key: cord-1019946-rnkti5s2 authors: Robinson, E.; Sutin, A. R.; Daly, M.; Jones, A. title: A systematic review and meta-analysis of longitudinal cohort studies comparing mental health before versus during the COVID-19 pandemic date: 2021-03-08 journal: nan DOI: 10.1101/2021.03.04.21252921 sha: 88c63e52852890e28a94d1fd59d274b0f991e37c doc_id: 1019946 cord_uid: rnkti5s2 Background: Increases in mental health problems have been observed in some studies during the COVID-19 pandemic. It is unclear whether changes have been large and experienced by most population sub-groups, persisted over time or been symptom specific. Methods: We systematically reviewed and meta-analysed longitudinal cohort studies that examined changes in mental health among the same group of participants before and during the pandemic (PROSPERO: CRD42021231256). Searches for published and unpublished studies were conducted in January 2021. Changes in mental health (standardised mean change; SMC) were examined using meta-analyses. Findings: Sixty-five studies were included. There was an overall increase in mental health symptoms that was most pronounced during March-April 2020 (SMC = .102 [95% CI: .026 to .192], p = 0.03) before significantly declining over time (May-July SMC = .067 [95% CI: -.022 to .157], p = .141). Compared to measures of anxiety (SMC = 0.13, p = 0.02) and general mental health (SMC = -.03, p = 0.65), increases in depression and mood disorder symptoms tended to be larger (SMC = 0.22, p < .001) and reductions over time appeared less pronounced. Increased mental health symptoms were observed across most population subgroups examined but there was no evidence of any change in symptoms among samples with a pre-existing mental health condition. Interpretation: There was a small increase in mental health symptoms soon after the outbreak of the COVID-19 pandemic that decreased and was comparable to pre-pandemic levels by mid-2020 among most population sub-groups and symptom types. Funding: N/A There have been reported increases in mental health problems during the outbreak of the COVID-19 pandemic. However, it is unclear whether changes in mental health problems have been symptom specific, how changes have differed across populations and whether increased mental health problems have persisted over time. We systematically reviewed and meta-analysed longitudinal cohort studies that examined mental health among the same participants prior to and during the pandemic in 2020. This approach allowed us to quantify the mental health burden associated with the outbreak of the pandemic and how it has changed over time. We searched Pubmed, SCOPUS, Web of Science and PsychInfo from January 2020 to January 11, 2021 and identified eligible unpublished articles available on pre-print servers. We identified 65 eligible articles that reported 201 comparisons of mental health pre vs. post pandemic outbreak. Meta-analysis indicated that longitudinal cohort studies that examined mental health prior to and during the COVID-19 pandemic in 2020 showed a significant but statistically small increase in mental health symptoms. The overall increase in mental health symptoms was most pronounced during the early stages of the pandemic (March-April), before decreasing and being generally comparable to pre-pandemic levels by mid-2020. Compared to anxiety and general measures of mental health functioning, increases tended to be larger in depressive symptoms and although statistically small, remained elevated past the early stages of the pandemic. Increases in mental health symptoms were observed across most population sub-groups, but there was no evidence of a change in mental health symptoms among samples of participants with a pre-existing mental health condition. Findings confirm that the initial outbreak of the pandemic was associated with a significant but statistically small increase in mental health symptoms. Given that small effects may have meaningful cumulative consequences at the population level, there is a need for continued mental health provision and monitoring particularly during periods of the pandemic when infection rates and deaths are high. Further into the pandemic, mental health problems decreased significantly, which indicated recovery and resilience in overall mental health. Contrary to predictions made early in the pandemic, there was also no evidence of a worsening of mental health symptoms among samples of participants with a pre-existing mental health condition. Overall the results of the present analyses suggest that the pandemic may not have caused an unprecedented and long lasting mental health crisis, instead there appears to have been resilience in mental health. As of February 2020, the COVID-19 pandemic has been responsible for more than 2.5 million deaths worldwide (1) . Soon after the COVID-19 crisis was declared a pandemic by the World Health Organisation (WHO) on the 11 th March, concerns were raised over a potential parallel mental health crisis fuelled by the pandemic and the associated social restrictions imposed by governments to reduce virus transmission (2, 3) . The mental health of 'at risk' groups during the pandemic, such as those with pre-existing mental health conditions, has also been highlighted as cause for concern (3, 4). There are considerable methodological challenges to quantifying the impact that the pandemic has had on mental health. During the outbreak of the pandemic studies indicated that many participants perceived their mental health had