Introduction

Pandemics and other large-scale emergencies have the potential to negatively affect mental health during the event and long after. In response to the COVID-19 pandemic, mitigation measures have interrupted in-person learning, social and community networks, recreational activities and access to health care, challenging access to important routines, social structures, resources, and supports. While these measures are necessary to prevent an escalating public health emergency, prolonged social isolation and home confinement may lead to immediate and long-term mental health and well-being challenges (Kaufman, Petkova, Bhui, & Schulze, 2020; Wang, Zhang, Zhao, Zhang, & Jiang, 2020). Short-term factors contributing to mental distress during the COVID-19 pandemic include concerns about SARS-CoV-2 infection and subsequent health impacts, social isolation, and worsening social determinants of mental health such as socioeconomic stressors resulting in stress and increased mental illness (Aknin et al., 2021). For instance, a longitudinal survey in the United Kingdom of over 50,000 individuals found that the proportion reporting clinically significant levels of mental distress rose from 19% in 2018 to 27% in April 2020, one month into the COVID-19 lockdown (Pierce et al., 2020).

Mental health impacts of disasters, such as depression, anxiety, post-traumatic stress disorder, substance use disorder, as well as domestic violence and child abuse, have been identified in settings such as the aftermath of the SARS epidemic, 9–11, Hurricane Katrina, and other humanitarian emergencies (Furr, Comer, Edmunds, & Kendall, 2010; Galea, Merchant, & Lurie, 2020; Purgato et al., 2019; Sprang & Silman, 2013; B. Tang, Liu, Liu, Xue, & Zhang, 2014; Tang, Deng, Glik, Dong, & Zhang, 2017; Tol et al., 2011). The cumulative effect of multiple risk factors and inadequate protective factors can reduce mental well-being and increase vulnerability to mental illness emergence (World Health Organization, 2014). Individuals at risk may experience new onset of mental illness, while those with pre-existing mental health conditions may experience symptomatic worsening, especially if mental health service access is impeded (Moreno et al., 2020). The unprecedented reach of COVID-19 pandemic impacts necessitates urgent population-level monitoring of mental health to optimize efforts for prevention and mitigation of its effects.

Pre-existing vulnerabilities, such as socioeconomic disadvantage, neurodiverse needs, or disability may increase risk of poor mental health outcomes during the COVID-19 pandemic. Organizational closures and physical distancing requirements have reduced social contact and support, compromising food security in some cases and secondary oversight of child and adolescent emotional and physical safety (Poole, Fleischhacker, & Bleich, 2021; Salt et al., 2021; Swedo et al., 2020). Students who rely on special education, lack digital access or tools, or live in unstable home settings risk falling behind their (peers) as schools move online. In the rapidly evolving context of the pandemic as well as literature and evidence-generation elucidating mental health impacts of the pandemic on children and adolescents, we seek to build on reviews capturing early pandemic impacts (Araújo, Veloso, Souza, Azevedo, & Tarro, 2020; Meherali et al., 2021; Nearchou, Flinn, Niland, Subramaniam, & Hennessy, 2020) and report on child and adolescent mental health impacts one year into the COVID-19 pandemic. Our systematic review aims to summarize population-level impacts of the COVID-19 pandemic on global child and adolescent mental health as captured in the year following its onset, contextual factors influencing impacts, as well as to identify protective factors that may mitigate these impacts.

Methods Search strategy

In this review, we searched for peer-reviewed and preprint articles describing COVID-19-related mental health changes among children and adolescents (<19 years of age) made available in English from January 1, 2020, to February 22, 2021. We registered a protocol for this review (https://doi.org/10.17605/OSF.IO/B94AC) with the Open Science Framework (Snell & Samji, 2021). We searched nine electronic databases: MEDLINE, PsycINFO, Scopus, PubMed, EMBASE, Web of Science, medRxiv, PsyArxiv, and Cumulative Index of Nursing and Allied Health Literature (CINAHL). We used a predefined search strategy (Table S1) to extract articles from MEDLINE, PsycINFO, PubMed, and EMBASE. This strategy did not use available database fields to filter results by language. We modified these keywords and patterns as necessary to identify articles using search engines offered by CINAHL, Web of Science, medRxiv, PsyArxiv, and Scopus.

