You are here

  • Home
  • Archive
  • Volume 11, Issue 12
  • Loneliness, coping, suicidal thoughts and self-harm during the COVID-19 pandemic: a repeat cross-sectional UK population survey

Article Text

Article menu
Mental health
Original research
Loneliness, coping, suicidal thoughts and self-harm during the COVID-19 pandemic: a repeat cross-sectional UK population survey
  1. Ann John1,
  2. Sze Chim Lee1,
  3. Susan Solomon2,
  4. David Crepaz-Keay2,
  5. Shari McDaid2,
  6. Alec Morton3,
  7. Gavin Davidson4,
  8. Tine Van Bortel5,6,
  9. Antonis A. Kousoulis2
  1. 1Health Data Research UK, Swansea University Medical School, Swansea University, Singleton Park, Swansea, UK
  2. 2Mental Health Foundation, Colechurch House, 1 London Bridge Walk, London, UK
  3. 3Department of Management Science, Strathclyde Business School, University of Strathclyde, 199 Cathedral Street, Glasgow, UK
  4. 4School of Social Sciences, Education and Social Work, Queen's University Belfast, Belfast, UK
  5. 5Cambridge Public Health, Department of Psychiatry, Cambridge School of Clinical Medicine, University of Cambridge, Cambridge, UK
  6. 6Leicester School of Allied Health Sciences, Faculty of Health and Life Sciences, De Montfort University, Leicester, UK
  1. Correspondence to Professor Ann John; a.john{at}swansea.ac.uk

Abstract

Objectives There has been speculation on the impact of the COVID-19 pandemic and the associated lockdown on suicidal thoughts and self-harm and the factors associated with any change. We aimed to assess the effects and change in effects of risk factors including loneliness and coping, as well as pre-existing mental health conditions on suicidal thoughts and self-harm during the COVID-19 pandemic.

Design This study was a repeated cross-sectional online population-based survey.

Participants and measures Non-probability quota sampling was adopted on the UK adult population and four waves of data were analysed during the pandemic (17 March 2020 to 29 May 2020). Outcomes were suicidal thoughts and self-harm associated with the pandemic while loneliness, coping, pre-existing mental health conditions, employment status and demographics were covariates. We ran binomial regressions to evaluate the adjusted risks of the studied covariates as well as the changes in effects over time.

Results The proportion of individuals who felt lonely increased sharply from 9.8% to 23.9% after the UK lockdown began. Young people (aged 18–24 years), females, students, those who were unemployed and individuals with pre-existing mental health conditions were more likely to report feeling lonely and not coping well. 7.7%–10.0% and 1.9%–2.2% of respondents reported having suicidal thoughts and self-harm associated with the pandemic respectively throughout the period studied. Results from cross-tabulation and adjusted regression analyses showed young adults, coping poorly and with pre-existing mental health conditions were significantly associated with suicidal thoughts and self-harm. Loneliness was significantly associated with suicidal thoughts but not self-harm.

Conclusions The association between suicidality, loneliness and coping was evident in young people during the early stages of the pandemic. Developing effective interventions designed and coproduced to address loneliness and promote coping strategies during prolonged social isolation may promote mental health and help mitigate suicidal thoughts and self-harm associated with the pandemic.

  • COVID-19
  • mental health
  • public health
  • suicide & self-harm

Data availability statement

Data are available upon reasonable request. The data that support the findings of this study are available from the Mental Health Foundation, contactable through the corresponding author upon reasonable request once the study is completed.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/bmjopen-2020-048123

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Strengths and limitations of this study

    • This is a novel and timely study exploring the determinants (and their changes) in suicidal thoughts and self-harm in the UK general population in the ongoing COVID-19 pandemic.

    • Potential sampling bias through use of an online survey with non-probability quota sampling.

    • Repeated cross-sectional design.

    • Variables regarding self-harm and suicidal thoughts were missing at baseline.

    • Validated measures assessing mental health conditions were lacking in this study.

    Introduction

    Worldwide research efforts have understandably been focused on preventive and therapeutic strategies1 against the COVID-19 pandemic.2 The immediate and potential long-term impacts on mental health, suicide and self-harm have not yet been widely addressed.3 4 Without effective treatment or vaccines, extensive lockdown and social/physical distancing measures were implemented worldwide to contain the pandemic. In addition to the threat of infection, public mental health and well-being were expected to deteriorate in tandem with these dramatic changes at personal (eg, restricted freedom), social (eg, due to isolation and distancing) and economic levels (eg, unemployment and financial hardship), particularly for vulnerable individuals.3 5 Studies from China in the very early stages of COVID-19 pandemic have found high levels of mental health problems and distress in the general population.6 7

    Previous epidemics where similar public health measures were taken, for example, SARS, resulted in an elevated suicide rate in elderly population (over 65 years) in Hong Kong in 2003.8 9 This was associated with anxiety and worry concerning their susceptibility to the pandemic, pre-existing health conditions, lack of social support and higher reliance on the health services that were under pressure during the outbreak. However, important psychological factors such as loneliness, perceived stress level and coping were not examined.

    Loneliness, the self-perceived deficiency of an individual’s social relation network in quantitative or qualitative terms,10 has been identified as an important factor associated with suicidal thoughts and behaviours.11–17 The effect of loneliness on suicidal thoughts and behaviours is more prominent in younger (16–20 years) and older (>58 years) populations because these two age groups similarly experience periods where drastic changes in social status occur.13 Previously, social isolation and loneliness were linked to health protection measures similar to those taken in the COVID-19 pandemic resulting in a deterioration in children and young people’s mental health.18 Similarly, individuals’ perceived stress level is often associated with suicidal behaviours alongside mental disorders such as depression.19–21 Perceived stress and poor coping mechanisms are associated with suicidal behaviours in both younger22–25 and older individuals.26

    A UK-wide lockdown was announced on 23 March 2020 which included instructions to the general population to stay at home, socially/physically distance and self-isolate if they had symptoms. This was accompanied by guidance regarding movements outside the home for exercise and grocery shopping. These restrictions were fully in place until 13 May 2020 when they were gradually eased. Given the potential for such measures to be implemented in any further waves of COVID-19 or other pandemics, it is important to understand the effects of these measures on mental health and well-being in order to mitigate them in the future and address them currently. A growing body of emerging research focuses on assessing mental health and well-being in response to the pandemic and measures taken to curb its spread at a population level.27 Outcomes related to self-harm and suicidality have not been commonly reported in repeated surveys although they are included in some.28 29 This study aimed to explore the risk factors for suicidal thoughts and self-harm in response to the COVID-19 pandemic in the UK population using a repeated cross-sectional online population survey representative of the adult UK population. We assessed the effects of known modifiable psychological risk factors including loneliness and coping, as well as pre-existing mental health conditions and other sociodemographic covariates.

