Fluent Essays

RESEARCH Open Access The relationship between physical activity, mental wellbeing and symptoms of mental health disorder in adolescents: a cohort study Sarah Louise Bell1* , Suzanne Audrey1, David Gunnell1,2, Ashley Cooper2,3 and Rona Campbell1 Abstract Background: Mental illness is a worldwide public health concern. In the UK, there is a high prevalence of mental illness and poor mental wellbeing among young people. The aim of this study was to investigate whether physical activity is associated with better mental wellbeing and reduced symptoms of mental health disorder in adolescents. Methods: A cohort of 928 12–13 year olds (Year 8) from six secondary schools in England, who had participated in the AHEAD trial, ‘Activity and Healthy Eating in Adolescence’, were followed up three years later (when 15–16 years old, Year 11). At baseline, physical activity was measured using accelerometers. At follow-up, mental wellbeing was measured using the ‘Warwick Edinburgh Mental Wellbeing Scale’ (WEMWBS) and symptoms of mental health disorder using the ‘Strengths and Difficulties Questionnaire’ (SDQ). Multivariable linear regression analyses were used to investigate associations between physical activity and both mental wellbeing and symptoms of mental health disorder. Results: 794 (86%) of the eligible 928 young people provided valid accelerometer data at baseline. 668 (72%) provided complete mental wellbeing data and 673 (73%) provided complete symptoms of mental health disorder data at follow-up. The multivariable analyses showed no evidence of an association between physical activity volume (counts per minute (cpm)) or intensity (Moderate to Vigorous Physical Activity (MVPA)) and mental wellbeing (WEMWBS overall score) or overall symptoms of mental health disorder (SDQ Total Difficulties Score). However, higher levels of physical activity volume at age 12–13 years were associated with lower scores on the emotional problems subscale of the SDQ at age 15–16 years. Conclusions: This cohort study found no strong evidence that physical activity is associated with better mental wellbeing or reduced symptoms of mental health disorder in adolescents. However, a protective association between physical activity and the emotional problems subscale of the SDQ was found. This suggests that physical activity has the potential to reduce symptoms of depression and anxiety in adolescents. Future cohort study designs should allow for repeated measures to fully explore the temporal nature of any relationship. Keywords: Physical activity, Mental wellbeing/mental health/mental illness/mental health disorder, Adolescents/ young people, Cohort study © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. * Correspondence: [email protected] 1Population Health Sciences, Bristol Medical School, University of Bristol, Bristol BS8 2PS, UK Full list of author information is available at the end of the article Bell et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:138 https://doi.org/10.1186/s12966-019-0901-7 http://crossmark.crossref.org/dialog/?doi=10.1186/s12966-019-0901-7&domain=pdf http://orcid.org/0000-0003-1181-9591 http://creativecommons.org/licenses/by/4.0/ http://creativecommons.org/publicdomain/zero/1.0/ mailto:[email protected] Background Mental illness is a worldwide public health concern [1]. It is currently the largest single cause of disability in the UK representing an estimated 28% of the total disease burden (compared to 16% each for cancer and heart disease) [2]. The World Health Organization (2013) estimates that, worldwide, 20% of adolescents in any given year may experience mental illness. In England, the most recent population survey (2017) re- ported 14.4% (1 in 7) of young people aged 11–16 years were identified with a mental health disorder [3]. Emotional disorders (present in 9%) were the most common type at this age followed by behav- ioural (conduct) disorders (6.2%) [3]. Mental health disorder has diverse and long-term negative effects on individuals, their families, and wider society [4]. Population surveys have also found increased levels of low wellbeing in young people [3]. Mental wellbeing is conceptualised as more than the absence of mental illness [5]. It has been described as encompassing hedonic (happiness, life satisfaction, and affect) and eudaimonic (positive functioning, sense of purpose, and self-acceptance) wellbeing [6–10]. Mental wellbeing is protective for a range of health outcomes [11–14] and found to be associated with higher educational outcomes in adolescence and better occupational functioning in adulthood [15–17]. Correlates of young peoples’ mental illness and mental wellbeing are reported to be largely distinct, stressing the importance