+          
+            
+              01 C1:
+              onks se myös et osallistetaan,
+              
+                is it also that a person is made to get involved,
+              
+            
+            
+              02 P1:
+              kyllä sitäki paljon käytetään mut ei se,
+              
+                that’s also used a lot but it isn’t,
+              
+            
+            
+              03  
+              (0.5)
+            
+            
+              04 P2:
+              tavallaan kuitenki myös vastuu,
+              
+                kind of a responsibility, too
+              
+            
+            
+              05 P1:
+              nii joo totta.
+              
+                yeah that’s right.
+              
+            
+            
+              06 P2:
+              riippuu mit- (0.2) oma vastuu omast itestäki.
+              
+                it depends wha- (0.2) responsibility for oneself.
+              
+            
+            
+              07 P1:
+              joo.
+              
+                yes.
+              
+            
+            
+              08 P2:
+              mä nyt laitan sen tähän mukaan.
+              
+                I’ll put it on here now.
+              
+            
+            
+              09 C1:
+              ymmärretyksi tuleminen
+              
+                being understood.
+              
+            
+          
+        onks se myös et osallistetaan,
+                is it also that a person is made to get involved,
+              kyllä sitäki paljon käytetään mut ei se,
+                that’s also used a lot but it isn’t,
+              (0.5)tavallaan kuitenki myös vastuu,
+                kind of a responsibility, too
+              nii joo totta.
+                yeah that’s right.
+              riippuu mit- (0.2) oma vastuu omast itestäki.
+                it depends wha- (0.2) responsibility for oneself.
+              joo.
+                yes.
+              mä nyt laitan sen tähän mukaan.
+                I’ll put it on here now.
+              ymmärretyksi tuleminen
+                being understood.
+              After the somewhat ambivalent reaction to the client’s proposal (line 2), P2 suggests the idea that client involvement also involves responsibility. This idea is immediately supported by P1 (yeah that’s right, line 5). At this point it is not yet entirely clear what the term “responsibility” entails, but in line 6, P2 makes it clear that she is talking about one’s responsibility for oneself. In this case, the implication is that the client takes responsibility for their own situation and care. Compared to the viewpoint of the same professionals expressed in Extract 4, this idea is radically different. Now the clients are seen in the active role of empowered actors, who have control over their lives. After P1′s agreement (line 7), P2 displays an orientation to a sufficient level of consensus among the participants by announcing that she will write it down on the note (line 8), see [65]. At this point, however, one of the clients (C1) takes a turn and expresses a different viewpoint: the client suggests the phrase “being understood” as an alternative conceptualization for client involvement. Thus, quite interestingly, when the professionals’ “unspoken” alternative was to consider clients as influential decision-makers and the professionals handing responsibility over to them, the clients agreed less and suggested something very different from “taking responsibility.”As shown above (see Extract 4), the professionals referred to their own responsibility and superior authority when deciding on a suitable treatment for their clients, and the clients abiding by this norm was seen as “collaboration.” Extract 6 below demonstrates that clients also orient towards compliance with professionals’ decisions as the one and only option for them to demonstrate their willingness and ability to cooperate. At the beginning of the extract, one of the professionals (P1) states that in health care it is the doctor who makes the decisions.Extract 6.
+          
+            
+              01 P1:
+              kyllähän mun mielest on ihan selvä etteihän ihminen
+              
+                I think it’s completely clear that a person can’t
+              
+            
+            
+              02  
+              määrittele siis terveydenhoidossa (.) ihminen ei
+              
+                determine things I mean in health care (.) people can’t
+              
+            
+            
+              03  
+              sinänsä voi määritellä miten häntä hoidetaan (.)
+              
+                in general determine how they’re treated (.)
+              
+            
+            
+              04  
+              lääkärihän sen päättää lääkäri vastaa siitä
+              
+                It’s a doctor who decides a doctor is responsible
+              
+            
+            
+              05  
+              mitä voi mut potilas voi hyväksyä sen tai ei tai
+              
+                but a patient can accept it or not or
+              
+            
+            
+              06  
+              ylipäänsä sitä neuvotellen kannattaa tehä,
+              
+                or in general it’s advisable to negotiate,
+              
+            
+            
+              07 C1:
+              sit taas jos sä et hyväksy ni sit sä et oo
+              
+                then again if you don’t accept it then you’re not
+              
+            
+            
+              08  
+              hoitomyönteinen (.) tätä oon kuullu tosi paljon kans.
+              
+                a compliant patient (.) this is something I’ve heard a lot too.
+              
+            
+            
+              09 P1:
+              joo se varmaan on mut tietyllä tavallahan
+              
+                yeah it probably is but somehow
+              
+            
+            
+              10  
+              se sit vaan on niin että tietyi asioita on sitten
+              
+                it just is so that some things just are like that
+              
+            
+            
+              11  
+              sellasii (.) et ihminen ei voi tilata hoitoa ku pizzaa.
+              
+                (.) people can’t order a treatment like a pizza.
+              
+            
+            
+              12 P2:
+              kaikkee ei voi hoitaa kaikel tapaa et jotku asiat pitää
+              
+                everything can’t be treated in every way there are some things
+              
+            
+            
+              13  
+              hoitaa tietyl tapaa et ne tulee hoidetuks.
+              
+                that have to be treated in a certain way so to that they will be taken care of.
+              
+            
+            
+              14 P1:
+              nii et semmosia hoitojuttuja mitkä yleisesti tiedetään
+              
+                so certain treatment things that are generally known to be
+              
+            
+            
+              15  
+              toimiviks et kyllä tässä se semmonen rajanveto tai
+              
+                effective that there is this kind of line to be drawn
+              
+            
+            
+              16  
+              käynti että ihminen tulee kuulluks ja saa sanoo oman
+              
+                that a person is heard and can state their own
+              
+            
+            
+              17  
+              sanansa mutta että tulee sit kuitenkin se hoito tietyl
+              
+                opinion but that the treatment is determined
+              
+            
+            
+              18  
+              taval määritellyks.
+              
+                in a certain way.
+              
+            
+            
+              19 C1:
+              nii eihän noi yksinkertaisia asioit oo.
+              
+                yes these are not simple things.
+              
+            
+            
+              20 P1:
+              ja siitä se kai periaatteessa se puhuminen vast alkaa
+              
+                and in principle that’s when the talking begins
+              
+            
+            
+              21  
+              jos ollaan kauheen eri mieltä et miten se sit hoidetaan.
+              
+                when we really disagree on how it’s to be handled.
+              
+            
+            
+              ((begins to talk about the organization moving to a new building))
+            
+          
+        kyllähän mun mielest on ihan selvä etteihän ihminen
+                I think it’s completely clear that a person can’t
+              määrittele siis terveydenhoidossa (.) ihminen ei
+                determine things I mean in health care (.) people can’t
+              sinänsä voi määritellä miten häntä hoidetaan (.)
+                in general determine how they’re treated (.)
+              lääkärihän sen päättää lääkäri vastaa siitä
+                It’s a doctor who decides a doctor is responsible
+              mitä voi mut potilas voi hyväksyä sen tai ei tai
+                but a patient can accept it or not or
+              ylipäänsä sitä neuvotellen kannattaa tehä,
+                or in general it’s advisable to negotiate,
+              sit taas jos sä et hyväksy ni sit sä et oo
+                then again if you don’t accept it then you’re not
+              hoitomyönteinen (.) tätä oon kuullu tosi paljon kans.
+                a compliant patient (.) this is something I’ve heard a lot too.
+              joo se varmaan on mut tietyllä tavallahan
+                yeah it probably is but somehow
+              se sit vaan on niin että tietyi asioita on sitten
+                it just is so that some things just are like that
+              sellasii (.) et ihminen ei voi tilata hoitoa ku pizzaa.
+                (.) people can’t order a treatment like a pizza.
+              kaikkee ei voi hoitaa kaikel tapaa et jotku asiat pitää
+                everything can’t be treated in every way there are some things
+              hoitaa tietyl tapaa et ne tulee hoidetuks.
+                that have to be treated in a certain way so to that they will be taken care of.
+              nii et semmosia hoitojuttuja mitkä yleisesti tiedetään
+                so certain treatment things that are generally known to be
+              toimiviks et kyllä tässä se semmonen rajanveto tai
+                effective that there is this kind of line to be drawn
+              käynti että ihminen tulee kuulluks ja saa sanoo oman
+                that a person is heard and can state their own
+              sanansa mutta että tulee sit kuitenkin se hoito tietyl
+                opinion but that the treatment is determined
+              taval määritellyks.
