Cohort Profile: Resilience, Ethnicity and AdolesCent mental Health (REACH)

Cohort Profile: Resilience, Ethnicity and AdolesCent mental Health (REACH) Gemma Knowles, Charlotte Gayer-Anderson , Rachel Blakey, Samantha Davis, Katie Lowis, Daniel Stanyon, Aisha Ofori, Alice Turner, Lynsey Dorn, Schools Working Group, Young Persons Advisory Group, Stephanie Beards, Vanessa Pinfold, Ulrich Reininghaus, Seeromanie Harding, Craig Morgan* Health Service and Population Research Department, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, UK, ESRC Centre for Society and Mental Health, King’s College London, London, UK, National Children’s Bureau, London, UK, McPin Foundation, London, UK, Department of Public Mental Health, Central Institute of Mental Health, Medical Faculty Mannheim, University of Heidelberg, Mannheim, Germany and Department of Nutritional Sciences, School of Life Course Sciences, Faculty of Life Sciences & Medicine, King’s College London, London, UK

Why was the cohort set up? REACH (Resilience, Ethnicity, and AdolesCent Mental Health) is a school-based accelerated cohort study in innercity London, UK. REACH comprises three cohorts which were established to provide detailed and extensive information on the nature, distribution, and determinants of mental health among young people from diverse backgrounds and densely populated inner-city areas. Specifically, REACH is designed to test several hypotheses concerning: (i) the extent, nature, and development of mental health problems among young people; (ii) variations by gender, ethnic group, and socioeconomic status; (iii) risk and protective factors; and (iv) mechanisms linking risk/protective factors and mental health. The first phase of REACH, early-to-mid adolescence , was funded by the European Research Council. The second phase (2020-21), funded by United Kingdom Research and Innovation (UKRI), examines the impacts of the covid-19 pandemic. The third phase , funded by the UK Economic and Social Research Council, will examine the transition to adulthood.
Mental health problems are a major public health issue, with an estimated lifetime prevalence of around 25%. 1 In the UK, depressive and anxiety disorders rank among the six leading causes of disability among men and women, and substance use disorders rank among the 10 leading causes among men. 2 In 2009/10, the estimated total economic and social costs of mental ill health in England was £105.2 billion. 3 There is an urgent need to better understand the development of mental health problems and how to prevent them.
Adolescence is a critical period in the development of mental health problems. Around 50% of mental health problems begin by age 14 and 75% by age 24. 4 Those who develop recurring or persistent problems during adolescence are at increased risk of a range of adverse social, economic, and health outcomes later in life. [5][6][7] In England, the most recent national survey suggests a prevalence of around 14% among those aged 11-16 years. 8 Data from national surveys and from other major cohort studies in the UK suggest that, for some groups of young people and some types of problems-most notably, depression and anxiety among young women aged 16-24 years-prevalence has increased over the past 15-20 years. 8,9 However, for some groups and other types of problems, the data suggest little change over time 8 -which is somewhat surprising against a backdrop of rapid and far-reaching social change (e.g. recession and austerity, rapid increases in the use of mobile technologies, social media, etc.).
Invariably, the most disadvantaged and at-risk groups in society-i.e. most minority ethnic groups, those who grow up in poverty and in challenging circumstances, etc.-are under-represented in national and other largescale studies. 8,10,11 Indeed, few prospective studies in the UK are sufficiently powered to make meaningful inferences about mental health trends and trajectories among those from minority ethnic groups. This is important because mental health is intimately connected to social, economic, and environmental contexts and experiences. Rates vary by geographical location, socioeconomic status, and ethnic group, 2,12 and risk is strongly associated with adverse experiences-discrimination, maltreatment, exposure to crime and violence, etc-which disproportionately affect many minority ethnic groups and the poorest in society. [13][14][15][16] To inform interventions and service provision, large studies in diverse urban areas are required. 12 To the best of our knowledge, REACH is the largest and most comprehensive contemporary study of mental health among young people from diverse inner-city areas in the UK. The highly diverse and representative REACH cohorts are drawn from two of the most densely populated and socioeconomically and ethnically diverse boroughs in England, [17][18][19] Lambeth and Southwark, London. These boroughs consistently rank among the 20% most deprived boroughs in the country. [17][18][19] The prevalence of adult mental disorders is around two times higher in these boroughs compared with national estimates. 12 REACH provides important new data about the development and trajectories of mental health problems in diverse groups and investigates why, despite similar experiences and circumstances, some young people develop mental health problems whereas others do not. In doing so, REACH will inform the development of interventions to promote mental health and prevent mental health problems in young people from all backgrounds.