Selection criteria

We included studies which assessed stratified mental health outcomes for young people 0-18 years of age related to the COVID-19 pandemic. We defined mental health outcomes as broadly as possible, including studies which measured changes in the prevalence or symptoms of mental illness, overall mental health or well-being, mental health service utilization, and other emotional or behavioral characteristics. Mental health outcomes could be (a) quantitatively or qualitatively measured, and/or (b) derived from child and adolescent self-report or from caregivers, caretakers, teachers, or other adults reporting on the mental health of young people they supervise. Gray literature and non-English language articles were not eligible for inclusion and were thus removed during screening. Studies focusing exclusively on the direct or indirect effects of COVID-19 on physical health or on mental health service adaptation in response to the pandemic were excluded. Controlled trials exclusively assessing mental health interventions and secondary analyses of primary data such as reviews, editorials, and letters were also excluded. Figure 1 describes the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart of our literature search.

PRISMA flow diagram describing study selection process

Article screening and extraction

We used Covidence systematic review software (https://www.covidence.org/) to remove duplicate entries and coordinate article screening. In a two-stage process, a team of five coders first reviewed article abstracts and then full-text manuscripts for eligibility. Two coders completed a data extraction template for each article screened for inclusion. This template included fields for study location, period, design, target population and sample size, measurement tools, and mental health outcomes. When possible, coders extracted relevant outputs of quantitative analyses, including any reported effect sizes, descriptions of measurement uncertainty, and results of significance tests. Coders also assigned keywords to each study to identify common factors and outcomes, which included protective and resilience factors. Resilience was conceptualized as a process, rather than a trait or outcome, and defined as an individual’s ability to cope with a crisis mentally or emotionally; to move forward in a positive manner despite adversity; or to adapt successfully to a pandemic that threatens their viability, function, or development (de Terte & Stephens, 2014; Southwick et al., 2014). We assessed evidence quality using a rating scheme modified from the Oxford Centre for Evidence-Based Medicine ranging from one to five, with lower scores denoting higher-quality evidence (Centre for Evidence-Based Medicine, n.d.). We graded qualitative studies at ‘4’ on this scale to denote that they collected cross-sectional data. We descriptively summarized results of data extraction considering high heterogeneity in study designs, populations, mental health outcomes, and measurement tools represented in included articles.

Results Study design and quality

A total of 116 articles representing more than 127,923 children and adolescents met inclusion criteria and were included in the systematic review. Using the Oxford quality scoring system, four articles were given a rating of two (prospective cohort design) (Gassman-Pines, Ananat, & Fitz-Henley, 2020; Munasinghe et al., 2020; Shek, Zhao, Dou, Zhu, & Xiao, 2021; Xiang, Yamamoto, & Mizoue, 2020) and 24 were given a rating of three (case-control, retrospective cohort, and chart review designs) (Abawi et al., 2020; Ademhan Tural et al., 2020; Alonso-Martínez, Ramírez-Vélez, García-Alonso, Izquierdo, & García-Hermoso, 2021; Amorim et al., 2020; Bothara et al., 2021; Breaux et al., 2021; Chahal, Kirshenbaum, Miller, Ho, & Gotlib, 2021; Cheek, Craig, West, Lewena, & Hiscock, 2020; Chen, Chen, et al., 2020; Chen, She, et al., 2020; Conti et al., 2020; Diaz de Neira et al., 2020; Ezpeleta, Navarro, de la Osa, Trepat, & Penelo, 2020; Ferrando et al., 2020; Gotlib et al., 2020; Janssen et al., 2020; Jefsen, Rohde, Nørremark, & Østergaard, 2020; Leeb et al., 2020; Leff, Setzer, Cicero, & Auerbach, 2021; Magson et al., 2021; Rogers, Ha, & Ockey, 2021; Tanaka & Okamoto, 2021; Tromans et al., 2020; Zhang, Zhang, et al., 2020). The remaining 88 articles were given a rating of four, indicating cross-sectional or qualitative study design. A variety of standardized tools were used across studies to assess mental health, psychological and psychiatric diagnostic outcomes. The most commonly used tools included the Generalized Anxiety Disorder Scale (GAD-7) (15/116), the Strengths and Difficulties Questionnaire (SDQ) for mental well-being (14/116), and the Patient Health Questionnaire (PHQ-9) (13/116), and the Centre for Epidemiologic Studies Depression Scale (CES-D) for depressive disorder (7/116).