+                in a certain way.
+              nii eihän noi yksinkertaisia asioit oo.
+                yes these are not simple things.
+              ja siitä se kai periaatteessa se puhuminen vast alkaa
+                and in principle that’s when the talking begins
+              jos ollaan kauheen eri mieltä et miten se sit hoidetaan.
+                when we really disagree on how it’s to be handled.
+              ((begins to talk about the organization moving to a new building))In lines 1–4, P1 makes a strong statement that, in health care, a person cannot determine how they are treated. P1 uses the clitic particle -hän (kyllähän, etteihän, line 1), which has been argued to indicate common knowledge [66] and expresses certainty in talk (ihan selvä “completely clear”). P1, in other words, presents her view as something that is self-evident. By announcing and reminding others about this state of affairs, P1 also positions herself as more authoritative and knowledgeable than the others. In line 4, P1 further states that it is a doctor who “decides.” However, she immediately corrects her own speech and states that it is a doctor who is “responsible” for the treatment. By this type of self-repair, P1 displays her normative orientation towards what is meant to be talked about in the given context [67]. It seems that in the contemporary “client involvement discourse,” professionals’ responsibilities may well be normatively easier to topicalize than their decision-making authority over the client. Indeed, in lines 5–6, P1 seems to seek to further mitigate her prior view on decision-making, emphasizing that it is not only the professional alone, but the client and professional together, who negotiate treatment decisions. The client’s role in this negotiation is, however, presented as narrow: the client can either accept or refuse the professionals’ decisions (line 5).At this point, however, one of the clients (C1) states—possibly sarcastically—that if a client does not accept the professional’s decision, then they are not seen as hoitomyönteinen “compliant” (lines 7–8). With this statement, the client seems to be referring to the traditional view according to which a “good patient” is passive and compliant, e.g., [68]. Thus, if a client wants to be a “good patient”—that is, to cooperate and play the game with its long-established rules [50,69]—they have no other option but to accept the professional’s decision. In this way, the client expresses doubt about their ability to genuinely have a say in the decisions made in social and health care encounters. The client substantiates his claim by also pointing out that this is something he has heard from others and does not represent (only) his own experience.In lines 9–11, P1 responds to the client. She reformulates her prior statement by using a figurative expression “people can’t order a treatment like a pizza” (line 11), which, in this context, comes across as highly defensive. These types of expressions have been observed in connection with complaints, for example, to enhance their legitimacy and to bring the complaint sequence to a close in the face of a lack of agreement [70]. At this point, another professional (P2) joins the conversations, supporting P1′s view (lines 12–13). He emphasizes the need to have hoitojuttuja “treatment things” managed in a certain way, which means favoring the solutions that have proven to be effective. He also highlights a need to draw the line between a client being heard and stating their opinion, on the one hand, and the professional determining the client’s treatment, on the other (lines 14–18). The client concedes by agreeing that these are not simple things to deal with (line 19), which is followed by P1 pointing out that client disagreements will be dealt with as they occur in the social and health care encounters. With this comment, P1 closes the discussion and moves on to a new topic.Thus, even if the professional in this situation expressed willingness to attend to clients’ concerns in the future once they become relevant during the consultations, in so doing, she ignored the client’s concern in the here and now of the client involvement workshop encounter. In this case, the client’s concern was on a meta level, being about his theoretical opportunity to have a say in decisions about his own care.In this paper, we have examined possible differences in the ways in which social and health care professionals and clients in co-development workshops perceived client involvement and unraveled the degree to which they share their perceptions. We found that both the clients and the professionals agreed that on a general level client participation should be promoted, but the main obstacle hindering its realization seemed to be the health care system. When considering client involvement as a future ideal, the realization of which was out of their hands, the views of the professional and client members of the workshop were mutually congruent. Both the clients and the professionals also agreed that being heard was a prerequisite of client involvement. Yet, content-wise, the professionals and clients emphasized slightly different ideas, which points to a subtle discrepancy between their views. The professionals stressed the importance of being heard when conceptualizing client involvement, whereas the clients asked for real power to influence the services. On the other hand, when the professionals handed the main responsibility over to the clients, the clients were not eager to agree with their view. Moreover, in contrast to giving the responsibility of the client’s own care to the clients themselves, the professionals referred to their own responsibility when deciding on a suitable treatment for a client. In this way, they expressed their superior authority to make the decisions. These themes of responsibility and authority were also intertwined with the question concerning collaboration. The clients considered compliance with professionals’ decisions as the only option to display their ability to cooperate. In other words, knowing how to play the “client involvement” game requires recognition of the limitations of that very involvement.These findings reflect the tensions around “expert” knowledge, control, responsibility, and power traditionally reported in social and health care, e.g., [71,72]. As shown in prior research [9,26], the clients in our data wished greater involvement in service delivery (Extract 3) but they also wanted the professionals to recognize this wish to be optional and varying according to the amount of responsibility the client can take (Extract 5). The clients also considered the participation in shared decision-making as crucial part of involvement [9,25,26] but suspected that being categorized as “non-compliant” prevents them from participating genuinely in decision-making [69]. As noted in Anthony and Crawford’s paper [28], the professionals seemed to value client involvement as such but to be reluctant to adopt it as a guiding clinical practice (as shown in Extract 5). The professionals in Extract 1 referred to systemic barriers for not being able to take their responsibility to make clients involved, and in Extract 5 they laid the responsibility to be involved on the client. It can also be argued that the professionals in Extract 6 present the traditional medical view rather than being adherent to client-centered care or the principles of shared decision making [10,12]. It might be that although the professionals value the client involvement as such, they might experience the greater client empowerment as threatening their professional boundaries [27,29].In analyzing the ways in which social and health care professionals and clients perceive the notion of client involvement, we found various tensions and discrepancies between their views. It is important to note, however, that unlike in certain conversational contexts, such as radio or television talk shows, in which explicit debates and overt controversies are common and even expected [73,74], people typically avoid argument and disagreement [75,76,77]. This was also the case in our data, in which all the discrepancies analyzed were implicit, occurring below the surface level of the interaction. More specifically, although the participants basically expressed agreement with each other’s views, simply building and elaborating on them in and through the turn-by-turn unfolding of interaction, they displayed differences in their orientations towards how knowledgeable they were, or were expected to be [78] and who was to define what should and what should not be done [79]. Such negotiations consist of participants dealing with each other’s turns, not entirely on their own terms, but in ways that slightly deviate from and refrain from appreciating the full interactional import of the earlier talk [40] (p. 260). The motivation for such negotiations, in turn, may be argued to be simply about the need for people to maintain their views about themselves [79] (p. 383)—in this case, either as clients who have control over their lives and who deserve to be heard and to participate in decisions concerning their own treatment, or as social and health care professionals who have the ultimate authority and responsibility to promote what they consider to be best for the clients. The analysis of the precise contents of these negotiations nonetheless allows us to obtain a deeper understanding of the process of social and cultural change in perceiving the role of the client in social and health care.The study has a number of limitations that have to be taken into account. We strived to increase the trustworthiness of the qualitative analysis by listening to the recordings while reading the transcript, conducting independent coding of pieces of data and discussing selected segments of the recordings with the research team to specify what we could reliably identify in the data. Another obvious limitation is the relatively small number of participants in our data, which constrains the generalizability of our results. In a similar vein, given that all our data came from a very specific context—client involvement workshops in two Finnish municipal social and health care organizations—our results cannot be freely applied to other contexts. Furthermore, the clients in our data were not randomly chosen; they were particularly active in participating in the organizational development activities and obviously did not represent the heterogeneous group of clients as a whole. It can also be argued that they did not represent the most marginalized and disadvantaged groups of clients. On the other hand, they had personal