worsened (5, 6), but retrospective reports of change in mental health are prone to substantial bias (7, 8) . Other studies have found a greater incidence of mental health problems in cohorts recruited during the pandemic compared to different cohorts who completed measures prior to the pandemic (9, 10). However, differences in how samples were recruited (e.g. greater reliance of online and nonprobability based samples during the pandemic) and differences in demographic profiles of pre vs. post-pandemic outbreak cohorts also make inferring change in mental health attributable to the pandemic difficult (11) . A number of longitudinal cohort studies have sampled the same participants before and during the pandemic to examine how mental health has changed. In a large nationally representative sample of UK adults, Daly et al. found that non-specific general mental health symptoms (GHQ-12) increased in April-June 2020 compared to a pre-pandemic baseline (12) . Other longitudinal cohort studies have found little change in mental health (13) or mixed evidence (14) . Longitudinal studies are uniquely placed to characterise the time course of the mental health burden associated with the COVID-19 pandemic. Although the effects of All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 8, 2021. ; https://doi.org/10.1101/2021.03.04.21252921 doi: medRxiv preprint the pandemic on mental health could be long lasting (15) , a recent multi-wave longitudinal study of US adults found that after an initial increase in distress during the early stages of the pandemic, distress reduced to pre-pandemic levels within a few months (16) . There has been no systematic review and meta-analysis of longitudinal cohort studies with pre-pandemic measurement of mental health (17) (18) (19) . We systematically reviewed and meta-analysed studies that examined longitudinal changes in mental health among the same sample of participants before vs. during the pandemic, in order to quantify the size of the mental health burden associated with the pandemic and how this burden changed over time. Secondary aims were to examine whether changes have been symptom specific and whether changes differed based on population demographics. Participants. To be eligible, studies were required to have sampled the same cohort of participants prior to 11/03/20 (date the WHO declared a pandemic) (20) and at least once after this date. Chinese studies were eligible (but analysed separately) if mental health was assessed prior to and after 23/01/20 because substantial social restriction measures were enforced across China from this point (21) . There were no limits on populations sampled. To be eligible, studies were required to have collected data using a validated multi-item measure of mental health symptoms or mental well-being, such as depression (e.g. Patient Health Questionnaire: PHQ9), anxiety (Depression, Anxiety, Stress Scale: DASS), non-specific general mental health related functioning and distress (General Health Questionnaire: GHQ12, Kessler) and well-being (Warwick-Edinburgh Mental Wellbeing Scale). As our focus was on mental health symptoms, ineligible measures included loneliness, stress and physical health related quality of life. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Outcome. Studies that examined continuous changes (i.e. standardised mean change; SMC) in mental health symptoms were eligible, as was change in the % of the sample meeting questionnaire specific cut-offs for clinically relevant/likely serious mental health problems were eligible (i.e. Odds Ratio). Studies were required to sample the same participants using the same measure of mental health pre and post-pandemic (repeated cross-sectional studies were not eligible). If only a sub-sample of participants were followed up across survey waves, only data from the sub-sample were eligible. Interventions to improve mental health during the pandemic were not eligible. If multiple articles reported on data from the same cohort of participants, the article with the largest number of post-pandemic follow-up data collection points was included. Journal articles and pre-prints were eligible. We searched Pubmed, SCOPUS, Web of Science and PsychInfo from January 2020 to January 11, 2021, using combinations of coronavirus and mental health relevant search terms (see online supplementary materials). One author conducted title and abstract screening and 25% were cross-checked by a second author (no discrepancies). Two authors conducted fulltext screening and disagreements in eligibility were resolved by discussion. We searched three databases for unpublished pre-prints; Open Science Framework (inclusive of 30 preprint archives, e.g. PsychArxiv), MedrXiv and the Social Science Research Network, and conducted forward citation tracking (Google Scholar) for all eligible articles. A single author conducted pre-print searches and forward citation tracking; a second author verified eligibility of identified articles. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The following information was extracted by a single author and checked for accuracy by a second author; country of study, participant group sampled, age group of sample, sampling strategy used (e.g. use of representative sampling vs. convenience), pre and post pandemic dates of data collection, mental health measure, analytic treatment of mental health symptomology change (e.g. use of change in % meeting questionnaire cut-off vs. continuous change in questionnaire score), level of attrition (%), effect size information (authors were contacted if information was missing) and whether the study was reported in a journal article or pre-print. We reviewed methodological quality scales and risk of bias measures to develop a list of bias indicators relevant to included studies. See online supplementary materials for full detail (including justifications for each indicator). Indicators were rated by two authors; i) whether the study reported representative sampling (yes = lower in risk of bias), ii) whether the study underwent peer review (yes = lower in risk of bias), iii) low level of attrition unlikely to affect results (yes = lower in risk of bias), iv) whether the study had a relatively large (N≥1000) sample size (yes = lower in risk of bias), v) whether the pre-pandemic measure of mental health was collected within the last year of the post-pandemic outbreak measure (yes = lower in risk of bias), vi), whether survey delivery mode (e.g. online) was consistent across pre and post outbreak waves of data collection (yes = lower in risk of bias), vii) whether conflicts of interest were reported (no conflicts = lower in risk of bias). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. To examine if heterogeneity of change in mental health was attributable to country level factors, for each eligible study, we identified the number of recorded COVID cases and deaths (by country) per million population for the month that mental health was measured during the pandemic. We also used data from the Oxford COVID-19 Government Response Tracker to characterise each country's severity of social restrictions, number of health measures and level of economic support in place during the month post-pandemic outbreak mental health was assessed. See supplementary materials for full information. For our main analysis on continuous data, we computed standardised mean change (SMC) in pooled SD units For analyses examining questionnaire cut-off data we computed Marginal Odds Ratios to quantify size of change. See online supplementary materials for full statistical information. As studies contributed multiple comparisons to analyses, we conducted a multilevel meta-analysis. Heterogeneity was assessed using I 2 statistic. We examined whether time (month post-pandemic outbreak measure was collected) predicted change in mental health symptoms using meta-regression. We also conducted sub-group analyses on symptom type (depression, anxiety, general mental health functioning (including distress), well-being, psychotic symptoms, other), gender, age (adult vs. child/adolescent), population sampled (general population, university students, mental and physical health conditions) and continent sample was from. A minimum of n=5 effects for each sub-group was required for analysis. All analyses were conducted in R. To address influential cases we conducted leave one-outanalyses and computed DFBETAS values for each effect size in the full models (without moderators). Funnel plots were inspected for potential publication bias, Egger's test of All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Article identification. After removal of duplicates, title and abstract screening of electronic database search results and identification of eligible articles through other sources, 153 articles were full-text screened. A total of 65 articles were eligible for inclusion. See Figure 1 for study selection flow chart. Not a research report indicates manuscript did not report on a study. Excluded as unable to include in metaanalysis (N=1) was due to authors not providing the required statistical information to compute effect size. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. See Table S1 in the supplementary materials for individual study information. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Table 1 for SMCs for each symptom type by time. Sub-group analyses. Across sub-groups, we found no evidence that change in mental health symptoms differed based on age, gender, or study continent. See Table 2 . Change in 1 Multi-level models remained a better fit of the data separately for anxiety, depression and general mental health measures, but not for psychosis, well-being or other symptoms which is likely due to smaller number of effect sizes. Regardless, treating the data as single or multilevel in these cases did not substantially influence the effect sizes or statistical significance of the models. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 8, 2021. ; https://doi.org/10.1101/2021.03.04.21252921 doi: medRxiv preprint symptoms tended to be larger among participants with a pre-existing physical health condition compared to the general population. Notably, change in mental health symptoms was non-significant in samples with pre-existing mental health conditions. No country-level data (number of COVID cases/deaths, stringency of government measures or level) explained heterogeneity between samples (ps > .05). See online supplementary materials for full results. Due to the small number of studies examining changes in % of sample meeting questionnaire cut-offs, risk of bias analyses were limited to SMC studies. See Table 2 for risk of bias summary information and results and Table S2 for study-level information. We found limited evidence that the risk of bias indicators predicted size of change in symptoms, except that change in mental health tended to be larger when delivery mode of questionnaire was consistent pre vs. post pandemic. In an