of considering these concepts separately and avoiding their conflation [18]. While mental illness and mental wellbeing may be re- lated, they are not necessarily distinct ends of a con- tinuum [19–21]. The dual continuum model views mental illness (or mental health disorder) and mental health (or mental wellbeing) as two separate continua rather than as opposite ends of the same continuum [20]. Keyes and Lopez (2002) depicted the dual con- tinuum model of mental illness and mental health and described four states: struggling (incomplete mental ill- ness i.e. mental illness and high wellbeing), floundering (complete mental illness i.e. mental illness and low well- being), languishing (incomplete mental health i.e. no mental illness and low wellbeing), and flourishing (complete mental health i.e. no mental illness and high wellbeing). A large number of adolescents are thought to suffer from poor mental wellbeing despite being free from mental illness [4, 22]. Therefore, promoting mental wellbeing alongside preventing and treating the symp- toms of mental illness, is a growing priority. The recent Green Paper (2018) focuses on schools finding low cost and low risk interventions to promote mental wellbeing and prevent symptoms of mental health disorder [23]. Although there is evidence of physical activity improv- ing mental wellbeing [24, 25] and having the potential to prevent symptoms of mental health disorder [26, 27] in adults, the evidence of any relationship in adolescents is weaker. The studies lack measurement consistency, hav- ing defined and assessed physical activity, mental well- being, and symptoms of mental health disorder in a variety of ways. Furthermore, few studies have used a multi-dimensional measure of mental wellbeing or symptoms of mental health disorder (most capture only one component of mental wellbeing such as self-esteem [28] or self-efficacy [29] or a specific mental health prob- lem such as depression [30–34]), and studies that have used an objective measure of physical activity to assess any relationship are limited [35–37]. Several reviews have attempted to analyse any associ- ation in young people [38–51]. A review of reviews by Biddle et al. (2011) [48] showed that physical activity has beneficial effects on mental health in children and ado- lescents. More recently there has been a significant in- crease in the number and quality of studies exploring any association, and when the review of reviews was up- dated in 2019 [49], physical activity continued to be shown to be associated with certain mental health out- comes in young people (a causal association was found with cognitive functioning, a partial association for de- pression, no association for self-esteem, and research fo- cusing on the association of physical activity with anxiety was reported to be variable but generally showed small beneficial effects) [49]. A review by Rodriguez- Ayllon et al. (2019) [50] analysed the effects of physical activity interventions (randomised controlled trials and non-randomised controlled trials) on mental health out- comes of adolescents and also synthesised the observa- tional evidence (both longitudinal and cross-sectional). Their review included studies that had at least one ‘psy- chological illbeing’ (i.e. depression, anxiety, stress or negative effect) and/or ‘psychological wellbeing’ (self-es- teem, self-concept, self-efficacy, self-image, positive affect, optimism, happiness and satisfaction with life) outcome. They concluded that there was a small positive effect of physical activity interventions on mental health outcomes in adolescents [50]. The SDQ is a unique screening tool of symptoms indi- cating overall mental health disorder in young people. This composite measure identifies symptoms of emo- tional problems, hyperactivity, and behavioural/conduct problems and has been used in the UK series of surveys of the mental health of children and young people (1999, 2004 and 2017) [3] alongside other significant population surveys [52]. Despite this, few studies have used this measure when looking at the association be- tween physical activity and symptoms of mental health disorder [31, 53–55]. Of the studies identified, neither of the two longitudinal studies used an objective measure of physical activity [31, 53]. Bell et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:138 Page 2 of 12 The WEMWBS is a relatively new scale, also used in the series of surveys in the UK [3], designed to capture popu- lation mental wellbeing. The reviews that have included studies assessing the relationship between physical activity and various aspects of mental wellbeing [38–44; 46–50] have concluded that there is evidence of promise, but fur- ther studies are