RESEARCH QUESTION 1:
Changes in Mental Health Symptoms
Eligible studies must report changes in symptom levels, proportion of participants above a cutoff threshold, or proportion of participants who change by a pre-defined magnitude (e.g., minimal clinically important difference) across a delineated COVID-19 related event. This could include comparisons of pre-COVID-19 and COVID-19 symptoms, symptoms at the initiation of the outbreak to the peak, or symptoms during highly restrictive isolation periods to subsequent periods, for instance. Studies with < 100 participants are excluded.
We are not including cross-sectional studies that report percentages of participants with scores above cutoff thresholds on commonly used symptom questionnaires. Conclusions that can be drawn from that type of data about mental health effects from COVID-19 and clinical implications, however, are limited, and, per our protocol, we have not included those studies. This is because percentages of people who score above a threshold on standardized questionnaires vary, sometimes dramatically, between populations, even in normal times. For example, the percentage of participants with scores of at least 10 on the Patient-Health Questionnaire-9, a commonly used measure of depressive symptoms, in large, randomly selected, regional or national general population samples, has been reported as 4% in Hong Kong (N = 6,028); 6% in Germany (N = 5,018); 7% in Shanghai, China (N = 1,045); 8% in the United States (N = 10,257); 8% in the province of Alberta, Canada (N = 3,304); 11% in Sweden (N = 3,001); and 22% in Jiangsu Province, China (N = 8,400). Even within populations from the same region, the percentage can vary dramatically depending on sample characteristics. In Jiangsu Province, for example, the percentage among rural residents (32%) is twice that of urban residents (16%); it is also several times higher for older adults (25% for 55-64 years; 87% for ≥ 65 years) than for young adults (8% for 18-34 years). Further complicating interpretation when there is not a time-based or other relevant comparator, percentages from symptom measures such as the PHQ-9 tend to dramatically overestimate prevalence that would be obtained from validated methods for ascertaining prevalence of mental health disorders, and there is too much heterogeneity between samples in the difference to correct for this statistically.
Summary of Results
Nineteen studies have compared mental health symptoms prior to and during COVID-19; five of the studies were on undergraduate students (7890, 7899, 18242, 19533, 24660), eleven (17066, 21717, 9941, 22825, 24703, 27727, 27776, 30127, 41151, 27707, 75265) were from population-based surveys, two (24680, 52414) surveyed 435 and 131 people respectively, with pre-existing chronic medical conditions from a combination of countries including Canada, the United Kingdom, the United States, France, Denmark, Belgium, and Italy, and one (23486) surveyed 2288 sexual and gender minority people in the United States. The five studies of university students were from Switzerland, China, and the United States (N = 3); they included between 178 and 555 participants. The eleven population-based studies included samples from Brazil, Canada, China, the Czech Republic, Denmark, Ireland, New Zealand, the United Kingdom (N=5), and the United States (N=2); they included between 102 and 36,520 participants.
All studies of undergraduate students compared symptoms from pre-COVID-19 outbreak timepoints to symptoms at post-COVID-19 timepoints. In study 7890, which included 209 undergraduate students from Switzerland, symptoms of depression increased substantially by 0.53 (0.33 to 0.72) standard deviations, stress by 0.40 (0.20 to 0.59), loneliness, slightly, by 0.29 (0.10 to 0.49), and symptoms of anxiety, slightly, by 0.17 (-0.02 to 0.37). A second study (7899), of 178 undergraduate students from the United States reported that there were statistically significant changes in symptoms of both anxiety and depression, but it did not report data that allowed estimation of effect size; regression estimates of changes in depression symptoms, though, were substantially larger than those for anxiety. Another study (19533) from the United States, with 487 undergraduates, found that anxiety symptoms decreased significantly, though by a small amount, from the beginning of the semester and prior to COVID-19 to the end of the semester during COVID-19 (0.17 standard deviations, 95% CI 0.04 to 0.31). A third study (24660) from the United States, with 205 undergraduates, found large and moderate increases in both symptoms of depression (0.48 standard deviations, 95% CI 0.28 to 0.67 ), and symptoms of anxiety (0.35 standard deviations, 95% CI 0.15 to 0.54 ), respectively. In the study (18242) from China, there were small increases in combined anxiety and depressive symptoms and in negative affect.