    Methods

    Study design and participants

    This study, a part of a larger mixed-methods study to investigate mental impacts of COVID-19 pandemic in the UK,30 was a repeated cross-sectional online population survey using a quota survey design and a sampling frame allowing recruitment of a national sample. We examined the public’s mental health by assessing emotional responses, sources of social distress, coping, suicidal thoughts and self-harm in relation to the COVID-19 pandemic from individuals aged ≥18 years living in the UK. Data for the ‘Coronavirus: Mental Health in the Pandemic’ study were first collected shortly before the UK-wide lockdown was announced and repeated approximately every 3–4 weeks. In this study, we report on the first four waves with wave 1, wave 2, wave 3 and wave 4 conducted on 17–18 March 2020, 2–3 April 2020, 24–26 April 2020 and 28–29 May 2020, respectively. We aimed at a sample size of ~2000 in waves 1 and 2 but this was then doubled to ~4000 in later waves to increase statistical power for subgroups. Data collection is still underway although at increased intervals and all surveys were administered online by YouGov, a social market research company that recruited participants to form a panel containing over 1 million individuals from the UK.27 Quota sampling was used and data were weighted to be representative to the UK adult population. Detailed procedures for sampling and weighting were described in the Suppl Methods (online supplemental file 1). Participants were different in each wave but taken from the same panel and representative of the UK adult population. Weighted bases, counts and percentages were reported unless otherwise specified to avoid identification of individuals in the case of small counts.

    Participants signed up to YouGov to participate in surveys and they read and agreed to the terms and conditions of use and privacy policy before responding.31

    Patient and public involvement

    The survey forms part of a larger programme of work by the Mental Health Foundation (MHF) which includes focus groups of individuals (general population, risk groups, those with pre-existing mental illness) from the extensive networks of the MHF where themes identified from the survey are further discussed and questions for the next survey suggested.

    Measures

    Outcomes

    The outcomes of this study were self-reported experience of suicidal thoughts and self-harm behaviours, which were available in wave 2, wave 3 and wave 4 (see online supplemental table 1, see also a copy of the survey in online supplemental file 2). Participants were prompted first about the sensitivity of the topic and provided an option to skip the related questions. For participants who proceeded, they were prompted ‘Have you done or experienced any of the following, as a result of the Coronavirus (COVID-19) pandemic in the past 2 weeks? (Please select one option on each row)’ and then presented with two descriptions with one for suicidal thoughts (‘experienced suicidal thoughts/feelings’) and the other for self-harm (‘deliberately hurt myself’). Respondents were given three options: ‘yes’, ‘no’ and ‘prefer not to say’ for these two questions.

    Covariates

    The main covariates were loneliness and coping with stress. For loneliness, participants were first prompted ‘Which, if any, of the following emotions have you felt as a result of the Coronavirus (COVID-19) pandemic in the past 2 weeks? (Please select all that apply)’ and loneliness was one of the emotions. As a result, we analysed this binary variable reflecting whether individuals felt lonely due to the pandemic. For coping, participants were asked ‘For the following question, if you have not experienced any stress related to the Coronavirus pandemic, please select the “Not applicable” option. Overall, how well do you think you are coping with stress related to the Coronavirus (COVID-19) pandemic?’. Respondents were given options of ‘very well’, ‘fairly well’, ‘not very well’, ‘not at all well’, ‘don’t know’, ‘prefer not to say’ and ‘not applicable I have not experienced any stress related to the Coronavirus’ to answer. We combined ‘very well’ and ‘fairly well’ into ‘well’ category, as well as ‘not very well’ and ‘not at all well’ into a ‘not well’ category.

    Other covariates included sex, age, marital, work and socioeconomic status and household information. Detailed descriptions of these variables were summarised in Suppl Methods (online supplemental file 1). We also included variables concerning pre-existing mental health conditions, living arrangements and urban/rural classification starting from wave 3. For waves 3 and 4, we asked the participants who were employed whether they were temporarily furloughed by their employer. We combined working and furlough status as a single categorical variable for analyses whenever appropriate (Suppl Methods in online supplemental file 1). For pre-existing mental health conditions, participants were asked ‘Do you have a current pre-existing mental health condition or psychiatric diagnosis?’ and participants were given options ‘Yes’, ‘No’, ‘Don’t know’, ‘Prefer not to say’ and ‘Refused’ to respond (Suppl Methods in online supplemental file 1).

    Statistical analyses

    Sample weighting was incorporated in all statistical analyses to obtain UK representative estimates. The level of statistical significance was set at p=0.05 and all analyses were performed with Stata V.16.1.32 We reported sample characteristics and percentage of suicidal thoughts, self-harm with 95% CIs stratified by loneliness, coping and other sample characteristics for each wave of data collection where possible. For two-way and three-way cross-tabulations (figures 1–3), the significance of the association variables including wave, loneliness, coping, sex, age and working status was assessed by Wald χ2 statistics after conducting full factorial binomial regression analyses. In these analyses, coping variables were grouped into a binary variable (‘not well’ vs ‘well’/‘no stress’).

    Figure 1
    Figure 1

    Distribution of loneliness (A–C), not coping well with stress (D–F), suicidal thoughts (G–I) and self-harm (J–L) in the past 2 weeks during the COVID-19 pandemic stratified by sex (B, E, H, K), age (B, E, H, K) and working status (C, F, I, L).

    Figure 2
    Figure 2

    Association of suicidal thoughts in the past 2 weeks during the COVID-19 pandemic and loneliness stratified by sex (A–C), age (D–F), working status (G–I) and coping (J–L). NA, not applicable.

    Figure 3
    Figure 3

    Association of self-harm in the past 2 weeks during the COVID-19 pandemic and loneliness stratified by sex (A–C), age (D–F), working status (G–I) and coping (J–L). NA, not applicable.

    Binomial regression analyses were conducted using the ‘BINREG’ procedure in Stata33 with suicidal thoughts and self-harm separately as the dependent variable. We combined the data collected in wave 3 and wave 4 only for the regression because these two waves contained the largest number of variables of interest as well as the largest sample size. To pool data from two waves, sampling weights for each wave were reweighted using the sample population of each wave according to previously described methodology for analysing multiple independent surveys.34

    Both unadjusted and adjusted analyses for suicidal thoughts and self-harm were performed with the independent variables including time (wave 3 vs wave 4), loneliness, coping, sex, age, marital status, social grade, working status, pre-existing mental health conditions, household information, variables for living arrangements and urban/rural characteristics. All independent variables were included in the models as binary or categorical variables. We also included wave-by-loneliness, wave-by-coping, wave-by-sex, wave-by-age and wave-by-working status interaction terms in the adjusted model to explore change in effects of risk factors over time. Due to model convergence issues, the ‘employed’ category of the working status variable was not further stratified by furlough status in the regression analyses. We reported both risk ratios (RR) and risk difference (RD) with the corresponding 95% CIs and p values for independent variables. For the adjusted analysis, we also reported the Wald χ2 statistics to examine the significance of overall effects for all independent variables and interactions. To assess the trend of the effect sizes over age, we performed the χ2 test for linear trend to the model coefficients for the age variable in the adjusted analyses. To examine the robustness of the main analyses, we conducted a sensitivity analysis by repeating the regression on the combined data from wave 2 to wave 4, where suicidal thoughts and self-harm were available in these waves (online supplemental table 1). Details of the sensitivity analysis are described in Suppl Methods (online supplemental file 1).

    Results

    Sample and participant characteristics

    Between March 2020 and May 2020, the number of respondents for wave 1, wave 2, wave 3 and wave 4 of data collection was 2126, 2221, 4246 and 4382, respectively (table 1 for sample characteristics, see also online supplemental table 2). Missing data and inconclusive responses, including answers of ‘prefer not to say’, ‘don’t know’ and ‘refused to answer’, were considered minimal (online supplemental table 1) and were thus excluded in cross-tabulations and regression analyses, given the assumptions of negligible effect of missing data.35 The highest proportion of missing data was found for suicidal thoughts, self-harm and pre-existing mental health conditions. Cross-tabulation of these variables showed that missing data/inconclusive responses on suicidal thoughts/self-harm were more likely to be associated with missing data/inconclusive responses on having pre-existing mental health conditions (online supplemental table 1A). However, missing data/inconclusive responses on having mental health conditions were more likely to have suicidal thoughts or self-harm.