experience of being in that position, having subsequently also gained the ability to voice their concerns and participate in the “officially” driven development workshops [21].In addition to the limitations described above, it is also worth noting that the number of the clients participating in the workshops was smaller than the number of the professionals. As the workshops were organized to develop organizational work practices, and the participants were thus recruited within the organizations, it was surprisingly challenging for them to get the clients signed into the workshops. It might be that the actual participation of clients in the co-development of services is still quite scarce, regardless of how big a trend client involvement is in the Finnish social and healthcare services. This imbalance was also visible in our analysis, as we did not have small-group discussions with only clients as participants. The imbalance might have also affected the dynamics of the workshop discussion, as the professionals, who traditionally have the authority to dominate the interaction, were outnumbered. There therefore exists the possibility that this imbalance has reinforced the very power imbalance that the study was set out to examine. However, as we hope has become clear from our analysis, the aim of this study was not assessing the degree to which power imbalances exist or not. Instead, through the means of conversational analytic methodology, our goal was to unravel those nuanced practices of interaction by which power imbalances between professionals and clients are realized in interaction. This, in turn, might have a practical value in informing future co-development processes between professionals and clients.In terms of clinical practice, our paper highlights the importance of being aware of differing expectations the professionals and clients may have on the client involvement. As these expectations are not easy to negotiate in clinical encounters, some aspects of the client involvement, such as participation in decision-making and taking the responsibility over the care, may be treated as one and the same aspect of client involvement. As this might cause even more confusion and misalignments between the participants, we suggest that the different dimensions of client involvement would not be overly simplified and, as the client in Extract 6 states, treated as simple things.Our analysis of the “client involvement” workshops has mostly highlighted the differences between the clients and professionals’ views on what client participation entails. In addition to the social and cultural change in the client’s role, such differences also point to a lack of extensive contact between the two participant groups. Although professional–client relations may well be taken into the sphere of meta-level reflection in informal encounters among professionals at the workplace, and clients may have analogous conversations with their friends and family members, our everyday life entails very few situations in which such relations could be discussed by clients and professionals together. The “client involvement” workshops from which our data were collected therefore seem to fulfil an important function in advancing the emergence of a shared understanding of what may be expected from the client. Although in this respect, subtle implicit discrepancies easily escape the eye, our analysis suggests that the participants themselves nonetheless orient towards them. The precise experiential consequences of having to constantly negotiate your self-understandings is an empirical question to be addressed in future research, but a preference may well exist for remarks by recipients that validate the first speakers’ claims of rights to knowledge and decision-making [79,80], while remarks that challenge the speakers’ self-concepts may increase their anxiety [81] (p. 474). Fostering a shared understanding of the role of the client may therefore be a worthwhile goal.In this paper, we have analyzed conversations between clients and professionals in social and health care on “client involvement.” As all meta-level reflections in terms of “conversations about conversations,” our data—demonstrating client involvement in talk about client involvement—also showed that what is happening at the level of the content of talk may or may not be in line with what is happening at the level of interaction here and now. When a client in a workshop expresses doubt about the ability of the client to genuinely have a say in the decisions made in the social and health care encounters, a professional—as we saw in Extract 6—may circumvent the client’s criticism by pointing out that client disagreements will certainly be dealt will as they occur in the actual social and health care encounters. Intriguingly, however, by highlighting and drawing on the normative ideal according to which such disagreements cannot be ignored by professionals, the professional actually ignored the concern of the client in situ. It is thus a considerable paradox that in cases such as this it is the client involvement rhetoric and discourses themselves that provide the professionals with resources to actually hinder client involvement.Allowing client involvement to emerge now (and not in the future) is a critical challenge for any social and health care professional. At the same time, sequences of social interaction are essentially held together by the participants carefully attending to what each of them has just said when designing their responses. Systematically, giving such attention to the client—a phenomenon that some authors have referred to as “nexting” [82,83,84]—allows new insights to emerge, but also implies a degree of lack of control over the outcome of the encounter—something that a professional might not feel comfortable about. Concern over the effective routine functioning of the institution might thus generate a barrier for the professional to engage in practices of “letting the other happen to me” [85] (p. 232). However, determining what truly ethical conduct in social and health care interaction looks like may actually necessitate doing just that.Conceptualization, E.W., S.K., and M.S.; Methodology, E.W., S.K., and M.S.; Validation, E.W., S.K., M.S., and L.-L.U.; Formal Analysis, E.W., S.K., and L.-L.U.; Investigation, E.W., S.K., and L.-L.U.; Resources, E.W.; Data Curation, E.W.; Writing—Original Draft Preparation, E.W., S.K., M.S., and L.-L.U.; Writing—Review and Editing, E.W. and M.S.; Visualization, L.-L.U.; Project Administration, E.W.; Funding Acquisition, E.W. All authors have approved the submitted manuscript and agreed to be personally accountable for their contributions. This research was financially supported by the European Union Social Fund (grant number S21564) via the Finnish Ministry of Social Affairs and Health.Open access funding provided by University of Helsinki.The authors declare no conflict of interest. The authors alone are responsible for the content and writing of the paper.Simplified transcription symbols[ ] Overlapping talk(.) A pause of less than 0.2 seconds(.) Pause: silence measured in seconds and tenths of a secondword Accented sound or syllable((word)) Transcriber’s comments- Abrupt cut-off of preceding sound? Final rising intonation, Final level intonation. Final falling intonation= Continuous talk between speakersThe number of interaction segments with mutual congruence/discrepancy of viewpoints in the groups of professionals only and in the groups of both clients and professionals.

Med-MDPI/ijerph_5/ijerph-17-16-05654.txt

@@ -0,0 +1 @@
+Differentiation of true from mimicking Eagle’s syndrome based on conventional radiography is difficult; however, cone beam computed tomography (CBCT) images can contribute to proper diagnosis of mimicking Eagle’s syndrome. The aim was to study radiological images of a 37-year old female patient (patient #1), with chronic cervicofacial pain who underwent radiological diagnosis with a conventional panoramic machine; another 75-year old male patient (patient #2), with chronic oropharyngeal pain, underwent a radiological diagnosis with the CBCT machine, with a field of a view of 16 × 12 cm. Exposure factors were 120 kVp, 7 mA, with a 20 s exposure time of acquisition. The results show a panoramic image (patient #1) with a pathologically elongated styloid process 46 mm of length, which was surgically removed, releasing the patient from further pain episodes. CBCT acquisition (patient #2) showed an impacted left maxillary canine in the edentulous maxilla and a peculiar elongation of both stylohyoid complexes as impressive, “collar-like”, bilateral, elongated, multiple segmented, calcified stylohyoid complexes, without pressure on the vital neurovascular neck structures, mimicking true Eagle’s syndrome. The impacted maxillary canine was surgically extracted with a subsequent resolution of pain episodes and the cessation of neurological complaints. The conclusions suggest that the use of CBCT images can contribute to differentiating mimicking from true Eagle’s syndrome, which has been rarely reported in the literature.Today, three dimensional (3D) cone beam computed tomography (CBCT), as a specialist medical tool, has become an important and inevitable source of 3D volumetric data in orofacial clinical practice, even though the initial purpose of CBCT was to perform angiography, mammography, and serve as guidance in radiotherapy [1]. CBCT can also serve to detect hidden or unclear anatomy, as well as occult maladies, thus reducing omissions in the early diagnosis of a relevant disease [2].The first description of the pathologically elongated temporal styloid process was made by American ENT specialist, Dr. Watt Weems Eagle, in 1937 [3,4,5]. Eagle’s syndrome (styloid syndrome or styloid–carotid artery syndrome) is a rare condition where an elongated temporal styloid process (apex) is in proximity to the important neuro-vascular anatomical structures of the neck, causing pharyngodynia, and chronic head and neck pain (cervico-facial pain) [6]. The average normal length of the styloid process is 20–30 mm [7], projected to the lesser horns of the hyoid bone. The apex of the styloid process is interposed between the medially located internal carotid arteries and the laterally positioned external carotid arteries, and is in close proximity to the facial, accessory and vagus nerves, the internal jugular vein (medially located), and the glossopharyngeal nerve (under the styloid process), accounting for the glossopharyngeal neurological symptoms often observed in Eagle’s syndrome [7]. The base of the styloid process (tympanohyal), the styloid process itself (stylohyal), the stylohyoid/stylomandibular ligaments (ceratohyal) and lesser horn of hyoid bone (hypohyal) form the stylohyoid complex (SHC) [8,9], which is affected, and mineralized (ossified) in Eagle’s syndrome [3,4,5,7]. According to Fromer [10], embryological data has shown that the hyoid bone and styloid process, with its attached ligaments, originate from the first and second branchial cord, in addition to the cartilage structure, i.e., the so-called Reichert’s cartilage [11]. The stylohyoid complex can solidify during the course of life and be radiologically seen as an elongation, which is a more frequent variant, or as a complete ossified structure, which is a very rare variant; these abnormalities can cause pharyngodynia, which could be considered as pathognomonic for Eagle’s syndrome [12].However, it is still clinically important to differentiate the true symptomatic Eagle’s syndrome and the asymptomatic elongated and calcified stylohyoid complex. Even though there is a plethora of reports of Eagle’s syndrome in the medical literature [3,4,5,8,10,11], only a few clinical cases of the “collar-like” symptom-free bilaterally elongated and calcified stylohyoid complex have been published [12].It should be noted that the two-dimensional orthopantomographic technique has numerous shortcomings for radiographic analysis of anatomical structures and assessment of available bone tissue for various orofacial surgeries, primarily due to radiological enlargement of the anterior region of the mandible, possible distortions of the image, false image impression of enlarged anatomical structures, and overlapping of the close bony structure in the proximity of the styloid process [9]; all these shortcomings can lead to extremely erroneous radiological assessments of the true dimensions of the anatomical pathological structures of the oropharyngeal skeleton [13]. A radiological machine for computerized imaging of the jaws, which uses digital volumetric tomography (3D) with conical beam radiation (cone beam CT-CBCT), has shown superiority in detecting specific changes in the orofacial region, primarily due to low radiation dose, its exceptional precision and practicality, and the ability to make real-time radiological cross-sections of tissues in the axial, coronary, and sagittal planes, as well as panoramic cross-sections, cross-sections, oblique cross-sections, and temporomandibular joint imaging [14].A dilemma in clinical practice is as follows: is it possible to use CBCT acquisition as radiological tool to help clinicians to differentiate a mimicking (false) from true, very rare oropharyngeal skeletal pathologies, i.e., a pathologically elongated temporal styloid process (Eagle’s syndrome).Consequently, the aim of this paper is to present and describe the use of panoramic and CBCT images in differentiation of the true from the very rare, asymptomatic, “mimicking” Eagle’s syndrome patients.This case study has been approved by the ETHICAL BOARD of the Clinic of Stomatology, Nis, Serbia, by its decision # 146/2-6. All procedures performed involving human participants were in accordance with the ethical standards of the institutional and/or national research committee, following the rules of the Declaration of Helsinki of 1975 (https://www.wma.net/what-we-do/medical-ethics/declaration-of-helsinki/), revised in 2013. We obtained patient’s consents at the time of admission at our department.A 37-year-old woman (patient #1), was referred to our institution because of pain in the throat and the right side of the neck and face. Her pain episodes started four years ago, prior to planned mandibular teeth decay and root canal re-treatment of a tooth in the mandible and maxilla. She was constantly experiencing severe cervico-facial pain on the right side of the head, with the pain attacks lasting approximately five-ten minutes, and with a pain frequency of three-five times per day. The irradiation of pain was from the external auditory canal to the right side of the throat, with pain episodes during swallowing and moving tongue during speech. The patient had no past medical history significant for systemic or local disease, with the laboratory values of blood and urine within normal range.The patient’s orofacial region was thoroughly examined with no clinical findings indicating the cause of her pain episodes. However, oropharyngeal examination showed severe tenderness and pain on palpation in the right tonsillar region, with motion pain occurrence during head horizontal rotation. Panoramic X-ray acquisition showed pathological elongation of the right styloid process, like a “thorn” with a length of 49 mm by manual measurement with a ruler (Figure 1).A 72-year-old man (patient #2), was referred to our institution for the assessment of “inexplicable pain” in the left oropharyngeal region, with mild dysphagia and vague throat discomfort of a few minutes’ duration, first pain episodes having started two-three years before. The patient’s medical history was contributory for coronary heart disease and his laboratory data for calcium, vitamin D, and phosphorus metabolism were within normal range. Intraorally, avoiding the gag reflex, physical examination revealed that there was a palpable “asymptomatic foreign body” mass in the bilateral posterior alveololingual grooves. During the assessment of the patient’s edentulous left maxilla, intraoral clinical evidence of a possible unerupted maxillary canine emerged, barely showing the edge of the cusp of the maxillary canine. Panoramic X-ray acquisition was initially performed. However, in order to differentiate true from mimicking Eagles’ syndrome, additional CBCT scans were obtained using the 3DCT® (from manufacturer MEDICA Systems and Products, Nis, Serbia), previously used successfully in the imaging of rare pathologies. The CBCT machine, 3DCT®, consists of an amorphous-silicon flat-panel detector, X-ray tube, and a source collimator. CBCT scans were performed in the patient’s sitting-up position, and with straps over the forehead in order to prevent movement of the head. During the gantry rotation of 360°, for the purpose of data acquisition, the mandible rested in a prefabricated chin cup. The technical parameters were standardized as follows: used field of view–16 (w) × 13 (h) cm, exposure time of 20 s, X-ray generator of 120 kVp with 7 mA, and a radiation dose of 37 micro-sievert (µSv). The obtained images were reconstructed by using the previously described algorithm and technical data which represents a combination of the high spatial frequency reconstruction algorithm and the acquired image data, consisting of a projection image matrix of 1560 × 1900 (pixels), with 14 bits that delivered 16,384 shades of gray for a better image contrast, by using an amorphous silicon digital X-ray imager (receptor), manufactured for the high-speed dental cone beam CT (PaxScan 2520D/CL; Varian Medical Systems, Inc., Salt Lake City, UT, USA); other technical parameters consisted of a volume matrix of 800 × 800 × 650 (voxels), volume size of 18 × 18 × 13 cm (i.e., 3615.840 cm3), and 0.3 × 0.3 × 0.3 mm (i.e., voxel size of 0.027 mm3). The image slices were reconstructed in the resolution mode, with the isotropic cubic voxel size of up to 300 µm (0.3 mm).Patient #1 is categorized in the group confirming diagnosis of true Eagle’s syndrome and scheduled for surgery. After preparation for general anesthesia, surgery was performed by the extraoral trans cervical per via approach into the parapharyngeal space, for identification of styloid process, which was subsequently surgically removed for half of its length (~25 mm), seen in repeated panoramic X-ray acquisition (Figure 2).The postoperative course was uneventful and the patient was discharged from our institution on the seventh postoperative day. Over the next few weeks the patient stopped experiencing any pain in the right cervico-facial region with cessation of pain episodes.Patient #2′s panoramic X-ray acquisition showed bilaterally pathological nodular elongation of the left and right styloid processes extending beyond the lower border of the mandible (Figure 3); an unerupted impacted maxillary left canine and, on the right side, the sinus mucocele was also observed. A tentative diagnosis was made that the patient had true Eagle’s syndrome due to the unusual panoramic image and pain experiences.The extracted CBCT’s panoramic view shows an atypical bilateral elongation of the styloid process, with the unerupted impacted maxillary left canine and, on the right side, the sinus mucocele; the CBCT’s extracted panoramic and axial views of the maxilla shows a more detailed unerupted impacted maxillary left canine and two additional sinus pathologies in the maxillary sinuses; in the left maxillary sinus, there was chronic inflammatory hyperplastic mucosa, while the right maxillary sinus exhibited a sinus mucocele (Figure 4).Apart from these findings, there were multiple astonishing CBCT images of the different planes, which show an impressive bilateral mineralization of the styloid complex. All the taken CBCT radiographic images showed a “collar-like” multi-interrupted and segmented calcified stylohyoid complex (SHC). On the left side, the segmented calcified styloid ligaments fully stretched to the lesser cornu of the hyoid bone and on sagittal CBCT images, measuring 85.87 mm in length with an average width of 3.63 mm, while the right side barely reached the lesser horns of the hyoid bone, measuring 81.13 mm in length with an average width of 2.58 mm (Figure 5).Due to suspicion that this patient belongs to the Eagle’s syndrome group, additional angle measurements were performed. The measurements of important angles for 3D visualization of the SHC showed: right 53.54° (short) and left 66.8° (~normal) for the mediolateral angle (MLA), right 80.63° and left 78.78° (elongated) for the anteroposterior angle (APA), indicating wide both angles [9]. Measurements of maximum thickness with values of 4.56 mm and 3.6 mm (normal) for the right and left sides were seen, respectively. On both sides, the pattern of calcification/mineralization was almost identical, involving the tympanohyal, stylohyal, ceratohyal and hypohyal parts of the SHC (Figure 6).In this clinical situation, we consider that pain of other origin, for example, dental pain from an impacted tooth, can contribute to this asymptomatic pain free patient with calcified styloid process, experiencing painful episodes, which can in fact be a “mimicking”, but not a symptomatic/painful true Eagle’s syndrome.The patient was scheduled for the odontectomy of the impacted non-erupted left maxillary canine, which was surgically extracted. The patient experienced