Discussion
To address child and adolescent mental health during COVID-19, we conducted an overview of 18 systematic reviews and meta-analysis of eligible studies from 366 unique primary studies. Reviews were included from any setting, although the majority included primary studies conducted in China (48.8% of included studies from quantitative syntheses). Most reviews focused on the prevalence of depression and anxiety across a range of child (5–9.9 years) and adolescent (10–19.9 years) populations. This unfortunately made it impossible to disaggregate and analyse data by younger versus older age groups. Other mental health disorders including post-traumatic stress disorder, sleep disturbances (i.e., insomnia), substance, eating and addictive disorders were less often explored. We rated the quality of the 18 included systematic reviews using the AMSTAR 2 tool and found 61.1% of reviews to be critically low quality and 38.9% to be low quality. Our rating of the quality of included reviews as low or critically low underscores the urgent need to invest in improving mental health measurement tools and systematic data collection for children and adolescents across varying contexts.
To ensure comprehensiveness, we reviewed all primary studies within included systematic reviews and calculated the CCA to avoid duplication of data, which was found to be only a slight overlap of 4.0%. Furthermore, to limit confounding that may arise from including children and adolescents with previous mental health or chronic conditions and to address the need to understand the mental health impacts for those without pre-existing conditions, we re-analysed primary data from eligible studies with strict exclusion criteria. From our re-analysis of primary data, the overall PP of depression was 32% (95% CI: 27 to 38) (n=161 673), while the PP of anxiety was 32% (95% CI: 27 to 37) (n=143 928) in children and adolescents globally without disclosed prior mental health conditions, or positive for SARS-CoV-2. In comparison, our reported prevalence for depression and anxiety during the pandemic is higher than prepandemic estimates, where WHO estimated 3.6% of adolescents aged 10–14 years and 4.6% of those aged 15–19 years experienced an anxiety disorder prior to the pandemic, globally.36 Depression was estimated to occur among 1.1% of adolescents aged 10–14 years and 2.8% in those aged 15–19 years.36 These findings add to the extant literature on the detrimental effects of COVID-19, the spread of the virus, lockdown stay-at-home orders, school closures and decreased social interactions, but are likely an underestimation. The global underestimation of mental health conditions is recognised widely in the literature and has been hypothesised to stem from sociocultural factors (i.e., stigma), access and affordability of care and current measurement approaches (i.e., overlap between psychiatric and neurological disorders, conflation of chronic pain, exclusion of personality disorders).37 Furthermore, discrepant reporting across the world, with fewer studies from LMICs, may contribute to a lack of representative information from the countries where mental health issues are underestimated. Given the heterogeneity of included studies, instruments used and populations samples, results should be interpreted with caution. However, prevalence data still provides critical population-level information for service planning.
Furthermore, our review found subgroup differences for both depression and anxiety by WHO region in children and adolescents without prior conditions, but not by country income level or by gender. However, subgroup analyses should be interpreted with caution, given the low number of studies contributing to some subgroups (i.e., only three studies contributed to the Eastern Mediterranean region subgroup analysis compared with 30 studies for the Western Pacific). In comparison, a study by the COVID-19 Mental Disorders Collaborators reported increased prevalence of depressive and anxiety disorders due to the pandemic, where the PP of anxiety (B: –0.003, 95% uncertainty level (UL): –0.005 to –0.0002; p=0.0001) and major depressive disorder (B: −0.007, 95% UL: 0.009 to –0.0006; p=0.0001) was greater in younger age groups as compared with older age groups.8 Their findings also suggest that females were affected more by the pandemic than males (B: 0.1, 95% UL: 0.1 to 0.2; p=0.0001) for major depressive disorder and (B: 0.1, 95% UL: 0.1 to 0.2; p=0.0001) for anxiety disorders, which is hypothesised to be a result of compounding social and economic inequities that females face, which have been further exacerbated by the pandemic.8 The report by the COVID-19 Mental Disorders Collaborators also included ages into adulthood, so differences in our results may be due to increased inequities that females face later in life, compared with in childhood and adolescence. Systematic reviews by Racine et al,19 Ma et al21 and Chai et al34 all reported differences in the prevalence of anxiety and/or depression by female sex. It is likely that our findings did not reach statistical significance as our sample was limited to studies that reported disaggregated prevalence of depression and/or anxiety, resulting in a smaller sample size, and thus results may not have been powered to detect a difference.