exploratory analysis limited to a more homogenous collection of studies that were of lower risk of bias results were consistent with the main analyses. See Figure 2 . All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Across sixty-five longitudinal cohort studies comparing mental health prior to and during the COVID-19 pandemic there was a statistically small overall increase in mental health symptoms. The increase in mental health symptoms was largest among studies that sampled participants in the early stages of the pandemic (March-April) and severity of mental health symptoms decreased significantly over the following months (May-July). This pattern of results may represent an acute and normal response to an unforeseen and distressing traumatic event (22) , which was then followed by a period of psychological adaptation and resilience (16, 23) . In line with this interpretation, a large study of US adults found that All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. perceived risk and worries about financial instability, infection and death were highest during the early stages of the pandemic (24) . However, as more information about the pandemic became available perceived risks decreased and this predicted recovery to pre-pandemic levels of distress (24) . Similarly, in a large sample of UK adults recruited after the pandemic outbreak, both anxiety and depressive symptoms showed a trajectory of recovery from the beginning of April onwards (25) . Worsening of mental health symptoms was largest in studies examining depression and mood disorders symptoms, and a small increase in depression symptoms was still observed during May-June. Change in anxiety symptoms was smaller and was not significant in samples measured in May-June. Changes in non-specific general mental health functioning (including distress) and well-being were small and non-significant. The more pronounced change in depressive symptoms observed may reflect the effects of isolation and sadness caused by social restrictions and loss of life during the pandemic (26) . Increases in mental health problems were observed across most population sub-groups sampled (e.g. general population, university students, existing physical health condition). Contrary to concerns raised early in the pandemic (3, 4) , changes in mental health were less pronounced among people with pre-existing mental health conditions and overall there was no statistically significant change in mental symptoms in this group. These findings may in part be explained by regression to the mean and naturally occurring recovery of more severe mental health symptoms over time (27) , as well as stay at home restrictions that potentially provided a more structured routine and reduced exposure to external stressors (e.g. large social gatherings) among those with severe mental health conditions (22) . A number of countries, such as China, introduced new mental health services in response to the pandemic (4) and this may have relieved symptoms among those with and without pre-existing mental health conditions. In contrast, mental health symptoms increased among those with pre-All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. A smaller sub-set of studies examined change in proportion of sampled population exceeding questionnaire cut-offs for clinically relevant mental health symptoms. In line with the main analyses, there was a statistically small increase in likelihood of meeting questionnaire cut-offs. A limitation of these studies is that although questionnaires used have been shown to be valid indicators of clinically relevant mental health disorder symptomology, it was common for response formats to be altered to ask participants to report on shorter time frames (e.g. the last week). Therefore, these studies provide an indication of acute symptom severity rather than clinical diagnostic value (e.g. anxiety disorder diagnosis typically requires symptoms for several months (28) ). In contrast to the fear that the COVID-19 pandemic would cause a parallel and longstanding mental health crisis (2, 3) , the present findings suggest that overall there has been considerable resilience in mental health. Data on recorded suicides align with this, as All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. there have been stable rates or decreases reported across a number of countries (29) . However, there is a need for continued mental health provision and monitoring particularly during periods of increased COVID-19 infection and death. The increase in depression and mood disorder symptoms that did not return to pre-pandemic levels warrants attention, as even a small upward shift in depressive symptoms may have meaningful cumulative consequences on the population-level. There are limitations to the studies included. The majority of studies sampled populations in developed countries during the early stages of the pandemic. Given that there have been second waves of the pandemic in many countries during early 2021, it will be important to continue to monitor mental health. Heterogeneity of models ended to be high and although some of this variability was explained by symptom type and population groups sampled, it will be important to identify further explanatory factors. For example, there are population sub-groups who may be at greater risk of declines in mental health that we were unable to examine and who are likely to be under-represented in studies of the general population. Due to stressful working conditions frontline