needed that use a multi-dimensional measure such as the WEMWBS. Given the limitations of the evidence base, the aim of this study was to determine whether physical activity is associated with mental wellbeing and symptoms of men- tal health disorder in adolescents. This is the first study to investigate any potential relationship longitudinally using an objective measure of physical activity and valid and reliable self-report measures of both mental well- being, using the WEMWBS, and symptoms of mental health disorder, using the SDQ, in adolescents. Methods A prospective cohort was formed based on the 928 par- ticipants from six secondary schools in the South West of England who took part in a two-year school-based ex- ploratory randomised controlled trial (RCT) of an Activ- ity and Healthy Eating intervention for use in ADolescence: the AHEAD trial [56]. All state secondary schools in the selected local authorities were invited to participate in the study. Schools first to express an inter- est (ensuring variation in size, geographical area, Ofsted rating, Free School Meal entitlement, and achievement rating) were recruited to the study. Physical activity was measured in 2008 when the participants were aged 12– 13 years (Year 8); mental wellbeing and symptoms of mental health disorder was measured three years later (2011) when the participants were aged 15–16 years (Year 11). The inclusion criteria were participation in the AHEAD trial and continued attendance at a study school three years later. Data collections at baseline (2008) and follow-up (2011) were conducted by a team of researchers in the schools (classrooms or school halls) during a usual lesson (approximately 60 min). There was no evidence of promise that the AHEAD intervention improved physical activity or diet. Physical activity measure The ActiGraph GT1M accelerometer (ActiGraph, LLC, Penscola, FL) was used to measure physical ac- tivity. Participants were instructed in the use of the accelerometers in school, and then asked to wear the instrument for seven days during waking hours, ex- cept for water-based activities such as bathing and swimming. Accelerometer data were downloaded using ActiLife software (Lifestyle Monitor System software Version 3.3.0) and processed using Kinesoft software (Version 3.3.62; Kinesoft, Saskatchewan, Canada) to generate outcome variables (10 s epochs were used to capture the sporadic nature of adoles- cent physical activity). Physical activity volume was computed as mean accelerometer counts per minute (cpm), and physical activity intensity (mean minutes per day of Moderate to Vigorous Physical Activity (MVPA)) was computed using established thresholds [57]. A valid day of measurement was defined as re- cording at least 480 min (8 h) of data (monitoring period from 7 am until 11 pm; periods of ≥60 min of consecutive zeros, with allowance for 2 min of inter- ruption, was classed as nonwear time) and at least three valid days were required for inclusion in analyses. Mental wellbeing and symptoms of mental health disorder measures The ‘Warwick Edinburgh Mental Wellbeing Scale’ (WEMWBS) [58], validated for use in adolescents aged 13–16 years, was used to measure mental wellbeing. The WEMWBS has 14 positively worded items with a five- point Likert scoring scale for each item (with scores ran- ging from 1 = none of the time to 5 = all of the time). The responses to each item were summed to give an overall WEMWBS score; a minimum score of 14 (i.e. poor mental wellbeing) and a maximum of 70 (i.e. good mental wellbeing). The higher score, indicative of better mental wellbeing, reflects more positive thoughts, feel- ings and behaviours. Where scores for three or less items were missing the mean value of responses for completed items for that individual was used to replace the score for missing items, enabling a total score for that individual to be computed [59]. If more than three items were missing the data for that participant were not used. The WEMWBS was selected due to it being the first multi-dimensional scale to measure population mental wellbeing in adolescents, based on established in- dicators. It covers most aspects of mental wellbeing in- cluding both hedonic and eudaimonic perspectives and is suitable for looking at the relationship between phys- ical activity and mental wellbeing. The ‘Strengths and Difficulties Questionnaire’ (SDQ) [60] was used to measure symptoms of mental health disorder. It is a behavioural screening tool used to assess social, emotional, and physical aspects of behaviour in young people [61] and has been shown to be valid and reliable for completion by 11–16 year olds [62]. The questionnaire has 25 items which comprise five sub- scales: (i) emotional