Among population studies, almost all compared symptoms from pre-COVID-19 outbreak timepoints to symptoms at post-COVID-19 timepoints. In study 17066, which analyses 12,090 responses from a representative sample of British adults (UK Household Longitudinal Study, UKHLS), there was a small increase in general mental health symptoms of 0.31 standard deviations (95% CI 0.29 to 0.34; 1 point on GHQ – dichotomized items). Study 27707, analyses 9,748 responses to the same survey. It reports that there was an increase in the proportion of participants who scored above a mental health function threshold (GHQ-12 ≥ 4). The authors report an increase of 11% scoring above the threshold, from 19% at wave 9 of the UKHLS, to 30% in the COVID-19 wave, though there was not enough information provided to calculate a standardized mean difference. It was not possible to determine, though, the relative changes in anxiety and depressive symptoms. Study 21717, which recruited a small international sample via an online platform (N=218), reports negligible changes in anxiety (-0.11 standard deviations, 95% CI -0.30 to 0.08), depression (0.08 standard deviations, 95% CI -0.11 to 0.27), rumination (-0.07 standard deviations, 95% CI -0.25 to 0.12), and distress (-0.09 standard deviations, 95% CI -0.28 to 0.10). Study 9941 compared mental health symptoms from the beginning of quarantine to symptoms 4 weeks later. Results indicated that among participants in Brazil (N=360), there were small increases in depression (0.20 standard deviations, 95% CI 0.05 to 0.34), anxiety (0.30 standard deviations, 95% CI 0.15 to 0.45), and stress (0.22 standard deviations, 95% CI 0.07 to 0.37). The study did not report how participants were recruited. Study 22825 recruited the 28 year old children of women residing in Avon, England who had been recruited during their pregnancies for the ALSPAC longitudinal study. Compared to participant responses completed at age 26, the COVID-19 data demonstrated negligible changes in continuous depression scores (N= 2219, -0.11 standard deviations, 95% CI -0.06 to -0.15). There were small increases in continuous anxiety scores (N= 1811, 0.26 standard deviations, 95% CI 0.21 to 0.30). Continuous well-being scores showed a large decrease (N=2231, -0.51 standard deviations, 95% CI -0.47 to -0.55). Study 24703, which recruited a moderate sized, nation-wide sample from the United States (N= 2088) via an online recruitment service, reports a negligible change in loneliness (0.02 standard deviations, 95% CI -0.04 to 0.08). Study 27727 recruited 102 Chinese participants from chat groups on social media platforms. Results indicate that from the peak of the COVID-19 outbreak in China, to the trough, there were negligible changes in stress (-0.02 standard deviations, 95% CI -0.30 to 0.25). Study 27776, which compares data following the Christchurch earthquake (2018) to data collected post-COVID-19 outbreak, also reports negligible changes in psychological distress (0.09 standard deviations, 95% CI 0.00 to 0.18). The 940 participants are part of a random, nationally diverse cohort which responds to the longitudinal New Zealand Attitudes and Values Study. Study 30127, commissioned the survey agency Epinion to conduct a two-wave COVID-19 population survey of 2,149 Danish adults. Results comparing symptoms before and after the phase 1 reopening in Denmark indicate a small increase in well-being (0.17 standard deviations, 95% CI 0.11 to 0.23). Study 41151 followed up with 715 participants from a randomly selected, 1% population-based sample of Brno, Czech Republic residents. After the national COVID-19 lockdown, results indicate a large increase in depression (0.44 standard deviations, 95% CI 0.34 to 0.55) and moderate increase in stress (0.29 standard deviations, 95% CI 0.19 to 0.40). Study 75265 compared responses of 36,520 adults in the UCL COVID -19 Social Study, a panel study weighted to population proportions. Data were collected for 20 weeks during the course of the pandemic and latent growth models were fitted, estimating an average of 0.11 weekly point decrease in depression symptoms and 0.10 weekly point decrease in anxiety symptoms over the 20 weeks. Not enough information was provided to calculate standardized mean differences.
Study 24680 surveyed a well-characterized, international cohort of people with the rare disease systemic sclerosis (N= 388-435). Between pre-COVID-19 and post-COVID-19 timepoints, there was a large increase in anxiety (0.51 standard deviations, 95% CI 0.37 to 0.64) and a negligible decrease in depression (-0.05 standard deviations, 95% CI -0.19 to 0.09). Study 52414 surveyed another international cohort of 131 people with progressive multiple sclerosis who were originally recruited for rehabilitation clinical trial. Results indicate that between pre-COVID-19 and post-COVID-19 timepoints, there were negligible changes in anxiety (0.02 standard deviations, 95% CI -0.22 to 0.27) and depression measured by BDI-II (0.09 standard deviations, 95% CI -0.15 to 0.34). The same study found a non-significant increase in depression using the HADS-D measure (0.21 standard deviations, 95% CI -0.03 to 0.46).