    View this table:
    Table 1

    Sample characteristics

    Loneliness

    The proportion of participants feeling lonely at wave 1, prior to the UK lockdown announcement, was 9.8% (95% CI 8.6% to 11.2%). This increased to 23.9% (95% CI 22.1% to 25.8%), 26.7% (95% CI 25.4% to 28.1%) and 25.7% (95% CI 24.3% to 27.1%) at wave 2, wave 3 and wave 4, respectively. The effect of wave was statistically significant (χ2=186.4, df=3, p<0.001) reflecting the dramatic rise of proportion of individuals feeling lonely between wave 1 and wave 2 (figure 1A–C). Further details of the proportion of respondents who felt lonely in the past 2 weeks for each wave stratified by sample characteristics are available in online supplemental table 3.

    Respondents who were female (χ2=103.0, df=1, p<0.001), younger, that is, 18–24 years (χ2=455.5, df=4, p<0.001 for the effect of age), full-time students or unemployed (χ2=265.2, df=4, p<0.001 for the effect of working status) were more likely to report feeling lonely across all waves (figure 1A–C). Respondents who had pre-existing mental health conditions/psychiatric diagnosis were more likely to report feeling lonely (wave 3: 44.0% vs 21.6%; wave 4: 40.7% vs 20.8%).

    Coping with stress during the COVID-19 pandemic

    For coping with stress during the pandemic in the past 2 weeks, 11.8% (out of 2159; 95% CI 10.5% to 13.3%), 14.9% (out of 4132; 95% CI 13.8% to 16.0%) and 16.2% (out of 4280; 95% CI 15.1% to 17.4%) did not feel stressed due to the pandemic at wave 2, wave 3 and wave 4, respectively (online supplemental table 4–6). The proportion of respondents who responded coping ‘not well’ for wave 2 to wave 4 was 13.5% (95% CI 12.1% to 15.0%),12.9% (95% CI 11.9% to 14.0%) and 14.2% (95% CI 13.2% to 15.4%), respectively. These proportions did not vary significantly across waves (χ2=3.4, df=2, p=0.18). We found individuals who were female (χ2=28.8, df=1, p<0.001), younger (χ2=216.4, df=4, p<0.001 for the effect of age), unemployed and full-time students (χ2=244.0, df=4, p<0.001 for the effect of working status) were more likely to respond coping ‘not well’ (figure 1D–F). From wave 3 and wave 4, we found respectively 27.3% and 32.1% of individuals with pre-existing mental health conditions responded coping ‘not well’ compared with only 7.8% and 8.6% for respondents without mental health conditions (online supplemental table 5–6). We also found participants who were employed and temporarily furloughed had only slightly higher proportion of not coping well with stress (15.2% vs 11.9% and 16.2% vs 13.2% for wave 3 and wave 4, respectively, online supplemental table 7).

    Suicidal thoughts and self-harm during the COVID-19 pandemic

    We analysed individual responses on suicidal thoughts (online supplemental table 8) and self-harm behaviours (online supplemental table 9) in the past 2 weeks due to the COVID-19 pandemic collected at wave 2 to wave 4. The proportion of participants with suicidal thoughts slightly increased from 7.7% (out of 2097; 95% CI 6.6% to 9.0%) at wave 2 to 9.4% (out of 3968; 95% CI 8.5% to 10.3%) at wave 3 and 10.0% (out of 4135; 95% CI 9.1% to 11.1%) at wave 4. The proportion of participants who self-harmed was similar across waves at around 2% (wave 2: 1.9% out of 2111, 95% CI 1.3% to 2.6%; wave 3: 2.2% out of 3988, 95% CI 1.8% to 2.8%; wave 4: 1.9% out of 4156, 95% CI 1.5% to 2.4%; online supplemental table 9). The change in proportion of suicidal thoughts (χ2=5.3, df=2, p=0.069) and change in proportion of self-harm (χ2=1.5, df=2, p=0.480) were not significantly different across waves. Effects of being female (χ2=8.6, df=1, p=0.034 on suicidal thoughts; χ2=4.8, df=1, p=0.029 on self-harm), younger age (χ2=228.4, df=4, p<0.001 on suicidal thoughts; χ2=87.6, df=4, p<0.001 on self-harm) and working status (χ2=199.0, df=4, p<0.001 on suicidal thoughts; χ2=80.8, df=4, p<0.001 on self-harm) were statistically evident on suicidal thoughts (figure 1G–I, online supplemental table 8) and self-harm (figure 1J–L, online supplemental table 9). Participants who were employed and those furloughed had only slightly higher proportion of having suicidal thoughts (9.1% vs 8.7% and 11.5% vs 9.3% for wave 3 and wave 4, respectively, online supplemental table 7) and self-harm (2.1% vs 2.1% and 2.3% vs 1.4% for wave 3 and wave 4, respectively, online supplemental table 7) compared with those not being furloughed. Although we observed the proportion of having suicidal thoughts increased over time only in age groups 18–24 and 25–34 years (figure 1H), the wave-by-age interaction was not statistically significant (χ2=9.8, df=8, p=0.280).

    Suicidal thoughts and self-harm by loneliness and coping

    Respondents who felt lonely were more likely to have suicidal thoughts (17.7% out of 2097 for wave 2, 21.0% out of 3968 for wave 3 and 23.5% out of 4135 for wave 4) and to have harmed themselves (5.4% out of 2111 for wave 2, 4.1% out of 3988 for wave 3 and 4.0% out of 4156 for wave 4) than the general population (online supplemental tables 8 and 9). Compared with respondents who did not feel stressed during the pandemic, we found higher proportions of suicidal thoughts and self-harm for respondents who responded coping ‘not well’ (online supplemental tables 8 and 9). We also performed three-way cross-tabulations on suicidal thoughts (figure 2) and self-harm (figure 3) by loneliness, coping, sex, age and working status to elucidate the effects from these factors across waves. Individuals with the highest risks of suicidal thoughts and self-harm in response to the pandemic were those feeling lonely (χ2=379.9, df=1, p<0.001 on suicidal thoughts; χ2=11.0, df=1, p<0.001 on self-harm) and responded coping ‘not well’ (χ2=511.8, df=1, p<0.001 on suicidal thoughts; χ2=162.0, df=1, p<0.001 on self-harm). Both younger age (χ2=119.0, df=4, p<0.001 on suicidal thoughts; χ2=53.2, df=4, p<0.001 on self-harm) and working status (χ2=111.9, df=4, p<0.001 on suicidal thoughts; χ2=62.5, df=1, p<0.001 on self-harm) were also significant risk factors. We did not observe significant effect of sex, wave nor wave-related interactions on suicidal thoughts/self-harm from all three-way cross-tabulations.