symptoms (anxiety and depressive symptoms); (ii) conduct problems; (iii) hyperactivity/in- attention; (iv) peer relationship problems; and (v) pro- social behaviour (positive behaviours such as being kind and helpful, scored in reverse of the other subscales). Re- sponse options are ‘not true, somewhat true, or certainly Bell et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:138 Page 3 of 12 true’ (scored 0, 1 or 2). The SDQ ‘Total Difficulties Score’ (SDQ TDS) was generated by adding together the scores from the first four subscales and can range from 0 (low difficulties) to 40 (high difficulties). The five sub- scales whose scores can vary from 0 to 10 were also in- vestigated independently. Items scores which were missing were imputed only if at least three out of five items were complete on each subscale. In this case the total subscale score for each participant was divided by the number of complete items to get the mean score and used to replace the missing item score [60, 63]. If more than two items were missing from any sub scale the data for that subscale for that participant were not used. Participants required a score for each of the SDQ subscales to be able to compute their SDQ TDS. The SDQ was selected as the SDQ TDS provides a useful in- dicator of the level of symptoms of mental health dis- order overall. Furthermore, the subscale items may be used to indicate specific clinical disorders in adolescents: depression, anxiety, hyperactivity attention deficit dis- order (ADHD) and behavioural/conduct disorder. The SDQ is a useful screening tool for identifying young people with raised scores thus potentially at risk. Possible confounders and mediators The self-report behavioural questionnaires recorded a number of potential confounders and mediators of any potential relationship: age; gender; ethnicity; socioeco- nomic status (SES) (measured using the ‘Family Afflu- ence Scale’ FAS II [64]); study school; number of daylight minutes (a proxy for season); baseline symp- toms of mental health disorder (SDQ TDS); sleep (fre- quency of feeling tired when going to school in the morning); number of friends; belonging to teams and clubs; smoking; drinking alcohol; and intervention arm of the AHEAD trial [56] (the participants were rando- mised into two groups - the intervention arm received a physical activity and healthy eating intervention and the control arm continued with usual practice). Confounders included were determined by the con- struction of a directed acyclic graph (DAG) and the availability of relevant data on study participants. Accel- erometer wear time was computed from the participants accelerometer data. Data from the self-report behavioural questionnaires were entered into a secure Access database and the ac- celerometer data were stored as anonymised files on a secure drive. All analyses were conducted using Stata 13 MP [65]. Statistical analysis The analyses assessed the association between physical activity and the measures of mental wellbeing and symp- toms of mental health disorder. Multivariable linear regression analyses were used to estimate exposure ef- fects controlling for potential confounders and media- tors which were investigated by grouping them as clusters of related factors in the models: i.e. socioeco- nomic factors (ethnic group, SES, study school); factors that may influence the physical activity data processing (daylight minutes (when volume or MVPA exposure), minutes of wear time (when MPVA exposure only); life- style factors (sleep, friends, belonging to teams or clubs, drinking alcohol, smoking; measured at follow-up only); baseline symptoms of mental health disorder); and then by producing a fully adjusted model containing all of these factors. Where there was evidence of confounding or mediation (associations weakened or enhanced), fur- ther models were fitted to investigate this in more detail. Confounders adjusted for were determined by the avail- ability of relevant data on study participants. This some- what crude epidemiological approach was used due to there being no clear evidence of associations (and whether confounders or mediators in the relationship) in the literature. Coefficients represent the linear rela- tionship- change in WEMWBS overall score, SDQ TDS or SDQ subscale score per unit increase in physical ac- tivity (volume or intensity). Physical activity volume was defined by accelerometer counts (a dimensionless output from the accelerometer) per minute of recording (com- puted as total counts recorded divided by the total mi- nutes of valid recording over the measurement period, described as counts per minute (cpm)), whilst physical activity intensity was defined as daily minutes of Moder- ate to Vigorous