the complete resolution of previous cervicofacial pain episodes one month after odontectomy, with a subsequent cessation of further visits to our institution.There are numerous previously diagnosed reports of Eagle’ s syndrome in the literature [3,4,5,8,10,11,15], but 3-D CBCT images as an aid in differentiation of the bilateral asymptomatic (non-syndromic) “collar-like” calcified stylohyoid complexes from true, symptomatic (syndromic), painful Eagle’s syndrome are uncommon [16].Pathological 3-D angulation and pressure towards the vital neuro-vascular neck structure of the elongated and mineralized SHC process is the paramount anatomical cause for “classic” painful Eagle’s syndrome [6].It should be stressed, that the elongated styloid process that is mineralized is not pathognomonic for Eagle’s syndrome, because many patients with a mineralized, and elongated styloid process are asymptomatic-pain free patients [17].In the panoramic view, when the styloid process is elongated, it attains over one third of the mandibular ramus length [18], which is presented on the panoramic image in patient #1. Taking all the obtained clinical facts into account (clinical symptoms and panoramic image) in the described case, diagnosis with proper subsequent surgical treatment of true Eagle’s syndrome was made in patient #1.It is necessary to understand better and to clarify the importance of the origin of pain, which is an inseparable part in defining true Eagle’s syndrome. We define that a person with “mimicking/imitating Eagle’s syndrome” is a patient with calcified stylohyoid complex who experiences a pain of other origin (non-Eagle’s pain), which is not due to mineralized stylohyoid complex (SHC) [8,9], as a cause of SHC’s pressure on the neck’s nerve complexes; especially unilateral pain, which may came from another origin, such as an impacted maxillary canine in our case, which was in fact a contributing factor for “mimicking/imitating true (symptomatic, painful) Eagle’s syndrome.”The diagnosis of “mimicking Eagle’s” syndrome is made due to the results of the 3–D left mediolateral angle (MLA) measurement, which showed a normal range on the left side with the value of 66.80° (~normal), with wide anteroposterior angle (APA), so we concluded by using 3–D CBCT visualization and MLA and APA measuring, that this case is “mimicking” but not a true Eagle’s case. The unilateral pain then must have come from another origin, which was an impacted maxillary canine in our case. The term “mimicking” is a literary term, which describes, in this case, a mimicking of a true painful Eagle’s syndrome with the pain origin due to pressure of the mineralized stylohyoid complex (SHC) on the neck’s nerves. Contributing to our proper diagnosis of mimicking Eagle’s syndrome is a pain free clinical course after surgical extraction of the maxillary impacted canine; all patient’s pain complaints disappeared within a few postoperative weeks, which means that there was no pressure on the vital neurological and vascular anatomic neck elements and it is more likely that the pain episodes were provoked and caused by the impacted maxillary canine.The CBCT acquisition performed on patient #2 showed remarkable images of a segmented and completely calcified bilateral SHC, proving the necessity for CBCT imaging as an essential tool in the diagnosis of the cause of pharyngodynia, suspecting Eagle’s syndrome [17]. According to Langlais et al. [16], the pathological mineralization of the styloid complex is radiographically classified into three types. Type I is the elongated uninterrupted styloid process (more than 30 mm); type II is the pseudo-articulated styloid complex with stylo-mandibular and stylohyoid ligaments joined to the styloid process; and type III is the segmented and mineralized styloid complex. This classification could lead us to a diagnosis of the left type II and right type III mineralized styloid ligaments, in this case [15].However, even though these CBCT radiographic images are impressive and support Eagle’s syndrome [19], it is still a debatable issue whether this is true Eagle’s syndrome or “mimicking” Eagle’s syndrome, due to the resolution of pharyngodynia (cervico-facial pain) after the surgical extraction of the impacted canine in patient #2. It is important to emphasize that some literature reports consider true Eagle’s syndrome only when the facial or neck pain originates from a SHC that is mineralized [20]. A pathologically elongated and mineralized SHC should be bonded with a symptomatic, palpable and painful “foreign body” mass in the throat, with constant chronic head and neck pain [6], due to the chronic irritation and pressure of the elongated styloid process on the vital neuro-anatomic elements in the neck, as in patient #1; these clinical findings were not discovered in their full capacity in patient #2. It is a known fact that orofacial pain is a pain within the trigeminal sensory neurological system incorporating also, among the oral anatomical structure, the pharynx and infratemporal fossa. Occurrence of micro movement of the impacted maxillary canine can be provoked by wearing a full denture, and subsequent local pericoronitis and its inflammation, as a trigger, contribute to subsequent irradiation of the pain stimulus via trigeminal nerve branches; all these pathological events can contribute in the presented case with calcified styloid process and impacted maxillary canine, to patient’s experience of pain, which can in fact be a “mimicking/imitating Eagle’s syndrome, but not a true Eagle’s syndrome”. There is a distinct possibility that a clinician, when faced with the impressive 3D “collar” like images of elongated stylohyoid complex/stalk extension, like the one we described, could firstly reach the tentative diagnosis of true Eagle’s syndrome without elaborating the possibilities, of other causes of unilateral cervicofacial pain, as was our case. That is why, in this patient with an impacted canine, we primarily surgically extracted this impacted canine and the patient in fact had clinically experienced various pain symptoms, which could trigger pain episodes manifested as sensory disturbances in the nasopharynx, mimicking pain like that in true Eagle’s syndrome. It is obvious that for these suspicious and unclear clinical cases, there is need for “active” 3-D radiological assessment of SHC and its angles, in order to make an accurate diagnosis of pain origin. Lengele and Dhem [20], suggested that a long styloid process with a downward, ventral and medial direction may be responsible for the creation of pain in true Eagle’s syndrome. Ramadan et al. [9] proposed several new parameters for analyzing the “normal vs. pathological” stylohyoid chain related to the mediolateral angle (MLA), the anteroposterior angle (APA), the angle of the styloid process’s base and the tip of the projection of SHC at the skull’s base bending, and the maximum thickness and length. Although the MLA and APA corners should show that there is a possibility of SHC pressure on the vital neurovascular neck structures with a consequent cervicofacial pain, which would support true Eagle’s syndrome, this has not been proved with CBCT visualization and measurement in our patient #2.The presented cases of true and “mimicking” Eagle’s syndrome are good examples how the creation of obscure and rare pathological entities could be misdiagnosed or incompletely diagnosed with conventional 2-D radiological apparatus. CBCT can contribute to correct diagnosis of mimicking Eagle’s syndrome and the origin of pain episodes i.e., impacted canine in this case, showing numerous impressive images of the “collar-like” non-syndromic bilateral calcified stylohyoid complex, which has not been described often in the literature.M.T.: acquisition of data, writing the manuscript, analysis and interpretation of data; N.B.: acquisition of data, writing—original draft preparation, review and editing the manuscript, analysis and interpretation of data, revision of the manuscript for significant intellectual content, providing original images; K.B.: acquisition of data, technical support, and assistance in English language draft’s managing. All authors have read and agreed to the published version of the manuscript.This research received no external funding.The authors declare no conflict of interest.Cropped panoramic image showing pathologically elongated styloid process ending at the inner side of angle of mandible. The length of styloid process is 46 mm (conventional manual measurement with straight ruler).Cropped panoramic image showing resected styloid process.Panoramic X-ray acquisition showing bilaterally pathological nodular elongation of the left and right styloid processes extending beyond the lower border of the mandible; an unerupted impacted maxillary left canine and, on the right side, the sinus mucocele can also be seen.The axial cone beam computed tomography (CBCT) slice shows (upper image): panoramic radiography showing an unerupted maxillary left canine with a cyst-like formation around the canine crown and sinus pathology in the right maxillary sinuses. Elongated bilateral stylohyoid complex (SHC) processes are present and attain over one third of the mandibular ramus length (white arrows); (lower image): CBCT reconstruction images showing chronic inflammatory hyperplastic mucosa in the left maxillary sinus, impacted canine, while the right maxillary sinus exhibits a mucocele affecting the latero-basal portion of the sinus.Top image–CBCT three-dimensional reconstruction shows an impressive calcification of the SHC in different image modes (left–default 2 mode, center–default mode, right-skin/red mode throughout antero-posterior slice sectioning). On the left side, the CBCT radiographic appearance of a “collar-like” and elongated SHC shows a multi-interrupted and segmented calcified SHC, fully stretched to the lesser cornu of the hyoid bone (see arrows). Bottom image–in a different image mode, on the right side, CBCT three-dimensional reconstruction shows the segmented calcified SHC which barely reaches the lesser cornu of the hyoid bone (left–default 2 mode, center–default mode, right–skin/red mode throughout antero-posterior slice sectioning) (see arrows).3D measurement of angles for analyzing a “normal-pathological” SHC: (left) measurements of mediolateral angling (MLA) of a calcified/ossified coronal 3D-CBCT image; (middle) measurements of right anteroposterior angling (APA) of a calcified/ossified SHC with the longitudinal axis of the SHC and Mc Rae’s line; (right) measurements of left anteroposterior angling (APA) of a calcified/ossified SHC with the longitudinal axis of the SHC and Mc Rae’s line.