    Factors associated with suicidal thoughts and self-harm during the COVID-19 pandemic

    We built binomial regression models to quantify the effect size (RR and RD) of loneliness, coping and other covariates on suicidal thoughts (table 2, online supplemental table 10) and self-harm (table 2, online supplemental table 11) for data collected at wave 3 and wave 4 only. In the unadjusted analyses for suicidal thoughts, RRs and RDs were all significantly greater than 1 and 0 respectively for loneliness, coping ‘not well’, young age, unemployment and being full-time students compared with the respective references. Similar patterns were observed for self-harm although the RD between unemployed and employed categories was marginally non-significant (RD=0.022, 95% CI −0.001 to 0.045, p=0.055). The effect of having pre-existing mental health conditions was also strong and statistically significant for both suicidal thoughts (RR=4.9, 95% CI 4.2 to 5.7, p<0.001; RD=0.191, 95% CI 0.169 to 0.214, p<0.001, online supplemental table 10) and self-harm (RR=6.9, 95% CI 4.8 to 10.0, p<0.001; RD=0.047, 95% CI 0.036 to 0.059, p<0.001, online supplemental table 11).

    View this table:
    Table 2

    Summary of risk ratios (RR) and risk differences (RD) (with 95% CIs) of time (wave), loneliness, coping, age and working status on suicide thoughts and self-harm in unadjusted and adjusted binomial regression models based on data from wave 3 and wave 4

    The effect sizes of the variables on suicidal thoughts and self-harm were reduced in the adjusted models (table 2, online supplemental tables 10 and 11). For suicidal thoughts, effect sizes of loneliness and coping ‘not well’ with stress remained statistically significant. The effect sizes were the largest in the younger age group (18–24, 25–34 and 35–44 years) and remained significant in the adjusted analysis but decreased monotonically with older age groups (for RR: χ2 for testing linear trend=8.8, df=1, p=0.003; for RD: χ2 for testing linear trend=8.5, df=1, p=0.004). Effect of working status was no longer a significant risk factor in the adjusted model. For self-harm, only coping ‘not well’ and age remained strong and robust risk factors in the adjusted model, with the effect sizes of age decreased for older age groups (table 2 and online supplemental table 11; for RR: χ2 for testing linear trend=15.3, df=1, p<0.001; for RD: χ2 for testing linear trend=9.5, df=1, p=0.002). Loneliness and working status were no longer significant effects. Not surprisingly, having pre-existing mental health conditions remained a statistically robust factor in the adjusted analysis for both suicidal thoughts (RR=2.2, 95% CI 1.9 to 2.7, p<0.001; RD=0.077, 95% CI 0.059 to 0.094, p<0.001, table 2 and online supplemental table 10) and self-harm (RR=2.8, 95% CI 1.9 to 4.2, p<0.001; RD=0.020, 95% CI 0.012 to 0.028, p<0.001, table 2 and online supplemental table 11). Between wave 2 and wave 4, we did not find statistical evidence supporting for change in the risk of having suicidal thoughts/self-harm, nor change in the time-related interactions with loneliness, coping, sex, age or working status.

    Results from the sensitivity analysis which combined data from wave 2 to wave 4 and contained only a subset of independent variables from the main analysis were generally consistent with the main analysis with minor exceptions (online supplemental tables 12 and 13). For the adjusted analysis on suicidal thoughts (online supplemental table 12), the statistically robust effects of loneliness, coping and younger age were consistent with the respective model of the main analysis (online supplemental table 10). However, the effect of working status became significant in the adjusted model of the sensitivity analysis, particularly for the unemployed and not working/others groups. In keeping with the main analysis, we found no significant effect of wave nor effects of wave-related interactions except the wave-by-sex interaction. This reflected the differential trends of the risk of having suicidal thoughts between sexes from wave 2 (male: 8.2%, 95% CI 6.5% to 10.2%; female: 7.3%, 95% CI 5.9% to 9.0%) to wave 3 (male: 8.7%, 95% CI 7.4% to 10.1%; female: 10.0%, 95% CI 8.8% to 11.4%, online supplemental table 8). For self-harm, we found significant effects of coping and younger age but not working status in the adjusted model of the sensitivity analysis (online supplemental table 13), concurring with the respective main analysis (online supplemental table 11). However, the effect of loneliness was statistically significant in the adjusted model of the sensitivity analysis but not in the respective model of the main analysis. Neither the effect of wave nor the effects of wave-related interactions reached the level of statistical significance.

    Discussion

    Main findings

    This study is one of the few examining the mental health and well-being of the UK population with baseline data from a week before lockdown was announced (wave 1) and for the next 10 weeks (wave 2 to wave 4). We documented a sharp increase in loneliness among the UK population between wave 1 (9.8%) and wave 2 (23.9%) of data collection (late March 2020) when lockdown measures were implemented, particularly in young people. The proportion of the population coping ‘not well’ across waves remained stable at around 13.5%. Suicidal thoughts and self-harm similarly remained stable between wave 2 and wave 4 at 7.7%–10.0% and 1.9%–2.2%, respectively. We found that feeling lonely, coping ‘not well’ with stress, as well as younger age groups and those with pre-existing mental health conditions were associated with the highest risk of having suicidal thoughts and self-harm. For individuals aged between 18 and 24 years, risk of suicidal thoughts and self-harm were ~15%–20% and 5%, respectively, and no statistically significant increase was found over time. Around 25% and 5% of individuals with pre-existing mental health conditions had suicidal thoughts and harmed themselves, respectively. Rates of suicidal thoughts and self-harm were slightly higher for individuals from the lower socioeconomic category (8.0%–11.0% and 2.1%–2.3%, respectively) compared with those from the higher category but social grade was not a significant predictor when we controlled for other covariates.

    Loneliness and coping remained significantly associated with suicidal thoughts whereas coping with stress but not loneliness was an important factor for self-harm when we adjusted our analyses for other covariates at a personal, household and area level. Having pre-existing mental health conditions or psychiatric diagnosis was a strong and statistically robust risk factor for both outcomes. For sociodemographic factors, we found that effect sizes remained the largest and statistically significant for the youngest (18–24 years) and monotonically decreased with older age groups for both suicidal thoughts and self-harm. In contrast, the effect of employment status was not statistically evident. Our results did not identify significant temporal changes in the risk of having suicidal thoughts/self-harm and the effects of the interested risk factors on the outcomes remained constant between wave 3 and wave 4. Our sensitivity analysis was in agreement with the main analysis with minor exceptions. We believed that those disagreements stem from the omission of important risk factors (eg, pre-existing mental health conditions) due to data unavailability at wave 2 in the sensitivity analysis, resulting in overestimation of other effects and interactions. Interestingly, loneliness was a statistically robust risk factor for self-harm only in the sensitivity but not in the main analysis where more covariates, including pre-existing mental health conditions, were included in the adjusted model. This difference and our other results are consistent with the findings of a recent review on the impact of loneliness and social isolation measures related to disease containment which showed high correlation of loneliness and mental health symptoms in children and young people.18

    Due to differences in sampling methodologies and measures used, direct comparison of prevalences found in our study with other studies is not robust. Our measures on suicidal thoughts/self-harm were asked in relation to them being a consequence of the pandemic. Thus, the respective prevalences are lower than those from similar UK population-based survey (eg, Iob and colleagues28) since instances not related to the pandemic might not be captured. Nonetheless, our prevalence of having mental health conditions (~20%) is in keeping with other UK studies (~17%–26%28 29 36–38). After the UK lockdown from early April 2020 (wave 2 onwards), we did not observe any further significant increase in the proportion of participants with suicidal thoughts, self-harm, loneliness and coping not well. This is consistent with the constant or even decreasing trends of various mental health outcomes for the general population shortly after lockdown.29 37 38 Notably, our increasing, although non-significant trend of suicidal thoughts from early April to the end of May 2020 qualitatively tallies with the increase of suicidal ideation over the same period in a similar study.29

    Results from our unadjusted analyses are similar to others suggesting suicidal thoughts and self-harm are higher among women, younger ages, people experiencing socioeconomic disadvantage and those with pre-existing mental conditions.28 29 These factors are strongly associated with other adverse mental health outcomes during the pandemic as reported elsewhere.37–41 Comparison of effect sizes from multivariable analyses across studies is more challenging because of the differences in the use and definitions of outcomes and covariates. Nonetheless, pre-existing mental health conditions and lack of coping are strongly associated with suicidal thoughts/self-harm. Young people with pre-existing mental health issues and not coping well were unsurprisingly more likely to experience adverse outcomes, the effect size of loneliness reduced (especially in relation to self-harm) when all measured psychosocial, demographical and environmental factors were adjusted for.

