COVID‐19 partial school closures and mental health problems: A cross‐sectional survey of 11,000 adolescents to determine those most at risk

Abstract Background Understanding adolescents' mental health during lockdown and identifying those most at risk is an urgent public health challenge. This study surveyed school pupils across Southern England during the first COVID‐19 school lockdown to investigate situational factors associated with mental health difficulties and how they relate to pupils' access to in‐school educational provision. Methods A total of 11,765 pupils in years 8–13 completed a survey in June–July 2020, including questions on mental health, risk indicators and access to school provision. Pupils at home were compared to those accessing in‐school provision on risk and contextual factors and mental health outcomes. Multilevel logistic regression analyses compared the effect of eight risk and contextual factors, including access to in‐school provision, on depression, anxiety and self‐reported deterioration in mental wellbeing. Results Females, pupils who had experienced food poverty and those who had previously accessed mental health support were at greatest risk of depression, anxiety and a deterioration in wellbeing. Pupils whose parents were going out to work and those preparing for national examinations in the subsequent school year were also at increased risk. Pupils accessing in‐school provision had poorer mental health, but this was accounted for by the background risk and contextual factors assessed, in line with the allocation of in‐school places to more vulnerable pupils. Conclusions Although the strongest associations with poor mental health during school closures were established risk factors, further contextual factors of particular relevance during lockdown had negative impacts on wellbeing. Identifying those pupils at greatest risk for poor outcomes is critical for ensuring that appropriate educational and social support can be given to pupils either at home or in‐school during subsequent lockdowns.


INTRODUCTION
The impact of the COVID-19 pandemic and resulting lockdowns on the mental health of young people is a significant societal concern (Courtney et al., 2020;Holmes et al., 2020;Lee, 2020;Townsend, 2020). Young people might be affected by the immediate consequences of full or partial school closures and changes to daily routine, potentially leading to reduced social contact, loneliness and negative impacts on their wellbeing (Brooks et al., 2020;Courtney et al., 2020;Orben et al., 2020). These detrimental effects are especially likely for school pupils who are already at risk of mental health difficulties.
In March 2020, the first UK COVID-19 national lockdown commenced and schools were closed except to children whose parents were essential workers (also known as 'critical' or key workers), or those who were considered 'vulnerable', such as those with mental health needs or living in care (Cabinet Office, 2020; Department for Education, 2020). All other pupils were provided with varying degrees of educational support while at home. From 1 June, pupils in some year groups were also invited back to school, such as those approaching key national examinations. This implies that those offered places in school could be more at risk of mental health difficulties than those who remained at home for one or more reasons; either because they met the criteria for 'vulnerability', because their parents were performing essential roles outside the family home or because they needed to prepare for national examinations in the forthcoming academic year.
There was a concern that partial school closures may reinforce or exacerbate pre-existing inequalities by disproportionally affecting young people who are already at increased risk of poor mental health (Armitage & Nellums, 2020;Van Lancker & Parolin, 2020;Viner et al., 2020). School closures might have some unique benefits to those pupils who frequently experience behavioural difficulties or peer victimisation at school, but it might be detrimental for many others with mental health problems due to loss of support (Courtney et al., 2020;Golberstein et al., 2020;Lee, 2020;YoungMinds, 2020). Cumulative risk factors within homes, such as domestic abuse, limited physical space, economic challenges and single parenthood, may contribute to increased adversity, potentially leading to both immediate and long-term consequences for mental health (Clemens et al., 2020;Cluver et al., 2020;Courtney et al., 2020;Crawley et al., 2020;Gilbert et al., 2009;Usher et al., 2020).
To inform practice and policy for future periods of school lockdown and long-term consequences, this study investigates which young people were most at risk of negative impacts during school closures on their mental health and wellbeing (Golberstein et al., 2020;Holmes et al., 2020;Lee, 2020; and how increased risk relates to whether or not pupils were getting access to in-school educational provision. We used a large crosssectional survey of school pupils during June-July 2020 and compared established factors and lockdown-specific demographic and situational factors on three different outcomes: clinical depression, clinical anxiety and pupils' self-reported deterioration of their mental wellbeing.

Design
OxWell is an annual cross-sectional survey of schools and further education colleges (FECs) in Southern England (Mansfield & Fazel, 2020). This study analyses responses from pupils in years 8-13 (age 12 years and over), for whom the survey included the 25-item Revised Children's Anxiety and Depression Scales (RCADS) (Ebesutani et al., 2012) and multiple questions to assess the risk of mental health disorders.

Population
School pupils in years 8-13 (age 12-25 years) at state-maintained and independent secondary schools and FECs (excluding special schools) in Oxfordshire, Berkshire, Buckinghamshire, Gloucestershire and Wiltshire, plus six schools in North Somerset and Bristol.

Recruitment
Head teachers were invited to sign up their school via an email from their local authority in May-July 2020. All participating schools sent study information and opt-out instructions to parents/guardians before providing information and login instructions directly to pupils or in some cases via parents.