Med-MDPI/ijerph_5/ijerph-17-16-05656.txt

@@ -0,0 +1 @@
+Background/objectives: The longitudinal effect of abdominal weight status (AWS) defined by waist circumference (WC) on healthy aging has not yet been comprehensively examined. Therefore, the purpose of the present study was to examine the temporal association between WC-defined AWS and a comprehensive assessment for healthy aging. Subjects/methods: This study utilized data from 5211 respondents aged 65+ who participated in the National Health and Aging Trends Study from 2011 to 2018. Mixed effects regression models were used to examine the association between baseline AWS and the annual change rate in healthy aging score (HAS) via interaction terms (AWS*round) adjusting for confounding effects. Further multiple mixed models examined the relationship of AWS and HAS over an 8-year period. Results: There were no annual change rate differences in HAS by baseline AWS, regardless of sex. However, males with abdominal obesity were more likely to have a lower HAS than males with normal AWS (β = −0.20, 95% CI: −0.30, −0.10, p < 0.001) but no difference in HAS was observed between males with overweight and normal AWS. A similar pattern was observed among females. Conclusions: Study results indicate that AWS was associated with HAS but it did not modify annual HAS change rate over time.The United States (US) population is steadily aging. The percentage of adults aged 65+ increased 34.2% over the past decade [1] and thus it is important to understand predictors of healthy aging in older adults. Healthy aging is a complex and multifactorial concept that incorporates factors associated with the aging process, including physical, mental, and social wellbeing [2,3]. The World Health Organization (WHO) describes healthy aging as the absence of major chronic diseases (e.g., diabetes, cancer, cardiovascular disease), the presence of good physical/cognitive function, and wellbeing [2]. Recent research indicates that additional health indicators are important for healthy aging, such as health-related limitations in social life, function limiting pain, mental health, and perceived health status [3,4,5,6]. Despite efforts to identify predictors of healthy aging, existing longitudinal studies have utilized only a limited range of health indicators to measure healthy aging and have not incorporated other factors such as major chronic diseases [3], wellbeing [5,6], function limiting pain [3,6], and perceived health [4,6]. To better understand factors associated with healthy aging, there is a critical need for longitudinal research that comprehensively measures healthy aging and considers key health indicators identified from the existing literature.The relationship between weight status defined by body mass index (BMI) and aging has been studied extensively [7,8]. Even though existing longitudinal studies have primarily focused on a limited number of aging-related health indicators such as physical function, disability, dementia, and comorbidities, these studies provide critical evidence of the relationships between weight status categories based on BMI and the health of older adults [7,8]. However, in recent years, waist circumference (WC) has been recognized as a better measure than BMI in older adults for longitudinal aging studies [9]. The reason for this is that older adults are likely to experience age-related decreases in muscle and bone mass and BMI weight classifications would view these as healthy changes while abdominal weight status (AWS) defined by WC would not and it was less influenced by aging than BMI [10,11]. In addition, BMI does not distinguish abdominal obesity, which is the most metabolically deleterious, whereas WC directly reflects abdominal fat mass [9,10]. Accordingly, the purpose of the present study was to examine associations between AWS defined by WC and a comprehensive assessment of healthy aging over an eight-year period in older adults aged 65+ in the US using nationally representative longitudinal data. All analyses were stratified by sex due to the sex difference identified in the existing literature [12].The present study (see Supplementary Table S1 for study design summary) used data drawn from the 2011–2018 National Health and Aging Trends Study (NHATS), which was a longitudinal study of nationally representative Medicare beneficiaries aged 65+ years in the USA NHATS conducted surveys and measurements annually with response rates ranging from 70.9% to 94.8% [13,14]. A total of 8245 individuals participated in 2011 NHATS and 8077 of these had positive analytic weight [15]. Respondents were excluded from the current study for the following reasons: (1) missing baseline WC and BMI data (n = 1477) in 2011; (2) BMI < 18.5 kg/m2 in any round of data collection (n = 436) due to the possible underweight related physical and psychological pathology [16]; and (3) missing baseline health indicators (see method for detailed information) related to healthy aging (n = 953). Using these criteria, 5211 respondents were included in the present study, which was approved by the Institutional Review Board of the University of Rhode Island (IRB# 1551268-1).The present study utilized WC to determine AWS. At baseline and each of the following seven rounds, WC was measured by trained staff using a flexible tape measure around each respondent’s waist at the level of their umbilicus following standardized protocol [17]. WC was used to create the following AWS categories: normal (males WC < 37 inches, females WC < 31.5 inches), overweight (males 37 inches ≤ WC < 40 inches, females 31.5 inches ≤ WC < 35 inches), and obese (males WC ≥ 40 inches, females WC ≥ 35 inches) [9,18].Healthy aging was assessed using 10 health indicators based on the WHO and Assmann and colleagues’ definition of healthy aging [2,5]. These indicators (see Table 1) encompassed physical and cognitive function, wellbeing, major chronic disease, depression, anxiety, instrumental activities of daily living (IADL), health-related limitations in social life, function limiting pain, and perceived health [2,5]. Each health indicator was scored “1” if the established or set criterion was met and “0” if it was not. Scores for each indicator were then summed to create a HAS (range 0–10). The HAS was then dichotomized based on score distribution as done in prior healthy aging research [6]: (1) good (above median, scores 7–10) and (2) poor (below median, scores 0–6). Each of the 10 health indicators and the scoring criteria are discussed below.Physical function was measured by the short physical performance battery (SPPB), which is comprised of three subtests: balance stand, walking speed, and repeated chair stand [14,17]. SPPB scores were calculated using the NHATS’ SAS programming statements and range 0–12 [19]. Respondents were classified as meeting the physical function criterion if their SPPB total score was 10–12 [20,21].The presence of cognitive impairment was identified by utilizing three measures: (1) previous diagnosis of dementia or Alzheimer’s disease (AD), (2) the AD8 dementia screening interview, and (3) cognitive tests of memory, orientation, and executive function [22]. The NHATS’ SAS program statement was used to calculate cognition impairment scores and dementia classifications [23,24]. Respondents were classified as meeting the criterion of no cognitive impairment/dementia if there had been no reported diagnosis of dementia or AD, did not meet the AD8 criteria, and were not identified as having impairment in any of the cognitive tests [22].Wellbeing was assessed by 11 items adapted from the National Survey of Midlife Development in the USA, which measured three aspects of wellbeing: positive and negative affect (4 items), sense of control (4 items), and psychological wellbeing (3 items) [15]. Wellbeing scores for the current study were based on criteria used in previous research [25]. Since no cut points are available for wellbeing, respondents’ scores were divided into tertiles (1) poor (scored 1–33), (2) fair (scored 34–37), and (3) good (scored 38–41). Wellbeing scores categorized as good were considered to be indicative of good wellbeing and meeting criterion.The presence of major chronic disease was assessed by three items that asked respondents if they had been diagnosed with heart disease, diabetes, or cancer [5,14]. Respondents who answered “no” to all of these questions were classified as having no major chronic disease and as meeting the criterion [5,14].Depression was assessed by the patient health questionnaire-2, which included two items asking respondents how frequently in the previous month they had “little interest or pleasure in doing things” or “felt down, depressed, or hopeless” [14]. Depression scores ranged from 0 to 6 and scores less than 3 were classified as meeting the criterion of no depression [14,26].Anxiety was assessed by generalized anxiety disorder-2, which included two items assessing respondents’ frequency in the last 30 days regarding “felt nervous, anxious, or on edge” and “been unable to stop or control worrying”. Anxiety scores ranged from 0 to 6 and respondents were classified as meeting the criterion of no anxiety if they scored less than 3 [14,27].The IADL limitations was measured via a validated instrument that asked respondents to identify their difficulty (none, a little, some, a lot) in completing five household activities (medication tracking, doing laundry, groceries shopping, making hot meals, handling bills, and banking) or the reasons why these activities were done by or with someone else [28]. Respondents were classified as having no IADL limitations if they had no difficulty carrying out these five activities or if these activities were done by or with someone else due to reasons that were not due to respondent’s health or functioning [5,28].Health-related limitations in one’s social life were assessed by 10 questions about participation in social events in the last month and whether participation was limited due to health reasons [14,28]. Respondents were asked if their health or