    It has been suggested that young adults are more susceptible to stressors because of their ongoing transitions across multiple aspects of life.13 42–44 These transitions involve personal (becoming financially independent), interpersonal (building of romantic relationships), educational (graduating from schools), professional (entering the labour market) and environmental (beginning living away from family) changes which are unprecedented and uncertain.13 45 Distress could be exacerbated by the additional uncertainty, hopelessness and worries associated with the COVID-19 pandemic and lockdown.45 The pandemic-associated economic decline may disproportionally impact young adults as they are already more likely to be at the margin of the labour market.46 47 Social isolation through stay at home orders likely impeded interactions with friends, colleagues or romantic partners thereby weakening existing social supports and promoting loneliness.13

    Strength and limitations

    This study is a novel and timely investigation that explores the determinants of suicidal thoughts and self-harm in the UK general population in the ongoing COVID-19 pandemic. Timeliness, including the need for prelockdown data, necessitated quota over probability sampling. We made use of an online population survey, the only feasible method of recruitment during lockdown that contained questions on suicidal thoughts, self-harm, loneliness coping with stress alongside a broad range of other factors. Since data were collected from panellists recruited from all UK nations with quota sampling and weighting, our results are representative for the UK adult general population. The survey is repeated approximately every 3–4 weeks, began prelockdown from mid-March 2020 so that change in outcomes and other factors due to the evolving nature of the outbreak (eg, increase in loneliness after lockdown) were captured. We included suicidal thoughts and self-harm behaviours as outcomes and examined their determinants. We adjusted for a number of covariates that cover psychosocial (loneliness and coping), demographic and environmental aspects. We believe that similar population-based repeat surveys are not commonly reported.

    A major limitation is the use of non-probability quota sampling through online recruitment.48 However, while an opt-in online panel is cost-effective for recruiting a sizeable number of diverse individuals within a short time frame,49 we could not determine the response rate for the survey. Doubts have also been cast on the representativeness of non-probability internet-based samples as the outcome of interest may differ from that of the target population even when quotas and weighting were applied.49 Digital poverty (particularly for older population) and the digital divide may hamper representativeness and data validity. It is worth noting that full-time students were over-represented (~7%) in our sample when compared with available statistics (~4%, from ~2 million students studied in higher education institutions out of ~52 million UK adult population in 2019/202050 51). In order to increase statistical power for subgroup analyses, sample sizes in waves 3 and 4 were approximately doubled compared with those in waves 1 and 2. Therefore, our prevalences should not be regarded as precise and conventional estimates. Nonetheless, online surveys are a viable method to recruit participants within a short period at multiple instances and to circumvent the impossibility of using other recruitment and sampling strategies that require actual fieldwork (eg, face-to-face interview) due to the pandemic outbreak.

    This study adopted a repeated cross-sectional approach where sample population were different at each wave. Caution is needed to interpret changes over time and between-subject variability across waves needed to be considered. The same limitation applies to establishing causation from data across time points. However, our results are still comparable across waves and representative of the adult UK population.

    Questions on suicidal thoughts and self-harm were asked as a result of the pandemic. Thus, interpretation of the prevalences requires caution as the actual prevalence of these outcomes may be underestimated due to under-reporting of instances not related to the pandemic. As a public mental health survey, we did not include validated measures for screening of mental disorders and psychiatric conditions such as depression and anxiety as adopted by others.52 Not all individuals with mental health conditions experienced the same level of stress during the pandemic in the same way (eg, those with social phobia, differences in symptom severity and chronicity53). Thus, our findings of a strong association between pre-existing mental health conditions and having suicidal thoughts/self-harm may vary across those with specific mental health issues. Not all participants provided conclusive answers on having pre-existing mental health conditions and we found similar relative risks of having suicidal thoughts/self-harm between participants having mental health conditions and those with missing/inconclusive responses. Given such similarity, we may surmise that those who made inconclusive responses were likely to have mental health conditions. However, we cannot reliably exclude other factors that may have contributed to the elevated risks. In any case, we demonstrated that having pre-existing mental health conditions/psychiatric diagnosis was a robust factor associated with suicidal thoughts and self-harm.

    Other limitations include the usual caveats of using self-reported data in surveys. We did not collect data on suicidal thoughts, self-harm outcomes and coping with stress at wave 1, thus baseline levels of these outcomes before the lockdown could not be assessed. Similarly, not all variables (eg, pre-existing mental health conditions) were available for analysis until wave 3 and wave 4. This could limit our ability to assess time trends and interactions between risk factors and time. If these variables were available at wave 1, we might be able to reveal temporal trends that may have been observed shortly before and after the lockdown (as was found for loneliness and in other studies that compare prepandemic and postpandemic/lockdown mental health outcomes39 40 54). We also considered our data as preliminary given the early stage of our study with limited number of waves of data collection and the rapidly changing situation of the ongoing pandemic and the associated policy response. Our socioeconomic variable assessing social grade was dichotomous and as such crude in the assessment of the impact of socioeconomic adversity which may account for our findings and working status provided a more robust assessment. We did not collect data on participants’ occupations, thus we were unable to analyse outcomes for specific occupation, for example, healthcare or key workers. Due to model convergence issues, we could not include furlough status into the regression models and thus comparisons of outcomes between individuals who were or were not furloughed during the pandemic remained descriptive.

    Implications for policy and practice

    The importance of loneliness for mental health and well-being has recently been recognised in policy.55 56 Our findings on loneliness, coping, young age and pre-existing mental health conditions being risk factors for suicidal thoughts and self-harm during the COVID-19 pandemic can better inform policies and intervention strategies for both vulnerable individuals and the general population. While a rise in suicide as a consequence of the pandemic is not inevitable,5 the rise in loneliness in young people and the association with suicidal thoughts and self-harm are of concern particularly given the findings of a recent report from the National Child Mortality Database which identifies a potential increase (although not significant) in suicides in under 18 year-olds in the early period of time after lockdown.57

    In view of our findings, we argue that any lockdown measures should be accompanied by universal interventions aimed at the general population that could: (1) protect mental health against loneliness and isolation; and (2) enhance coping and resilience under stress. The robust association between loneliness and suicidality in young people in our study could partially explain the deterioration of mental health subsequent to the pandemic containment measures which appears to have disproportionately affected young adults.18 40 Hence, efforts should be made to repurpose existing psychological strategies for promoting mental health and behavioural change in young people that are suitable for social isolation/quarantine settings.3 18 Recent reviews suggested benefits of using digital interventions, for example, computerised cognitive behavioural therapy for young people when conventional therapies are limited or delayed.58 59 However, further research evidence is required to evaluate the effectiveness of these digital interventions particularly for individuals lacking adequate digital resources, as well as to examine potential privacy issues.3 5 18 Selective interventions may include exploring more effective ways of signposting, delivering mental health services and assisting individuals in crisis particularly for vulnerable groups,3 5 as well as locally driven peer support programmes.