Study design
REACH is an accelerated cohort study ( Figure 1) comprising three cohorts of young people recruited from 12 statefunded mainstream secondary (high) schools in Lambeth and Southwark, London, UK. The cohorts were recruited and first assessed at age 11-12 years (Cohort 1, school year 7), 12-13 years (Cohort 2, school year 8) and 13-14 years (Cohort 3, school year 9), i.e. in Key Stage 3. Schools were selected to be representative of the 38 mainstream secondary schools within the two boroughs based on (i) the proportion of students eligible for free school meals (a marker of household socioeconomic disadvantage) and (ii) the proportion of students in minority ethnic groups. In the 12 participating schools, all students in school years 7-9 were invited to participate (n ¼ 4945).
In line with similar studies, 20-22 informed consent (for those aged 16 and over) and assent (for those aged under 16) was obtained from all participants using the following procedures. Approximately 2 weeks prior to data collection, researchers visited the school to deliver assemblies on REACH and to distribute information packs for parents and young people. Information was also made available via the study website and, where possible, school websites and mailing lists. Parents were asked to return a form or contact the school or research team if they did not want their child to take part. On the day of assessment, students received further verbal and written information from researchers and, if happy to take part, provided written assent before completing a computerized battery of validated questionnaires, in class, on study tablet computers. All baseline questionnaires were administered between February 2016 and January 2018. Those who completed the baseline questionnaires (i.e. 'Part 1') were then invited to take part in a nested sub-study (i.e. 'Part 2'), comprising a face-to-face interview, hair sample (for cortisol), and cognitive tasks, which we aimed to complete with a subsample of the cohort. Written information about the sub-study was distributed to these young people and to their parents/ carers; parents/carers were asked to provide written consent for their child to take part. On the day of Part 2 assessments, young people were given further verbal and written information by trained researchers and provided written assent before taking part. All study procedures were approved by the Psychiatry, Nursing and Midwifery Research Ethics Subcommittee (PNM-RESC), King's College London (ref: 15/162320).
Of the 4945 eligible students who were invited to participate in Part 1 at baseline, 4353 (88.0%) completed the baseline questionnaire ( Figure 2): Cohort 1: n ¼ 1593 (88.0%); Cohort 2: n ¼ 1421 (90.0%); Cohort 3: n ¼ 1339 (86.1%). Of those who did not take part, 353 (7.2%) were persistently absent, 167 (3.4%) parents refused, 57 (1.2%) young people refused, and 15 (0.3%) provided insufficient data due to technical issues with the study tablet. Of those who participated in Part 1, 85% were from minority ethnic groups and 24% were eligible for free school meals ( Table 1). The REACH cohorts are highly representative of the target population (Table 1). 23 For Part 2, our a priori target was to interview, at two time points, 552 young people and for this subsample to be broadly representative of the target population on core demographics (i.e. gender, ethnic group, age/cohort, and free school meals status) and with 25% experiencing mental health problems [i.e. a score of 18þ on the Strengths and Difficulties Questionnaire (SDQ) based on responses to the Part 1 questionnaire]. Of those who completed baseline Part 1 questionnaires (n ¼ 4353), consent for Part 2 was obtained for 1060 (21.4%) young people. To achieve our target, we stratified these 1060 participants by core demographics and mental health status and selected young people at random from within these strata to complete Part 2. Baseline interviews were completed with 803 young people. This exceeded our target sample size, allowing for attrition at follow-up interviews.
How often have they been followed up?