healthcare workers may be at increased risk and there may be long-term mental health effects among those who become seriously ill from COVID-19, live in nursing homes or those who suffered financially as a result of the pandemic (30) . Participants with pre-existing mental health conditions tended to be grouped in studies and therefore we were unable to examine changes in symptoms among specific patient groups (e.g. schizophrenia). We found little evidence that study outcomes were strongly related to individual risk of bias indicators and in an analysis limited to relatively high-quality studies results were similar to the main analyses. Level of attrition was high across studies and this is a limitation. Although there was no significant evidence of attrition being associated with change in symptoms in meta-regression or sub-group analyses All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 8, 2021. ; and lower levels of attrition tended to be associated with a smaller as opposed to larger increase in symptoms. Among longitudinal cohort studies examining mental health prior to and during the COVID-19 pandemic in 2020 there was a significant but statistically small increase in mental health symptoms. Increases were larger for depressive symptoms compared to anxiety and measures of overall mental health functioning. The overall increase in mental health symptoms was most pronounced during the first two months of the pandemic, before decreasing and being comparable to pre-pandemic levels by mid-2020. There was no evidence of a worsening of mental health symptoms among samples of participants with a pre-existing mental health condition. Data sharing. Study data file and pre-registered analysis protocol are available online https://osf.io/rg24v/. Analysis code and files are available on request. Contributors. The study was designed by all authors. AJ analysed the data. The first draft was written by ER. All authors critically revised the manuscript and agree to be accountable for all aspects of the work. Role of the funding source. There was no funding source for this study. All authors report no conflicts of interest. ER has previously received funding from the American Beverage Association and Unilever for projects unrelated to the present research. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. a Number of comparisons does not add up to n=165 because for a small number of effects it was unclear whether the pre-pandemic measures was collected more than 12 months prior to the post-pandemic outbreak measure. b We also examined whether attrition was associated with the change in mental health symptoms pre-post pandemic using meta-regression. There was no significant association (coefficient = .092 [95% CI: -.185 to .370] z = .653, p =.514). World Health Organisation WH. WHO COVID-19 dashboard Multidisciplinary research priorities for the COVID-19 pandemic: a call for action for mental health science Mental health and the Covid-19 pandemic Patients with mental health disorders in the COVID-19 epidemic Obesity, eating behavior and physical activity during COVID-19 lockdown: A study of UK adults Immediate Psychological Responses and Associated Factors during the Initial Stage of the 2019 Coronavirus Disease (COVID-19) Epidemic among the General Population in China Accuracy and bias in retrospective symptom reporting Retrospective recall of affect in clinically depressed individuals and controls Psychological Distress and Loneliness Reported by US Adults Depression reported by US Says who? The significance of sampling in mental health surveys during COVID-19. The Lancet Psychiatry Anxiety and depression symptoms, and lack of emotional support among the general population before and during the COVID-19 pandemic. A prospective national study on prevalence and risk factors Impact of COVID-19 on loneliness, mental health, and health service utilisation: A prospective cohort study of older adults with multimorbidity in primary care The long term mental health impact of covid-19 must not be ignored Psychological distress and adaptation to the COVID-19 crisis in the United States The psychological impact of COVID-19 pandemic lockdowns: a review and meta-analysis of longitudinal studies and natural experiments The prevalence of psychological consequences of COVID-19: A systematic review and meta-analysis of observational studies Impact of COVID-19 pandemic on mental health in the general population: A systematic review WHO Declares COVID-19 a Pandemic Modelling the effects of Wuhan's lockdown during COVID-19, China. Bull World Health Organ 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. The Lancet Psychiatry Re-evaluating the notion that resilience is commonplace: A review and distillation of directions for future research, practice, and policy Explaining the rise and fall of psychological distress during the COVID-19 crisis in the United States: Longitudinal evidence from the Understanding America Study Trajectories of anxiety and depressive symptoms during enforced isolation due to COVID-19 in England: a longitudinal observational study. The Lancet Psychiatry The clinical significance of loneliness: A literature review Regression toward the mean: its etiology, diagnosis, and treatment Anxiety and DSM-5 Trends in suicide during the covid-19 pandemic Bidirectional associations between COVID-19 and psychiatric disorder: retrospective cohort studies of 62 354 COVID-19 cases in the USA All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.