    Our study suggests that coping poorly, young age, pre-existing mental health conditions and loneliness are associated with suicidal thoughts while the same risk factors except loneliness are associated with self-harm behaviours during the early phases of the COVID-19 pandemic in the UK. While global attention is still dominated by the COVID-19 as the pandemic evolves, the associated impact on mental health, suicide and self-harm should not be overlooked and may be prolonged.4 5 Efforts to foster collaborative research between academics, government and service monitoring of trends in mental health and suicidal behaviours/self-harm and informing interventions should be prioritised in order to mitigate potential harms to mental health and support rapid translation in this evolving pandemic. Policy decisions about this and any future waves or pandemics should take into account the risks of loneliness.

    Data availability statement

    Data are available upon reasonable request. The data that support the findings of this study are available from the Mental Health Foundation, contactable through the corresponding author upon reasonable request once the study is completed.

    Ethics statements

    Patient consent for publication

    Ethics approval

    Ethics approval for this study has been granted by the Cambridge Psychology Research Ethics Committee of the University of Cambridge, UK (No PRE 2020.050).

    Acknowledgments

    Professor Mike Gravenor and Professor Alan Watkins provided statistical advice. Lucy Thorpe, Catherine Seymour, Richard Grange and Rachel Baird (Mental Health Foundation) provided conceptual and operational support for the main study.