Part 1: in-class questionnaires
In the first phase of REACH (i.e. adolescence), the cohorts are assessed annually at three time points. Baseline (T1) and 1-year (T2) and 2-year (T3) follow-up assessments have been completed. T3 data cleaning is ongoing. As of May 2020, all 12 schools remain in the study. Of the 4353 young people who completed the T1 questionnaire, 3735 (85.8%) completed the T2 questionnaire. T1-to-T2 attrition was 14.2%. Reasons for non-participation at T2 (among those who took part at T1) were: persistent absence despite repeated visits by researchers [n ¼ 330 (7.6%)]; present but unable to take part because of competing commitments at school [n ¼ 43 (1.0%)]; moved to a non-participating school [n ¼ 193 (4.5%)]; parents refused [n ¼ 36 (0.8%)]; young person refused (n ¼ 16 (0.4%)]. (See Supplementary File 1, available as Supplementary data at IJE online, for a breakdown of these numbers by cohort.) Compared with those who took part at T2, nonparticipants at T2 were more likely to be boys, from poorer households (i.e. eligible for free school meals) and to have had mental health difficulties at T1 (Table 2). However, the magnitude of these differences is small, and the cohorts remain highly representative of the target population (Tables 1 and 2). Due to the study design and the nature of doing research in schools, some of those who were not reassessed at T2 (i.e. the 14% T1-to-T2 attrition) return to the study and are reassessed at T3 (e.g. those who were absent from school at T2 but not at T3). To date, 4005 (92% of those who took part at T1) have completed at least one follow-up (i.e. T2 and/or T3).
T4 is ongoing, online, and is generating information about the heterogeneous impacts of the COVID-19 pandemic, including school closures and other social distancing measures, on young people from disadvantaged and diverse backgrounds. 24 Part 2: face-to-face interview In the first phase of REACH, Part 2 is conducted at two time points, approximately 1 year apart. Of the 803 young people who completed a baseline Part 2 assessment,   What has been measured? Table 3 provides a broad overview of data collected. Briefly, the Part 1 in-class questionnaire comprises validated and widely used measures and collects information on mental health, putative risk and protective factors, demographics, and social circumstances. Supplementary File 3, available as Supplementary data at IJE online, provides a more detailed breakdown of the types of information collected and the measures used at each time point and in each part of the study. The questionnaire takes 1 h to complete, and trained researchers (around 1 per 6 participants) are present in all sessions to answer questions. The Part 2 substudy collects more in-depth information (i.e. frequency, severity, duration, impact, and detailed descriptions of experiences, including support sought/received at the time) on mental health, putative risk and protective factors, and potential mechanisms linking risk and protective factors and mental health [e.g. neurocognition, social cognition, hypothalamic-pituitary-adrenal (HPA) axis activation (hair cortisol)]. All interviews are administered by two trained researchers and take on average 2 h to complete. In addition, linkage of REACH data to data routinely collected by schools for the National Pupil Database (i.e. academic attainment and progress, attendance, exclusions, etc.) is ongoing. In future waves, we will explore the feasibility of linkages to medical records and collection of samples for DNA. Moreover, the REACH cohorts and our ongoing partnerships with schools provide a strong platform for innovative nested studies to further examine the mechanisms linking socioenvironmental risk and protective factors with mental health. For example, 480 young people from the REACH cohorts and schools are taking part in an innovative longitudinal study using virtual reality to investigate the mechanisms linking difficult experiences (e.g. bullying, threat, violence) and state paranoia (i.e., context specific paranoia).

What has it found?
Baseline data on the extent and nature of mental health problems among young people in inner-city London are published 25,26 and core findings from later waves are under review or in press. Two key findings are summarized below.