functioning in the past 30 days ever kept them from doing the following activities that were somewhat or very important to them: (1) visiting with friends and family; (2) attending religious services; (3) participating in clubs, classes, or other organized activities; or 4) going out to dinner, a movie, or a musical/theatrical performance. Respondents were classified as having no health-related limitation in their social life if they reported to having no restrictions with regard to any activities that was claimed as “somewhat important’ or “very important” to them [28].Function limiting pain was assessed by three items. Respondents were classified as having no function limiting pain if they were not bothered by pain and pain had not limited their activities, or if they reported that they “rarely or never” took pain medication in the last month [5,17].Overall perceived health status was assessed by one item that asked participants to rate their health status (excellent, very good, good, fair, or poor) [14]. Respondents were classified as meeting criterion if they assessed their healthy status as being “excellent”, “very good”, or “good” [5].The demographic characteristics were examined including age (65–74, 75–84, 85+), sex (male, female), race/ethnicity (White, Black, Hispanic, Others), education (high school or less, college or above), annual income (<$27,600, $27,600–41,999, $42,000–63,999, $64,000–107,999, ≥$108,000), and homebound status (homebound, semi-homebound, not homebound) [14]. Homebound status was assessed by three variables: frequency, help needed, and difficulty going outside in the last 30 days [14,29].Analytic weights were applied to all analyses, as suggested by NHATS, to reduce possible sample bias [30,31]. For baseline sample characteristics, continuous and categorical variables were tabulated by weighted mean ± standard errors and frequencies and weighted proportions (%), respectively. Differences in respondent characteristic by HAS categories (good versus poor) were examined by performing linear regression or logistic regression models. Time trend plots were performed to check the pattern of change in HAS longitudinally by baseline AWS, whereas p-values for trends and interaction terms (AWS*round) were calculated using a univariate mixed effect regression model for HAS accounting for correlation with repeated measures and utilization of weighted data with the analytic weight, which adjusts for loss to follow-up. Round was treated as a continuous variable.Additionally, adjusted βs (95% confidence intervals (CIs)) and p-values for HAS trends by baseline AWS were estimated from three mixed effects regression models to look at annual HAS change rates for three AWS categories, respectively. Adjusted odds ratios (95% CIs) and p-values were estimated using three generalized estimating equation models to look at the proportion of annual HAS category change rates for three AWS categories, respectively. Every model accounted for clustering and round was treated as a continuous variable, adjusted for age, race/ethnicity, education, annual income, and homebound status. Then, the interaction terms stratified variables*round was added into the model to examine the interaction between AWS and round, as well as to investigate whether the annual changes in HAS and HAS categories differed by baseline AWS. The mixed regression model represents an outcome variable (HAS) as a function of an intercept (β0), the predictor variable (AWS), and a random error term (intercepts and slopes). The model was specified as follows: HASij = β0j + β1j x round + β2j AWS + β3j AWS x round + βij x covariates + Eij, where HASij represents the HAS value for person (i) at time (j); β0j (= γ00 + u0j) represents the person-specific intercept or baseline HAS value; γ00 represents the fixed average intercept across all individuals; u0j is random effect term; β1j (= γ10 + u1j) represents the person-specific slope of change in HAS over round; β2j represents the person-specific slope of change in HAS with the change of AWS; β3j represents the relationship between AWS and the rate of change in HAS over round, which represents the cross-level interaction between the level 1 (within-subjects) variable, round, and level 2 (between subjects) variable, AWS; Eij represents the residual error or deviation of the observed HAS values for each person (i) at time (j).Furthermore, the temporal associations of AWS and healthy aging were examined. For HAS outcomes, β (95% CIs) and p-values were estimated from mixed models; for HAS categorical outcomes, odds ratios (95% CIs) and p-values were estimated from generalized estimating equation models. All models accounted for the correlation with repeated measures. All multiple models were adjusted for age, race/ethnicity, education, annual income, homebound status, and round. All statistical analyses were stratified by sex except descriptive analysis of demographic characteristics and conducted using SAS 9.4 (SAS Institute Inc., Cary, NC, USA) and p < 0.05 was considered to be statistically significant.The study sample was 54.8% females wherein 9.5% were 85+ years of age. The sample was primarily whites with only 16% being classified as an ethnic minority. In addition, 18.3% of the sample had a high school degree or less, 37.3% had an annual income of less than $27,600, and 2.5% were homebound. A majority of respondents (69%) were categorized as having abdominal obesity while 11.4% had normal AWS. There was a difference in HAS score with AWS categories for normal and overweight respondents more likely to be categorized as having good HAS, while respondents with abdominal obesity was more likely to be categorized as having poor HAS (see Table 2).After examining differences in demographics at baseline, the next step of the analyses was to examine HAS trends by baseline AWS in males and females, respectively, over eight years. Results indicated that the decline in HAS was significant (all p < 0.001) (see Figure 1 and Supplementary Table S2). There was also a significant increase in the proportion of respondents who classified as having poor HAS (p < 0.001). At baseline, 44.3% of males and 54% of females were classified as having poor HAS, whereas 57.3% of males and 65.3% of females were classified as having poor HAS after eight years (see Supplementary Table S2). Furthermore, as shown in Supplementary Table S2, respondents with abdominal obesity started with lower HAS (6.31 for obese vs. 6.73 for both normal and overweight in males; 5.87 for obese vs. 6.63 and 6.59 for normal and overweight in females) and remained lower than those with normal and overweight AWS over 8 years regardless of sex. Similarly, a higher proportion of respondents with abdominal obesity were classified as having poor HAS than those with normal and overweight AWS at baseline (47.2% for obese vs. 37.5% and 40.4% for normal and overweight in males; 59.2% for obese vs. 37.3 % and 40.4% for normal and overweight in females) and remained higher in the proportion of poor HAS after eight years (62.3% for obese vs. 42.1% and 52.1% for normal and overweight in males; 71.1% for obese vs. 38.9 % and 54.4% for normal and overweight in females) (see Supplementary Table S2).The analyses then examined annual change rate in the HAS and HAS categories stratified by baseline AWS over eight years (Table 3). For males whose AWS was classified as normal, for every 1-year increase in age (each round), the HAS decreased by 0.09 (β = −0.09, 95% CI: −0.12, −0.06), and the odds of the proportion of respondents being classified as having poor HAS increased by 16% (OR = 1.16, 95% CI; 1.11–1.22). Similar patterns were observed for males whose AWS was classified as overweight (HAS: β = −0.09, 95% CI: −0.11, −0.07; poor HAS: OR = 1.14, 95% CI: 1.09, 1.19) and obese (HAS: β = −0.10, 95% CI: −0.11, −0.08; poor HAS: OR = 1.15, 95% CI: 1.12, 1.18), as well as in females regardless of AWS classifications. However, there was no significant annual change rate differences by AWS in HAS and HAS categories regardless of sex.The analyses then examined the temporal association between AWS and HAS, utilizing data from all eight rounds of data collection. As shown in Table 4, the HAS for male respondents with abdominal obesity was lower than male respondents with normal AWS (β = −0.20, 95% CI: −0.30, −0.10). Results were similar in females. Respondents with abdominal obesity had lower HAS than those with normal AWS (β = −0.15, 95% CI: −0.24, −0.05). There was no difference in HAS between respondents with abdominal overweight and normal AWS.While examining the temporal association between AWS and the HAS categories that utilized AWS and HAS data from all rounds (see Table 4), male respondents who were overweight (OR = 1.26, 95% CI: 1.08, 1.48) or obese (OR = 1.52, 95% CI: 1.29, 1.79) were more likely to be classified as having poor HAS than respondents with normal AWS. For females, respondents who were overweight (OR = 1.16, 95% CI: 1.01, 1.35) or obese (OR = 1.42, 95% CI: 1.21, 1.66) were also more likely to be classified as having poor HAS than those with normal AWS.This is the first study, to our knowledge, that has examined the temporal associations between AWS defined by WC and a comprehensive assessment of healthy aging in a nationally representative sample of older adults in the US. Study findings indicate that HAS in all AWS categories decreased annually but the rate of annual HAS decline in respondents with overweight or obese AWS was not significantly different than those with normal AWS. However, respondents with abdominal obesity always had lower HAS, and respondents whose AWS was classified as overweight and obese were more likely to have poor HAS than those with normal AWS.While a number of previous studies investigated health indicators among older adults—such as physical function and dementia [7,8,9,32,33,34,35]—the present study advances previous research by creating a more comprehensive measure of healthy aging that included measures of physical/cognitive function, wellbeing, major chronic diseases, mental