The severity of COVID-19, as well as economic status, and healthcare systems, vary greatly by country.38 39 Despite this, we did not find a difference in the prevalence of depression and/or anxiety between country income level. For example, large discrepancies exist between different contexts and access to mental health infrastructures, where some countries may have well-developed programming available, and others have limited and underfunded systems of care. Furthermore, the socioeconomic consequences of the pandemic have led to significant job losses globally.40 This is particularly problematic in LMICs where individuals are informally employed as daily wagers and weekly incomes are used to support family feeding.20 With poverty rates expected to soar due to the pandemic,41 this will further deteriorate the mental health of children and adolescents in LMICs. Our findings highlight a significant difference between anxiety and depression prevalence by WHO region. We report a significantly higher prevalence for anxiety and depression in the Eastern Mediterranean, compared with the Americas, Europe, South-East Asia and the Western Pacific. In addition to few studies contributing to these meta-analyses, the high prevalence reported for the Eastern Mediterranean, may represent an overlap with other factors such as conflict and displacement. There are inconsistent findings across the systematic reviews for subgroup analyses by region. For example, Racine et al19 and Brussières et al33 reported higher mental health problems in European countries compared with Asian countries. In contrast, Panda et al28 reported that most studies from Asia showed a higher prevalence of psychological morbidities, compared with HICs in Europe. The inconsistency in these results may be due to differences in when each primary study was conducted. Regardless of country region, studies conducted early in the pandemic may have higher self-reported mental health symptoms due to the uncertainty and fear experienced earlier on in the pandemic.
Experiences from past pandemics and crises indicate that population shocks (i.e., unexpected or unpredictable events that disrupt the environmental, health, economic or social circumstances within a population) can heighten stress responses. For children and adolescents, these shocks could have lifelong implications. For example, evidence from the Ebola epidemic of 2014–2016 in West Africa highlighted the prevailing fear and stigma of Ebola undermined the willingness of community members to help orphaned children, which ultimately had severe psychological repercussions for children orphaned by Ebola.42 Likewise, Maclean et al suggest that experiencing a natural disaster by age 5 has been shown to significantly increase the risk of mental health, particularly anxiety disorders in adults.43 This illustrates that the worsening of mental health seen in children and adolescents during the pandemic cannot be attributed to the indirect impacts alone, but is also related to the direct effects of the pandemic, such as fear of illness, family/caregiver death, loss of income or food insecurity. In addition to the direct experience of a crisis, a negative response to shocks is also likely due to changing parenting styles. Parenting styles during crises demonstrate increases in maltreatment behaviours and poor emotional support.44 45 Of note however, protective factors, (i.e., strengthening internal resources for coping, strong caregiver support, community resources) can help facilitate an adolescent’s ability to overcome difficult shocks or transitions. This highlights the importance of identifying families with higher vulnerability that are unable to cope with new stressors, as well as ensuring parent support interventions are universally available at policy and service levels.4 Furthermore, the consequences of poor mental health in young populations experienced during a crisis can also lead to emotional reactivity and engagement in health risk behaviours such as increased sedentariness and screen time, cyberbullying, substance abuse and self-harm.4 46 This highlights the need to support children and adolescents in developing both internal and external resources for more adaptive ways of coping.
Limitations
Our review is limited by the quality of the included studies, where the included reviews were rated as very low or low quality, underscoring a critical limitation. First, there was substantial heterogeneity between included studies in measurement and reporting of outcomes. Although many of the tools were validated, some studies developed their own questionnaires to measure outcomes of mental health which could lead to bias and inconsistent results. It should also be noted that although these tools allow for the quantification of symptoms, none provide a clinical diagnosis, thus estimates may be underestimated. Second, we relied on study disclosure of prior mental health conditions in children and adolescents, however many children with these conditions are underdiagnosed, thus it is likely that children with prior mental health conditions have been included in this sample. Third, as mitigation strategies were often bundled and implemented nationally or subnationally, it is challenging to disentangle individual strategies that may have led to worse mental health effects. Furthermore, evidence from the included studies did not allow us to disentangle the direct effects of the pandemic (i.e., fear of illness) on children’s and adolescent’s mental health. Much more refined research is necessary to determine what precisely about the pandemic is contributing to crises in children’s mental health in order to inform policies and interventions. Lastly, as most studies were cross-sectional in nature, compared with few studies that were longitudinal, differences in timing of measurement (i.e., beginning of the pandemic vs multiple years into the pandemic) could lead to inconsistent results. Almost all the included studies used surveys through online self-report, which increases the chance of bias and inaccurate reporting from subjectivity of caregiver perceptions about their child’s mental health state.