Extent and nature of mental health problems
Our data suggest that 19% of 11-14-year-olds in innercity London have a mental health problem (weighted prevalence 18.6%, 95% CI: 16.4, 20.8%). This is higher than reported in recent national studies in the UK, including those that have used the same self-report measures (e.g. 12% in Understanding Society 27 ). Moreover, comparing our estimates with those from similar ethnically diverse inner-city London studies conducted in the early 2000s, our data suggest that within inner-city London the prevalence of mental health problems has increased over the past 15 to 20 years, among both boys (from 10-12% to 16%) and girls (from 12-17% to 21%). 28,29 Mental health problems were more common among girls than boys, a difference that was more pronounced in older cohorts (Figures 3 and 4) and among those from economically disadvantaged backgrounds.
Arguably, the most striking observation-with regard to the extent and nature of mental health problems-is that the prevalence of conduct problems in inner-city London schools is around three times higher than reported in a recent national sample [16% (95% CI: 15.2, 17.5) vs 5% (95% CI: 4.6, 5.9]) using the same or similar self-report measures. 26 Variations-and similarities-in prevalence of mental health problems across diverse groups Interestingly, our data suggest many similarities-with some variations-in prevalence of emotional problems (i.e. anxiety, depression) and self-harm by ethnic group. For example, prevalence of mental health problems is similar for White British, Black African, and Black Caribbean groups in REACH. 25 These similarities are striking, as Black African and Black Caribbean groups in the UK experience, on average, greater social, economic, and environmental adversity than their White British peers. Understanding these similarities, despite variations in the distribution of risk factors, is central to our planned analyses. However, our data also hint at differences in the manifestation of distress across diverse groups. For example, our data suggest that prevalence of conduct/behavioural problems is higher among those from Black African and Black Caribbean backgrounds. Importantly though, our data suggest that modifiable social risk factors, including racial discrimination, contribute to variations in prevalence of conduct disorder by ethnic group. 26

Planned analyses
Supplementary File 2, available as Supplementary data at IJE online, outlines the core hypotheses and planned analyses for the first phase (Years 1-5) of REACH. Broadly, the planned analyses include: (i) prevalence and trajectories of mental health problems, overall and by social and ethnic What are the main strengths and weaknesses?
REACH is the largest contemporary UK-based study of mental health among young people from diverse backgrounds. The main strengths of REACH are: high baseline response rates; representative sample; 100% school retention; low participant attrition; the diversity of the cohorts and strong representation of groups that are invariably under-represented in large and national surveys; use of novel data collection methods (i.e. hair cortisol, virtual reality, video diaries) to examine mechanisms and pathways; corroboration of questionnaire data with more in-depth data collected via interviews; biological samples and ongoing data linkages; high frequency of data collection, through critical developmental periods; breadth and depth of information collected, including experiences that are not well documented or understood in the youth mental health literature (e.g. racism); and the accelerated cohort design, which enables rapid collection of data across a wider age range than would be possible in the same time frame with traditional prospective designs and, critically, allows age, period, and cohort effects to be disentangled. In addition, the next phase of REACH will provide robust new information about the impacts of the Covid-19 pandemic on young people from disadvantaged backgrounds and about risk and protective factors for mental health trajectories through the transition to adulthood. Finally, REACH's   extensive and ongoing public engagement programme has engaged over 15 000 local young people, parents, and teachers, and all our engagement materials are available for others conducting school-based research. Consistent with most large prospective studies, the main potential limitations are attrition bias, missing data, and misclassification due to the use of self-report measures. As presented in Table 2, those who did not participate at T2 differ slightly, on average, compared with those who did take part at T2, in terms of some basic characteristics. They also likely differ, on average, in their risk of mental health problems and in their experiences and social circumstances. This is an important limitation. However, due to the relatively high frequency of data collection in REACH and the school setting for fieldwork, we have been able to collect T3 data on many of those who did not complete the T2 questionnaire. Indeed, around 4000 provided data at two of the first three time points. These data are important for monitoring potential biases arising from missing data and will inform the development of multiple imputation models and inverse probability weights to restore representativeness. 30 Another key limitation is the potential for misclassification (e.g. mental health status) with the use of self-report measures in Part 1. However, the questionnaire comprises widely used and validated measures, and we will corroborate self-report questionnaire data with data collected via interviews with participants and through ongoing data linkages. Finally, the Part 1 questionnaire is detailed and takes 50-60 min to complete. Some students did not finish, so missing data due to item non-response is another potential limitation. Some schools allowed extra time for students who required it, but this was not possible at all schools. Nonetheless, the questionnaire content was deliberately structured to reflect our research priorities, such that mental health measures and information on core risk and protective factors were collected at the start of the questionnaire and lower priority questions were at the end of the questionnaire. Coverage of priority measures at baseline is excellent: for instance, the proportion of students with missing data for the baseline strengths and difficulties questionnaire (SDQ) is <0.1%.