Physical Activity (MVPA). The relation- ship was quantified as the change in mental wellbeing or symptoms of mental health disorder score associated with an increase of 100 cpm (e.g. an increase from 508 cpm to 608 cpm, an approximately 20% increase in phys- ical activity volume from baseline mean of the sample); or an additional 60 min of daily MVPA. Tests for inter- actions were carried out to investigate whether observed associations differed by gender. There was no evidence that associations between physical activity (volume and MVPA) and either mental wellbeing or symptoms of mental health disorder (WEMWBS and SDQ) differed in males and females. The test for interaction p-value ranged from p = 0.19–0.97 so all models were based on data for males and females combined. Ethics approval and consent to participate The University of Bristol Faculty of Medicine and Den- tistry Ethics Committee gave full approval in 2007 for the AHEAD feasibility study and pilot trial (reference number 060702) and in 2011 for the cohort study (refer- ence number 101119). Consent procedures were the same in the original feasibility study and pilot trial and in the subsequent Bell et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:138 Page 4 of 12 cohort study. Firstly, written consent to participation was sought from each schools’ headteacher. Secondly, letters were posted by school staff to the parents/carers of all eligible school pupils explaining the study and en- closing a reply slip to be returned if parents/carers did not want their child to participate. This ‘opt-out’ method of consent has been found to be an ethical and appropri- ate procedure in low-risk prevention research and avoids the low response rates and potential sampling bias when opt-in parental consent procedures are used [66, 67]. At all data collections, the young people were provided with information about the study and informed that they could ‘opt-out’ of some or all the study activities at any point and were asked to sign individual assent forms. Results Cohort study profile and baseline characteristics 794 (86%) of the 928 pupils provided complete, valid baseline physical activity data and were followed-up three years later to complete mental wellbeing and symptoms of mental health disorder measures. 673 (73%) completed the SDQ and 668 (72%) completed the WEMWBS at follow-up. Those lost to follow-up were more likely to be older, male and from one particular school (due to a new headteacher using alternative edu- cational placements for a large number of the school’s more challenging pupils). Table 1 shows the baseline characteristics of the participants included and excluded from the cohort. Figure 1 displays the study profile for the cohort study and Table 2 describes the baseline characteristics of the participants. Physical activity at baseline At baseline, overall physical activity volume (mean (SD) counts per minute (cpm)) was 508.3 cpm (169.42) and participants recorded 55.6 (21.5) (mean (SD)) daily mi- nutes of MVPA). Females were less active than males with regard to both physical activity volume (mean dif- ference 86.39 cpm (95% CI 111.2 cpm to 61.6 cpm, p < 0.001) and intensity (mean difference 11.7 min (95% CI − 15.41 to − 8.82), p < 0.001). Mental wellbeing and symptoms of mental health disorder at follow-up There was a negative association between the WEMWBS overall score and the SDQ TDS (r = − 0.41) at follow-up. This relatively weak correlation indicates the scales are measuring different things (the WEMWBS overall score only accounts for 16% of the total variation in the SDQ TDS). At follow-up, the participants’ WEMWBS overall mean (SD) score was 48.74 (8.66) (data from a similar study (13–16 year olds) 48.8 (8.66) [58]; females (n = 342) 46.93 (8.90) and males (n = 326) 50.63 (7.99) with strong evidence of a gender difference in WEMWBS overall score (mean difference in WEMWBS overall score − 3.70 (95% CI − 4.99 to − 2.24) p < 0.001). Females had a lower WEMWBS overall score indicating poorer mental wellbeing than males. The participants’ SDQ TDS mean score was 12.17 (5.56) (normative data (for 11–15 years) 10.3 (5.2)) [60]; females (n = 343) 12.66 (5.36) and males (n = 330) 11.66 (5.73). Again there was evidence of a gender difference in SDQ TDS (mean dif- ference in SDQ TDS 1.00 (95% CI 0.16 to 1.84) p = Table 1 Baseline characteristics of participants included and excluded from the cohort study Baseline characteristics Excluded (n = 242*) n (%) Included (n = 673) n (%) Difference in proportion between excluded versus included categories** Gender Males 149 (31.1) 330 (68.9) X2 = 11.21 (p = 0.001) Females 93 (21.0) 343 (79.0) Ethnicity White 217 (25.7) 629 (74.3) X2 = 3.67 (p = 0.06) Other 25 (36.0) 