    References

    1. ↵
      1. Lythgoe MP,
      2. Middleton P
      . Ongoing clinical trials for the management of the COVID-19 pandemic. Trends Pharmacol Sci 2020;41:363–82.doi:10.1016/j.tips.2020.03.006pmid:http://www.ncbi.nlm.nih.gov/pubmed/32291112
      OpenUrlCrossRefPubMed
    2. ↵
    3. ↵
      1. Holmes EA,
      2. O'Connor RC,
      3. Perry VH, et al
      . Multidisciplinary research priorities for the COVID-19 pandemic: a call for action for mental health science. Lancet Psychiatry 2020;7:547–60.doi:10.1016/S2215-0366(20)30168-1pmid:http://www.ncbi.nlm.nih.gov/pubmed/32304649
      OpenUrlPubMed
    4. ↵
      1. Torales J,
      2. O'Higgins M,
      3. Castaldelli-Maia JM, et al
      . The outbreak of COVID-19 coronavirus and its impact on global mental health. Int J Soc Psychiatry 2020;66:317–20.doi:10.1177/0020764020915212pmid:http://www.ncbi.nlm.nih.gov/pubmed/32233719
      OpenUrlPubMed
    5. ↵
      1. Gunnell D,
      2. Appleby L,
      3. Arensman E, et al
      . Suicide risk and prevention during the COVID-19 pandemic. Lancet Psychiatry 2020;7:468–71.doi:10.1016/S2215-0366(20)30171-1pmid:http://www.ncbi.nlm.nih.gov/pubmed/32330430
      OpenUrlPubMed
    6. ↵
      1. Tian F,
      2. Li H,
      3. Tian S, et al
      . Psychological symptoms of ordinary Chinese citizens based on SCL-90 during the level I emergency response to COVID-19. Psychiatry Res 2020;288:112992.doi:10.1016/j.psychres.2020.112992pmid:http://www.ncbi.nlm.nih.gov/pubmed/32302816
      OpenUrlCrossRefPubMed
    7. ↵
      1. Wang C,
      2. Pan R,
      3. Wan X, et al
      . A longitudinal study on the mental health of general population during the COVID-19 epidemic in China. Brain Behav Immun 2020;87:40–8.doi:10.1016/j.bbi.2020.04.028pmid:http://www.ncbi.nlm.nih.gov/pubmed/32298802
      OpenUrlCrossRefPubMed
    8. ↵
      1. Chan SMS,
      2. Chiu FKH,
      3. Lam CWL, et al
      . Elderly suicide and the 2003 SARS epidemic in Hong Kong. Int J Geriatr Psychiatry 2006;21:113–8.doi:10.1002/gps.1432pmid:http://www.ncbi.nlm.nih.gov/pubmed/16416469
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Cheung YT,
      2. Chau PH,
      3. Yip PSF
      , . A revisit on older adults suicides and severe acute respiratory syndrome (SARS) epidemic in Hong Kong. Int J Geriatr Psychiatry 2008;23:1231–8.doi:10.1002/gps.2056pmid:http://www.ncbi.nlm.nih.gov/pubmed/18500689
      OpenUrlCrossRefPubMed
    10. ↵
      1. Perlman D,
      2. Peplau L
      . Toward a social psychology of loneliness. Pers Relatsh 1981;3:31–56.
      OpenUrl
    11. ↵
      1. Calati R,
      2. Ferrari C,
      3. Brittner M, et al
      . Suicidal thoughts and behaviors and social isolation: a narrative review of the literature. J Affect Disord 2019;245:653–67.doi:10.1016/j.jad.2018.11.022pmid:http://www.ncbi.nlm.nih.gov/pubmed/30445391
      OpenUrlPubMed
    12. ↵
      1. Hedley D,
      2. Uljarević M,
      3. Wilmot M, et al
      . Understanding depression and thoughts of self-harm in autism: a potential mechanism involving loneliness. Res Autism Spectr Disord 2018;46:1–7.doi:10.1016/j.rasd.2017.11.003
      OpenUrl
    13. ↵
      1. McClelland H,
      2. Evans JJ,
      3. Nowland R, et al
      . Loneliness as a predictor of suicidal ideation and behaviour: a systematic review and meta-analysis of prospective studies. J Affect Disord 2020;274:880–96.doi:10.1016/j.jad.2020.05.004pmid:http://www.ncbi.nlm.nih.gov/pubmed/32664029
      OpenUrlCrossRefPubMed
    14. ↵
      1. Mushtaq R,
      2. Shoib S,
      3. Shah T
      . Relationship between loneliness, Psychiatric disorders and physical health ? A review on the psychological aspects of loneliness. J Clin Diagnostic Res 2014;8.doi:10.7860/JCDR/2014/10077.4828
    15. ↵
      1. Stickley A,
      2. Koyanagi A
      . Loneliness, common mental disorders and suicidal behavior: findings from a general population survey. J Affect Disord 2016;197:81-7.doi:10.1016/j.jad.2016.02.054pmid:http://www.ncbi.nlm.nih.gov/pubmed/26971125
      OpenUrlPubMed
    16. ↵
      1. Stravynski A,
      2. Boyer R
      . Loneliness in relation to suicide ideation and parasuicide: a population-wide study. Suicide Life Threat Behav 2001;31:32–40.doi:10.1521/suli.31.1.32.21312pmid:http://www.ncbi.nlm.nih.gov/pubmed/11326767
      OpenUrlPubMedWeb of Science
    17. ↵
      1. Teo AR,
      2. Marsh HE,
      3. Forsberg CW, et al
      . Loneliness is closely associated with depression outcomes and suicidal ideation among military veterans in primary care. J Affect Disord 2018;230:42–9.doi:10.1016/j.jad.2018.01.003pmid:http://www.ncbi.nlm.nih.gov/pubmed/29407537
      OpenUrlPubMed
    18. ↵
      1. Loades ME,
      2. Chatburn E,
      3. Higson-Sweeney N, et al
      . Rapid systematic review: the impact of social isolation and loneliness on the mental health of children and adolescents in the context of COVID-19. J Am Acad Child Adolesc Psychiatry 2020;59:1218-1239.e3.doi:10.1016/j.jaac.2020.05.009pmid:http://www.ncbi.nlm.nih.gov/pubmed/32504808
      OpenUrlPubMed
    19. ↵
      1. Elliott JL,
      2. Frude N,
      3. Stress FN
      . Stress, coping styles, and hopelessness in self-poisoners. Crisis 2001;22:20–6.doi:10.1027//0227-5910.22.1.20pmid:http://www.ncbi.nlm.nih.gov/pubmed/11548816
      OpenUrlPubMed
    20. ↵
      1. Mitchell AM,
      2. Crane PA,
      3. Kim Y
      . Perceived stress in survivors of suicide: psychometric properties of the perceived stress scale. Res Nurs Health 2008;31:576–85.doi:10.1002/nur.20284pmid:http://www.ncbi.nlm.nih.gov/pubmed/18449942
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Zweig RA,
      2. Hinrichsen GA
      . Factors associated with suicide attempts by depressed older adults: a prospective study. Am J Psychiatry 1993;150:1687–92.doi:10.1176/ajp.150.11.1687pmid:http://www.ncbi.nlm.nih.gov/pubmed/8214178
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Ellis JB,
      2. Lamis DA,
      3. Characteristics A
      . Adaptive characteristics and suicidal behavior: a gender comparison of young adults. Death Stud 2007;31:845–54.doi:10.1080/07481180701537303pmid:http://www.ncbi.nlm.nih.gov/pubmed/17886414
      OpenUrlPubMed
    23. ↵
      1. Garlow SJ,
      2. Rosenberg J,
      3. Moore JD, et al
      . Depression, desperation, and suicidal ideation in college students: results from the American foundation for suicide prevention College screening project at Emory University. Depress Anxiety 2008;25:482–8.doi:10.1002/da.20321pmid:http://www.ncbi.nlm.nih.gov/pubmed/17559087
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Heikkinen M,
      2. Aro H,
      3. Lönnqvist J
      . Life events and social support in suicide. Suicide Life Threat Behav 1993;23:343–58.pmid:http://www.ncbi.nlm.nih.gov/pubmed/8310468
      OpenUrlPubMedWeb of Science
    25. ↵
      1. Linda WP,
      2. Marroquín B,
      3. Miranda R
      . Active and passive problem solving as moderators of the relation between negative life event stress and suicidal ideation among suicide attempters and non-attempters. Arch Suicide Res 2012;16:183–97.doi:10.1080/13811118.2012.695233pmid:http://www.ncbi.nlm.nih.gov/pubmed/22852781
      OpenUrlPubMed
    26. ↵
      1. Bickford D,
      2. Morin RT,
      3. Woodworth C, et al
      . The relationship of frailty and disability with suicidal ideation in late life depression. Aging Ment Health 2021;25:1–6.doi:10.1080/13607863.2019.1698514pmid:http://www.ncbi.nlm.nih.gov/pubmed/31809584
      OpenUrlPubMed
    27. ↵
      COVID-19 research Tracker, 2021. Available: https://www.closer.ac.uk/covid19-longitudinal-research-hub/covid19-research-evidence/# [Accessed 10 Jun 2021].
    28. ↵
      1. Iob E,
      2. Steptoe A,
      3. Fancourt D
      . Abuse, self-harm and suicidal ideation in the UK during the COVID-19 pandemic. Br J Psychiatry 2020;217:1–4.doi:10.1192/bjp.2020.130pmid:http://www.ncbi.nlm.nih.gov/pubmed/32654678
      OpenUrlPubMed
    29. ↵
      1. O'Connor RC,
      2. Wetherall K,
      3. Cleare S, et al
      . Mental health and well-being during the COVID-19 pandemic: longitudinal analyses of adults in the UK COVID-19 Mental Health & Wellbeing study. Br J Psychiatry 2021;218:326–33.doi:10.1192/bjp.2020.212
      OpenUrlCrossRefPubMed
    30. ↵
      1. Van Bortel T,
      2. John A,
      3. Solomon S, et al
      . Mental health in the pandemic: a repeated cross-sectional mixed-method study protocol to investigate the mental health impacts of the coronavirus pandemic in the UK. BMJ Open 2021;11:e046422.doi:10.1136/bmjopen-2020-046422pmid:http://www.ncbi.nlm.nih.gov/pubmed/34452957
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. YouGov
      . YouGov’s client privacy and cookies notice, 2017. Available: https://yougov.co.uk/about/client-privacy-and-cookies-notice/ [Accessed 10 Jun 2021].