health, function limitation in household activities, and health-related limitations in social life, function limiting pain, and perceived health status [2,5]. When HAS were evaluated longitudinally, they decreased steadily regardless of AWS and sex. Although annual decrease in HAS (ranging from 0.08 to 0.10) is relatively small, the accumulation of this change overtime could have a significant impact on the trajectory of healthy aging. This finding aligns somewhat with prior research that examined certain aspects of healthy aging, such as limitations in social life, cognitive impairment, IADL, and pain [32,33,35,36]. However, it is difficult to make comparisons between the current study and previous studies due to different analytical approaches, as the current study examined annual change rate of the comprehensive assessment of healthy aging whereas previous studies did not. Nevertheless, the present study adds to the literature with important information on a comprehensive assessment of HAS changes over an eight-year period by baseline AWS in a nationally representative sample of adults aged 65+ in the US. Moreover, the use of WC is a unique strength of the present study. WC is a surrogate marker of central adiposity while BMI is simply a weighted ratio of height and body mass [9,10]. Previous studies have reported that WC is a marker of visceral fat associated with cardiometabolic risk [37], and that WC is also more closely related to morbidity and mortality in older adults than BMI [9]. This distinction is particularly important as older adults are at risk for sarcopenia-related muscle loss, which results in a loss of functional mass due to aging [11]. In addition, older adults are at risk for bone loss, which contributes to a loss of functional mass related to aging [10]. Both muscle and bone loss could result in healthier weight classification based on BMI despite increased relative adiposity and mortality [10,11]. Nonetheless, due to these differences between WC and BMI, caution needs to be applied when comparing the present study’s findings to findings of studies that used BMI as opposed to WC.The present study is also unique in its use of longitudinal AWS data while examining its associations with healthy aging overtime. AWS is important although AWS does not modify HAS annual change rate overtime. Respondents with abdominal obesity were more likely to have a lower HAS and to be classified as having poor HAS than those with normal AWS, regardless of sex. This finding is in agreement with previous studies [7,8,38] and the current study adds to the literature by examining the longitudinal association between AWS defined by WC and a comprehensive assessment of healthy aging. The high proportion of respondents (69%) classified as obese by AWS and the deleterious effect of this classification on HAS, indicate the need for strategies to prevent abdominal fat accumulation in older adults. It is worth noting that although there was no significant difference in HAS between respondents with overweight and normal AWS, respondents with overweight AWS were more likely to be classified as having poor HAS than those with normal AWS. This finding is not consistent with studies that used BMI to determine weight status that found being overweight is associated with lower mortality risk in older adults [39]. Nevertheless, the possible explanation for the difference observed in poor HAS classification between respondents with overweight and normal AWS is that being abdominally overweight after age 65 might still pose a risk to healthy aging in older adults. Difference may also be due to the precision of measurement although in this study, WC measurement was conducted by trained assessors using the standardized protocol [17]. Further research is warranted to examine possible healthy aging related factors in a large cohort of older adults to better understand the differential effect of AWS and BMI categorization on the trajectory of healthy aging.The present study has several strengths. First, to our knowledge, it is the first study to utilize a comprehensive definition of healthy aging to examine healthy aging trends and its association with AWS in older adults using representative data in the US. The longitudinal data collected over eight years allowed us to examine the holistic aspect of healthy aging overtime. In addition, the present study used WC to define AWS, which is a more appropriate measure in older adults [9]. Moreover, the present study adjusted for important confounders that have been identified as risk factors, including homebound status that has not been adjusted in prior research [14,29]. However, there could be other residual confounding factors not included that might bias our results. Some of the criteria used to create the HAS were based on self-reported measures (e.g., anxiety, depression, IADL), although these measures were validated [25,26,27,28]. Another study limitation is that WC is not as accurate as dual energy X-ray absorptiometry when measuring abdominal visceral fat [40]. However, WC has consistently been associated with abdominal fat measured via dual energy X-ray absorptiometry in older adults [41] and is an efficient and cost-effective measurement for general clinical assessment as well as for use with large cohort studies [9]. Furthermore, 84% of the sample were white, which presents a limitation for the generalization of our findings to a broader population. It is also possible that our sample might be skewed to those who were healthier and lived longer.Findings from the present study indicated that AWS was associated with healthy aging but did not modify the annual rate of change for HAS overtime. However, these findings still highlighted the importance of AWS because respondents with abdominal obesity had lower HAS at all points over an eight-year period compared to respondents with normal AWS. Results also indicated that both respondents with overweight and obese AWS were more likely to be categorized as having poor HAS than respondents with normal AWS. These findings indicate that abdominal obesity decreases the likelihood for successful aging and the effect of having overweight AWS on healthy aging is inconsistent. Study results suggest that, moving forward, studies examining weight in older adults should use WC defined weight status rather than BMI.The following are available online at https://www.mdpi.com/1660-4601/17/16/5656/s1, Table S1: Study design summary, Table S2: Healthy aging score trend by baseline AWS, NHATS 2011–2018.All authors contributed to the study design. F.X. completed the analyses with guidance from G.W.G. and S.A.C.; F.X., J.E.E., G.W.G., and M.L.G. drafted the manuscript; All authors have read and agreed to the published version of the manuscript.The present study is a secondary data analysis using data from NHATS. No funding was received for this study.The present study would not have been possible without publicly available data from NHATS which is being led by the Johns Hopkins University Bloomberg School of Public Health in collaboration with the University of Michigan. We would like to thank all their research team members for their time and effort and their generosity to share NHATS data publicly.The authors have no conflicts of interest to declare.Healthy aging score trends by baseline abdominal weight status, National Health and Aging Trends Study 2011–2018.Health indicators and its criteria for the definition of healthy aging.Note: SPPB = short physical performance battery; PHQ = patient health questionnaire; GAD = generalized anxiety disorder; IADL = instrumental activity of daily life; HRLS = health-related limitation in social life.Baseline characteristics of respondents stratified by HAS classification, NHATS 2011.Note: continuous and categorical variables were tabulated by weighted mean ± standard errors and frequencies and weighted proportions (%), p-values for continuous variables obtained by performing linear regression model, whereas logistic regression model for categorical variables. NHATS = National Health and Aging Trends Study; IADL= instrumental activities of daily living; HRLS = health-related limitation in social life; HAS = healthy aging score. $ Good HAS, defined as above the median; poor HAS defined as below the median. WC = waist circumference: # normal (WC < 37 inches/31.5 inches), overweight (37 inches/31.5 inches ≤ WC < 40 inches 35 inches), and obese (WC ≥ 40 inches/35 inches) in males and females, respectively, * p < 0.05.Annual changes in HAS and categories stratified by baseline AWS, NHATS 2011–2018.Note: & adjusted β (95% confidence intervals (CIs)) CI) and p for trend were estimated from mixed effects regression models. @ Adjusted odds ratio (OR) and p for trend were estimate using generalized estimating equation models; all the models accounted for clustering and round was treated as a continuous variable (1–8) adjusted by age, race/ethnicity, education, annual income, and homebound status. # The interaction terms stratified variables*round was added into the model to examine the effect of the interaction between stratified variables and round to investigate whether the changes over years in prevalence or mean differed between the stratified variables. NHATS = National Health and Aging Trends Study; AWS= abdominal weight status; HAS = healthy aging score. $ Good HAS defined as above the median whereas poor HAS defined as below the median; * p < 0.05.Temporal associations between AWS and HAS, NHATS 2011–2018.Note: # β (95% confidence intervals (CIs)) and p-values were estimated from mixed models; & odds ratios (ORs) (95% CIs) and p-values were estimated from generalized estimating equation models. All models were accounted for the correlation with repeated measures, and all analyses adjusted for age, race, education, annual income, homebound status, and round. NHATS = National Health and Aging Trends Study; AWS = abdominal weight status; HAS = healthy aging score. $ Good HAS defined as above the median whereas poor HAS defined as below the median, * p < 0.05.