44 (64.0) FAS Low and medium (0–5) 121 (25.5) 353 (74.5) X2 = 2.09 (p = 0.15) High (6–9) 87 (21.4) 320 (87.6) Study school 1 40 (22.0) 142 (78.0) X2 = 78.04 (p < 0.001) 2 58 (33.3) 116 (66.7) 3 31 (18.8) 134 (81.2) 4 15 (15.6) 81 (84.4) 5 28 (16.5) 142 (83.5) 6 70 (54.7) 58 (45.3) Physical activity*** Volume (mean counts per minute) 482.78 (158.12) 508.28 (169.42) −25.5 (95% CI −58.02 to 7.02) p = 0.12 MVPA (mean daily MVPA minutes) 51.90 (21.66) 55.59 (21.47) −3.69 (95% CI −7.86 to 0.48) p = 0.08 *13 pupils were excluded but did not have baseline characteristics data to compare **Pearson’s chi-squared tests ***Physical activity measures in excluded group n = 121 due to missing data (t-test mean difference) Bell et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:138 Page 5 of 12 0.02). Females had a higher SDQ TDS indicating higher symptoms of mental health disorder than males. Main findings The univariable (adjusted for gender, age and interven- tion arm of the AHEAD trial) and multivariable analyses showed no evidence of an association between physical activity (volume or intensity) and mental wellbeing (WEMWBS overall score) or overall symptoms of men- tal health disorder (SDQ TDS) (Table 3). When the five SDQ subscales were analysed independently, an associ- ation was found between both physical activity volume and intensity and the emotional problems subscale of the SDQ (scale range 0–10). However, the association found between MVPA and the emotional problems sub- scale of the SDQ was slightly attenuated in the fully ad- justed model (when controlling for symptoms of mental health disorder at baseline (SDQ TDS)), such that confi- dence intervals included the null value. For physical activity volume, a mean increase of 100 cpm (~ 20% increase in physical activity volume) was as- sociated with a decrease in the emotional problems sub- scale score of 0.12 (unadjusted model) and 0.11 (fully adjusted model). The confidence intervals (95% CI) ranged from − 0.23 to − 0.00 in the fully adjusted model implying that a potential reduction in the emotional problems subscale score of 0.23 could be achieved with an additional 100 mean cpm of physical activity (the ex- treme end of the effect estimate). For physical activity in- tensity, an additional 60 min of mean daily MVPA was associated with a decrease in the emotional problems subscale score of 0.54 (unadjusted model) and 0.49 (fully adjusted model; however confidence intervals crossed zero). Fig. 1 Study profile for cohort study Bell et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:138 Page 6 of 12 Discussion Main findings To the best of our knowledge, this is the first longitu- dinal study to investigate the relationship between phys- ical activity, mental wellbeing and symptoms of mental health disorder in adolescents. We also believe this is the first study to use an objective measure of physical activity (accelerometers) and composite measures of both mental wellbeing (WEMWBS) and symptoms of mental health disorder (SDQ) validated for use with young people alongside each other, to investigate any potential associations. We found no evidence of an asso- ciation between physical activity (volume or MVPA) and mental wellbeing (WEMWBS overall score) or overall symptoms of mental disorder (SDQ TDS). However, a protective association was found between physical activ- ity volume and the emotional problems subscale of the SDQ. This finding suggests that increasing physical ac- tivity volume in adolescents may have the potential to reduce their risk of emotional problems (items on the Table 2 Baseline characteristics of participants in the cohort study Continuous variables All Females Males n Mean (SD) n Mean (SD) n Mean (SD) Age in years 673 12.69 (0.34) 343 12.68 (0.34) 330 12.70 (0.34) SDQ Total Difficulties Score (SDQ TDS) 637 12.41 (5.76) 328 12.33 (5.67) 309 12.50 (5.87) Physical activity volume (mean counts per minute (cpm)) 673 508.28 (169.42) 343 465.93 (153.25) 330 552.31 (174.35) Physical activity intensity (mean daily MVPA minutes) 673 55.59 (21.47) 343 49.85 (19.24) 330 61.56 (22.07) Categorical variables n Category All Females Males n (%) n (%) n (%) Gender 673 343 (51.0) 330 (49.0) Ethnicity 673 White 629 (93.0) 320 (93.3) 309 (93.6) Other 44 (7.0) 23 (6.7) 21 (6.4) Family Affluence Scale (SES) 673 Low and Med (0–5) 353 (52.0) 187 (54.5) 166 (50.3) High (6–9) 320 (48.0) 156 (45.5) 164 (49.7) Study school 673 1 142 (21.0) 79 (23.0) 63 (19.1) 2 116 (17.0) 64 (18.7) 52 (15.8) 3 134 (20.0) 62 (18.1) 72 (21.8) 4 81 (12.0) 41 (12.0) 40 (12.1) 5 142 (21.0) 59 (17.2) 83 (25.2) 6 58 (9.0) 38 (11.1) 20 (6.1) Intervention arm of the AHEAD trial 673 Intervention 339 (50.4) 184 (54.3) 155 (45.7) Control 334 (49.6) 159 …