Ethical considerations
All participants included in these analyses gave active online assent to participate. Ethical approval (Ref. R62366/RE0010) was obtained

Measures
The OxWell survey includes over 200 core questions that are asked in annual surveys and additional questions that vary according to emerging hypotheses, described further in the study protocol (Mansfield et al., Submitted). All measures and questions selected for this study are detailed in online Tables S1 and S2. We selected mental health outcomes of depression and anxiety using RCADS (Chorpita et al., 2000;Ebesutani et al., 2012) and of the impact of school closures on mental wellbeing using a single item measure of pupils' perceived change to their wellbeing during lockdown. Predictors were selected that were background factors that could not have been influenced by the outcomes and were a proxy for established risk factors (relating to deprivation and vulnerability), or other characteristics relevant to the allocation of school places (essential worker parents and having upcoming exams). Accordingly, the indicators of socio-economic deprivation selected were access to free school meals and experience of food poverty. The indicators of potentially increased vulnerability were female gender; previous access to mental health support (including within school support); and living circumstances (with both parents in one house compared to other configurations). Measures relevant to the situational risk factors especially relevant to lockdown were how often parents were going out to work (a proxy for essential workers) or being in school years 10 or 12 (approaching UK national examinations in 2021 and thus invited back to school from June 2020). 'School' and 'school year' were selected as control variables in order to adjust for the variance due to these factors.

Analysis plan
For respondents who answered all 25 RCADS items, t-scores for depression, anxiety and combined depression and anxiety were calculated (Chorpita et al., 2000). All p-values were corrected for multiple testing using the false discovery rate, a recommended correction to minimise false positive rates for exploratory analyses in health studies (Glickman et al., 2014), and interpreted at the 95% confidence level.

Outcomes
To investigate which groups were most at risk of clinical depression and anxiety, we created binary RCADS outcomes using the diagnostic thresholds (t-scores ≥ 70) (Chorpita, 2020). To investigate which groups were most likely to perceive their wellbeing to have deteriorated during lockdown, we created a binary outcome defining participants who reported their wellbeing to be 'slightly worse' or 'much worse' during lockdown.

Predictors
All predictors assessed were modelled as binary indicators. Predictor variables included demographic measures (female gender), socioeconomic indicators (free school meals, experience of food poverty), contextual indicators (previous access to mental health support, not living with both parents) and situational factors relevant to lockdown (upcoming examinations, parents in essential roles). The binary variable for year groups with upcoming examinations and invited back to school from June (years 10 and 12) used the remainder of pupils (years 8, 9, 11 and 13) as the reference group. Children of essential workers were identified by how many days their parents left the house to go to work ('most days' or 'every day'), referenced against those whose parents only went out to work 'sometimes', 'once or twice' or 'never'. A further binary variable compared pupils who were 'in-school' during lockdown (those who had left the house to go to school at least once or twice) versus those who were 'at home' (pupils who had 'never' attended school during lockdown at the time of participation). This allowed a comparison of odds ratios (ORs) for those eligible (and taking up in-school places) to the other individual situational risk and contextual indicators.

Models
Multilevel logistic regression analysis (R; lme4 package) (Bates et al., 2014; R Core Team, 2020) was used to calculate ORs for each factor for the three outcomes. In order to adjust for the variance across schools and year groups, accounting for the nested structure of the data, school and year group were included as random intercepts. The set of predictors was identical for each outcome (model), such that the regression coefficients reflect the independent contribution of each variable on the outcome, adjusted for all other variables. As a robustness test, we ran a very preliminary sensitivity analysis aiming to account for non-response bias, which included weights based on ranking and auxiliary information for a subset of demographics that could be matched with the Office for National Statistics census data. We have reported results from the sensitivity analysis when this affected the interpretation of significance levels.

Participants
Of the 65,082 potentially eligible pupils in years 8-13 at the 84 secondary schools and 7 FECs, 14,352 accessed the survey. Pupils who did not access the survey were either not contacted by their school (some schools chose not to invite all relevant year groups), opted out by their parents (schools kept these records), did not read the survey information sent by their school (either due to lack of engagement or limited access to digital media) or chose not to participate. Participants were excluded from the analysis if they spent less than 10 min completing the survey or gave unrealistic or inconsistent responses, leaving 11,765 eligible participants (82% of MANSFIELD ET AL. -3 of 9 those who accessed the survey). Of these, 10,095 answered all 25 RCADS questions, 10,633 reported perceived change to mental wellbeing and 10,784 provided information on how often they physically attended school, which defined the study population (see Table 1). A further 1252 participants had missing data for one or more of the remaining predictors and were not included in the logistic regression models. Sixty-three percentage of the sample was female and response rates declined for older year groups. Participants were aged 12-21, and 315 (3% of the study population) were over 18 and technically adults. The prevalence in this sample of above the clinical threshold depression was 14% and for anxiety it was 10%, with 38% reporting their wellbeing to be worse during lockdown.