    32. ↵
      1. StataCorp
      . Stata statistical software: release 16. College Station, TX: StataCorp LLC, 2019.
    33. ↵
      1. Cummings P
      . Methods for estimating adjusted risk ratios. Stata J 2009;9:175–96.doi:10.1177/1536867X0900900201
      OpenUrlWeb of Science
    34. ↵
      1. Korn EL,
      2. Graubard BI
      . Analyses Using Multiple Surveys. In: Analysis of health surveys. Hoboken, NJ: John Wiley & Sons, 1999.
    35. ↵
      1. Dong Y,
      2. Peng C-YJ
      . Principled missing data methods for researchers. Springerplus 2013;2:222.doi:10.1186/2193-1801-2-222pmid:http://www.ncbi.nlm.nih.gov/pubmed/23853744
      OpenUrlCrossRefPubMed
    36. ↵
      1. Bebbington PE,
      2. McManus S
      . Revisiting the one in four: the prevalence of psychiatric disorder in the population of England 2000-2014. Br J Psychiatry 2020;216:55–7.doi:10.1192/bjp.2019.196pmid:http://www.ncbi.nlm.nih.gov/pubmed/31685036
      OpenUrlPubMed
    37. ↵
      1. Fancourt D,
      2. Steptoe A,
      3. Bu F
      . Trajectories of anxiety and depressive symptoms during enforced isolation due to COVID-19 in England: a longitudinal observational study. Lancet Psychiatry 2021;8:141–9.doi:10.1016/S2215-0366(20)30482-Xpmid:http://www.ncbi.nlm.nih.gov/pubmed/33308420
      OpenUrlPubMed
    38. ↵
      1. Iob E,
      2. Frank P,
      3. Steptoe A, et al
      . Levels of severity of depressive symptoms among at-risk groups in the UK during the COVID-19 pandemic. JAMA Netw Open 2020;3:e2026064.doi:10.1001/jamanetworkopen.2020.26064pmid:http://www.ncbi.nlm.nih.gov/pubmed/33104209
      OpenUrlPubMed
    39. ↵
      1. Kwong ASF,
      2. Pearson RM,
      3. Adams MJ, et al
      . Mental health before and during the COVID-19 pandemic in two longitudinal UK population cohorts. Br J Psychiatry 2021;218:334–43.doi:10.1192/bjp.2020.242
      OpenUrlPubMed
    40. ↵
      1. Pierce M,
      2. Hope H,
      3. Ford T, et al
      . Mental health before and during the COVID-19 pandemic: a longitudinal probability sample survey of the UK population. Lancet Psychiatry 2020;7:883-892.doi:10.1016/S2215-0366(20)30308-4pmid:http://www.ncbi.nlm.nih.gov/pubmed/32707037
      OpenUrlPubMed
    41. ↵
      1. Pierce M,
      2. McManus S,
      3. Hope H, et al
      . Mental health responses to the COVID-19 pandemic: a latent class trajectory analysis using longitudinal UK data. Lancet Psychiatry 2021;8:610-619.doi:10.1016/S2215-0366(21)00151-6pmid:http://www.ncbi.nlm.nih.gov/pubmed/33965057
      OpenUrlPubMed
    42. ↵
      1. Arnett JJ
      . Emerging adulthood. A theory of development from the late teens through the twenties. Am Psychol 2000;55:469-80.pmid:http://www.ncbi.nlm.nih.gov/pubmed/10842426
      OpenUrlPubMed
    43. ↵
      1. Duffy ME,
      2. Twenge JM,
      3. Joiner TE
      . Trends in mood and anxiety symptoms and suicide-related outcomes among U.S. undergraduates, 2007–2018: evidence from two national surveys. J Adolesc Heal 2019;65:590–8.doi:10.1016/j.jadohealth.2019.04.033
      OpenUrl
    44. ↵
      1. Shanahan MJ
      . Pathways to adulthood in changing societies: variability and mechanisms in life course perspective. Annu Rev Sociol 2000;26:667–92.doi:10.1146/annurev.soc.26.1.667
      OpenUrlCrossRefWeb of Science
    45. ↵
      1. Shanahan L,
      2. Steinhoff A,
      3. Bechtiger L, et al
      . Emotional distress in young adults during the COVID-19 pandemic: evidence of risk and resilience from a longitudinal cohort study. Psychol Med 2020:1–10.doi:10.1017/S003329172000241Xpmid:http://www.ncbi.nlm.nih.gov/pubmed/32571438
      OpenUrlPubMed
    46. ↵
      1. Bell DNF,
      2. Blanchflower DG
      . Us and UK labour markets before and during the COVID-19 crash. Natl Inst Econ Rev 2020;252:R52–69.doi:10.1017/nie.2020.14
      OpenUrlCrossRef
    47. ↵
      1. Cortes GM,
      2. Forsythe E
      . The heterogeneous labor market impacts of the Covid-19 pandemic. SSRN Electron J 2020.doi:10.2139/ssrn.3634715
    48. ↵
      1. Pierce M,
      2. McManus S,
      3. Jessop C, et al
      . Says who? the significance of sampling in mental health surveys during COVID-19. Lancet Psychiatry 2020;7:567–8.doi:10.1016/S2215-0366(20)30237-6pmid:http://www.ncbi.nlm.nih.gov/pubmed/32502467
      OpenUrlPubMed
    49. ↵
      1. Hays RD,
      2. Liu H,
      3. Kapteyn A
      . Use of Internet panels to conduct surveys. Behav Res Methods 2015;47:685–90.doi:10.3758/s13428-015-0617-9pmid:http://www.ncbi.nlm.nih.gov/pubmed/26170052
      OpenUrlPubMed
    50. ↵
      1. HESA
      . Higher education student statistics: UK, 2019/20, 2021. Available: https://www.hesa.ac.uk/news/27-01-2021/sb258-higher-education-student-statistics [Accessed 10 Jun 2021].
    51. ↵
      1. ONS
      . ONS Mid-Year Population Estimates - Custom Age Tables, 2020. Available: https://data.gov.uk/dataset/b1f548a4-694c-438b-9551-af7a121e3395/ons-mid-year-population-estimates-custom-age-tables [Accessed 10 Jun 2021].
    52. ↵
      1. Katikireddi SV,
      2. Niedzwiedz CL,
      3. Popham F
      . Trends in population mental health before and after the 2008 recession: a repeat cross-sectional analysis of the 1991-2010 health surveys of England. BMJ Open 2012;2:e001790.doi:10.1136/bmjopen-2012-001790pmid:http://www.ncbi.nlm.nih.gov/pubmed/23075569
      OpenUrlAbstract/FREE Full Text
    53. ↵
      1. Pan K-Y,
      2. Kok AAL,
      3. Eikelenboom M, et al
      . The mental health impact of the COVID-19 pandemic on people with and without depressive, anxiety, or obsessive-compulsive disorders: a longitudinal study of three Dutch case-control cohorts. Lancet Psychiatry 2021;8:121–9.doi:10.1016/S2215-0366(20)30491-0pmid:http://www.ncbi.nlm.nih.gov/pubmed/33306975
      OpenUrlPubMed
    54. ↵
      1. Daly M,
      2. Sutin AR,
      3. Robinson E
      . Longitudinal changes in mental health and the COVID-19 pandemic: evidence from the UK household longitudinal study. Psychol Med 2020:1–10.doi:10.1017/S0033291720004432pmid:http://www.ncbi.nlm.nih.gov/pubmed/33183370
      OpenUrlPubMed
    55. ↵
      1. HM Government
      . A connected Society: a strategy for tackling loneliness – laying the foundations for change, 2018. Available: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/750909/6.4882_DCMS_Loneliness_Strategy_web_Update.pdf
    56. ↵
      1. Jo Cox Commission on Loneliness
      . Combatting loneliness one conversation at a time: a call to action, 2017. Available: https://www.ageuk.org.uk/globalassets/age-uk/documents/reports-and-publications/reports-and-briefings/active-communities/rb_dec17_jocox_commission_finalreport.pdf
    57. ↵
      1. Odd D,
      2. Sleap V,
      3. Appleby L
      . Child suicide rates during the COVID-19 pandemic in England: real-time surveillance. Bristol, 2020. Available: https://www.ncmd.info/wp-content/uploads/2020/07/REF253-2020-NCMD-Summary-Report-on-Child-Suicide-July-2020.pdf
    58. ↵
      1. Grist R,
      2. Croker A,
      3. Denne M, et al
      . Technology delivered interventions for depression and anxiety in children and adolescents: a systematic review and meta-analysis. Clin Child Fam Psychol Rev 2019;22:147–71.doi:10.1007/s10567-018-0271-8pmid:http://www.ncbi.nlm.nih.gov/pubmed/30229343
      OpenUrlPubMed
    59. ↵
      1. Pennant ME,
      2. Loucas CE,
      3. Whittington C, et al
      . Computerised therapies for anxiety and depression in children and young people: a systematic review and meta-analysis. Behav Res Ther 2015;67:1–18.doi:10.1016/j.brat.2015.01.009pmid:http://www.ncbi.nlm.nih.gov/pubmed/25727678
      OpenUrlCrossRefPubMed

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Twitter @ProfAnnJohn, @TineVanBortel, @AKousoulis

    • Contributors AJ, AAK and TVB conceived the study. AJ, SCL, SS, DC-K, SM, ADM, GD, TVB and AAK designed the study. SS oversaw data collection. AJ supervised the study. SCL conducted the analysis. SCL and AJ wrote the initial draft. AJ, SCL, SS, DC-K, SM, ADM, GD, TVB and AAK commented on the interpretation of findings and the manuscript. AJ is the guarantor of this study who accepts full responsibility for the finished work and/or the conduct of the study, had access to the data, and controlled the decision to publish.

    • Funding The study is funded by MQ Transforming Mental Health (MQBF/3 ADP), the National Institute for Health Research Applied Research Collaboration (NIHR ARC EoE/U.SY.SYBJ.GAAB) and the Mental Health Foundation UK (MHF/G105979), with further in-kind or human resource contributions from the University of Cambridge, Swansea University, Strathclyde University and Queens University Belfast, UK. The Waterloo Foundation and the Manolo Blahnik International have also provided smaller funding contributions for the study.

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.