In a well-characterized cohort of sexual and gender minority people in the United States (N=2,288), study 23486 reported a small increase in depression (0.19 standard deviations, 95% CI 0.14 to 0.25) and a large increase in anxiety (0.54 standard deviations, 95% CI 0.48 to 0.60) following the COVID-19 outbreak.
Among university students for whom social relationships are likely highly valued and for whom risk of complications from COVID-19 is generally lower than among other adults, symptoms of depression increased more than anxiety in three studies where that was reported; in one study, anxiety symptoms actually decreased among university students. The study on university students in Switzerland also reports moderate increases in stress and smaller increases in loneliness. The large population study in the UK (17066) reported general mental health symptoms and a small increase, but this did not allow interpretation of the types of symptoms experienced. The additional population studies generally show either small increases or negligible changes in anxiety, depression, and other mental health functions. A significant decrease in well-being measured in Avon, UK and significant increase in depression, measured in Brno, Czech Republic, were the largest changes reported for population studies. Emerging evidence on vulnerable populations suggests anxiety may be more important. The study in the rare disease cohort showed a large increase in anxiety while there were negligible changes in depression. The study on sexual and gender minority people indicated a similarly large increase in anxiety and a small increase in depression. It will be important to understand to what degree these finding are replicated in other university, vulnerable, and general populations.
1. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med. 2001;16(9):606-613.
2. Yu X, Tam WWS, Wong PTK, Lam TH, Stewart SM. The Patient Health Questionnaire-9 for measuring depressive symptoms among the general population in Hong Kong. Compr Psychiatry. 2012;53:95-102.
3. Kocalevent RD, Hinz A, Brähler E. Standardization of the depression screener Patient Health Questionnaire (PHQ-9) in the general population. Gen Hosp Psychiatry. 2013;35:551-555.
4. Wang W, Bian Q, Zhao Y, et al. Reliability and validity of the Chinese version of the Patient Health Questionnaire (PHQ-9) in the general population. Gen Hosp Psychiatry. 2014;36:539-544.
5. Cao C, Hu L, Xu T, et al. Prevalence, correlates, and misperception of depression symptoms in the United States, NHANES 2015-2018. J Affect Disord. 2020;269:51-57.
6. Patten SB, Schopflocher D. Longitudinal epidemiology of major depression as assessed by the Brief Patient Health Questionnaire (PHQ-9). Compr Psychiatry. 2009;50:26-33.
7. Johansson R, Carlbring P, Heedman A, Paxling B, Andersson G. Depression, anxiety and their comorbidity in the Swedish general population: point prevalence and the effect on health-related quality of life. PeerJ. 2013;1:e98.
8. Lu S, Reavley N, Zhou J, et al. Depression among the general adult population in Jiangsu Province of China: prevalence, associated factors and impacts. Soc Psychiatry Psychiatr Epidemiol. 2018;53:1051-1061.
9. Levis B, Benedetti A, Ioannidis JPA, et al. Patient Health Questionnaire-9 scores do not accurately estimate depression prevalence: an individual participant data meta-analysis. J Clin Epidemiol. 2020;122:115-128.
*Abbreviations: BDI= Beck’s Depression Inventory; CES-D= Center for Epidemiologic Studies Depression Scale; DASS-21= Depression, Anxiety, and Stress Scale; DToS= Distress Tolerance Scale; FDI= Filgueiras Depression Inventory; GAD-7= Generalized Anxiety Disorder; GHQ= General Health Questionnaire; HADS-A= Hospital Anxiety and Depression Scale- Anxiety; HADS-D= Hospital Anxiety and Depression Scale- Depression; K6= Kessler Psychological Distress Scale; PANAS= Positive and Negative Affect Schedule; PHQ= Patient Health Questionnaire; PROMIS= Patient Reported Outcomes Measurement Information System; PSS= Perceived Stress Scale; RRQ= Reflection and Rumination Scale; SMFQ= Short Mood and Feelings Questionnaire; STAI= State Trait Anxiety Inventory; SWEMWBS= Warwick Edinburgh Mental Wellbeing Scale; ULS-9= University of California, Los Angeles (UCLA) Loneliness Scale; WHO-5= WHO (Five) Wellbeing Index; YUSS= Yang Uncertainty Stress Scale
*A positive effect size indicates an increase in the construct assessed