Educational provision
Those at home were compared with those in-school (unadjusted frequencies) to assess the extent to which the group in-school scored higher on the individual risk and contextual indicators (Table 1).
2,908 (27%) of the sample reported that they had left their homes to go to school, of which most were in years 10 (1136 = 39%) and 12 (568 = 20%). Of those receiving educational provision at home, 17% were in year 10 and 12% were in year 12. Most of the risk and contextual indicators were higher for those who had accessed inschool provision. Compared with pupils at home full-time, a slightly higher proportion of those in-school reported ever experiencing food poverty (11% vs. 9%, p = .0001), not living with both parents (23% vs. 19%, p < .0001), receiving previous mental health support (29% vs. 23%, p < .0001) and parents being essential workers (43% vs. 36%, p < .0001). The prevalence of above the clinical threshold depression in this sample was greater for the in-school group (17% vs. 13%, p < .0001), as was anxiety (13% vs. 9%, p < .0001). Pupils with inschool educational provision were also more likely to report that their wellbeing was worse during lockdown (42% vs. 36%, p < .0001).

Risk factors for mental health difficulties
Depression and anxiety, and self-reported deterioration to wellbeing, were each modelled against the demographic and situational indicators. Adjusted ORs and 95% confidence intervals were calculated (see Table 2), where all indicators were accounted for together in each model.

Depression and anxiety
Risk of depression was found to be higher for females (adjusted OR = 3.62, p < .0001), pupils who had previously accessed mental health support (adjusted OR = 3.91, p < .0001) and those who had ever experienced food poverty (adjusted OR = 3.36, p < .0001).
Increased risk of depression was also found for pupils whose parents were likely essential workers (adjusted OR = 1.34, p < .0001), pupils who were approaching national examinations (adjusted OR = 1.46, p < .0001) and a small increased risk for pupils not living with both  Note: The term "n" refers to the total number included in each reported comparison between pupils at home versus those accessing in-school provision; percentages reported are of the pupils who were in the relevant group (columns: home/school) and provided data for the relevant question (rows: factors).

DISCUSSION
This study provides insight into the mental health and wellbeing of different groups during the partial school closures of the first UK COVID-19 school lockdown in 2020 and how this relates to which pupils took up in-school places. The results show that those who were accessing on-site school were more likely to have depression or anxiety and were more likely to report a deterioration in their wellbeing. However, the poorer outcomes for the group in-school were accounted for by pre-existing vulnerability (e.g. experiencing T A B L E 2 Logistic regression of mental health and self-reported worse wellbeing with adjusted odds ratios for all risk and contextual factors  Parolin, 2020). We found that pupils who had previously accessed mental health support had almost four times the odds of reaching clinical thresholds for depression and anxiety and were also more likely to report a further deterioration of their wellbeing. This is consistent with the finding by YoungMinds that 80% of the 2036 adolescents surveyed with pre-existing mental health needs felt that their mental health had become worse during the pandemic (YoungMinds, 2020). This deterioration may be related to the reduced school provision during lockdown, missing many of the usual school rituals and structures, including interactions with peers and school staff, structured and unstructured daytime activities and systems of support (Courtney et al., 2020;Golberstein et al., 2020;Lee, 2020;YoungMinds, 2020). Notably, female pupils were especially at increased risk of depression, consistent with other studies (Altemus et al., 2014). Self-reported food poverty was higher than self-reported eligibility for free school meals, and pupils reporting food poverty had more than three times the odds of having depression or anxiety. In contrast, self-reported access to free school meals was not reliably related to any of the mental health outcomes and did not differ according to educational provision, possibly due to pupils not being fully aware of which category they might fall into.
In addition to the established risk factors of vulnerability, deprivation and gender, a slightly increased risk was identified on all three mental health outcomes for factors especially relevant to lockdown. The first of these is the increased risk for pupils in school years 10 and 12, who were invited back to school from June to help them prepare for key examinations in 2021. These pupils were F I G U R E 1 Adjusted odds ratios and 95% confidence intervals for depression, anxiety and self-reported worse wellbeing during school closures hypothesised to represent a group at increased risk during lockdown because they might experience longer term impacts of school closures on their educational and vocational outcomes, a hypothesis which is in line with responses to some smaller surveys suggesting that young people's mental health was impacted during lockdown by concerns about academic performance (NHS Digital, 2020; Young-Minds, 2020). These concerns might relate to reduced instruction time for key components of the curriculum, which could be more detrimental for pupils approaching national examinations. Pupils whose parents were leaving the house to go to work most days during lockdown were also more likely to report their wellbeing to be worse, potentially reflecting concerns about their parents or increased stress and reduced support at home (Clemens et al., 2020).
The fact that pupils of essential workers were at increased risk even when controlling for other key characteristics, including in-school provision, supports the allocation of school places to these pupils during lockdowns and has implications for targeting support.

CONFLICT OF INTEREST STATEMENT
The authors have declared that they have no competing or potential conflicts of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author as an anonymised data extract in accordance with research governance procedures. The data are not publicly available due to privacy or ethical restrictions.