Study populations

The 116 articles presented data on a total of 127,923 children, and adolescents; 50,984 child and adolescent proxy reports (i.e., parents, guardians, healthcare practitioners); and over 3,000 charts were reviewed across seven articles (although specific chart numbers reviewed were not reported in many studies frequently many did not specify the number of charts reviewed). The majority of studies described general population findings while a third of studies focused on specific population subgroups including those with neurodiverse conditions such as autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD), and obsessive-compulsive disorder (OCD) (Amorim et al., 2020; Asbury, Fox, Deniz, Code, & Toseeb, 2020; Breaux et al., 2021; Colizzi et al., 2020; Conte, Baglioni, Valente, Chiarotti, & Cardona, 2020; Conti et al., 2020; Evans et al., 2020; Graziola et al., 2020; Jefsen et al., 2020; Masi et al., 2021; Nonweiler, Rattray, Baulcomb, Happe, & Absoud, 2020; O’Sullivan et al., 2021; Patra, Patro, & Acharya, 2020; Paulauskaite et al., 2021; Sciberras et al., 2020; Secer & Ulas, 2020; Storch et al., 2021; Theis, Campbell, De Leeuw, Owen, & Schenke, 2021; Zhang, Shuai, et al., 2020). Other studies focused on specific populations such as children and youth with inflammatory bowel syndrome (IBS), chronic respiratory conditions, and neuromuscular disorders (Alshahrani et al., 2020), and LGBTQ-identifying adolescents (Fish et al., 2020).

Eight studies contained results on child and adolescent populations accessing psychiatric services during the pandemic (Bothara et al., 2021; Cheek et al., 2020; Chen, She, et al., 2020; Diaz de Neira et al., 2020; Ferrando et al., 2020; Leeb et al., 2020; Leff et al., 2021; Tromans et al., 2020).

All of the data collection in the studies included in this review was undertaken between January and November 2020. By key phases in the progression of the pandemic, studies collected COVID-19-related data from January to February (20/116), March to May (91/116), June to August (25/116), and September to November (2/116). The greatest number of studies were conducted in Europe (39/116), followed in diminishing order by East Asia (28/116), North America (21/116), South Asia (7/116), Australasia (7/116), West Asia (6/116), South America (2/116), South-East Asia (2/116), Sub-Saharan Africa (1/116), and North Africa (1/116). Two studies involved multiple countries or were international in focus. Study period, sample size, gender distribution, population, mental health-related assessment tool, and location for each included study are further detailed in Table S2.

Age distribution

The majority of articles focused entirely on child and adolescent populations ≤18 years of age. However, 24 studies also included populations ≥18 years (Alamrawy, Fadl, & Khaled, 2021; Banati, Jones, & Youssef, 2020; Buzzi et al., 2020; Chahal et al., 2021; Chen, She, et al., 2020; Commodari & La Rosa, 2020; Conte et al., 2020; de Matos et al., 2020; Esposito et al., 2020; Ferrando et al., 2020; Fish et al., 2020; Fitzpatrick, Carson, & Weisz, 2020; Janssen et al., 2020; Matovu, Kabwama, Ssekamatte, Ssenkusu, & Wanyenze, 2021; Munasinghe et al., 2020; Murata et al., 2021; Qi, Liu, et al., 2020; Rauschenberg et al., 2020; Scott et al., 2021; Shek et al., 2021; Storch et al., 2021; Tanaka & Okamoto, 2021; Tromans et al., 2020; Zhou, Wang, et al., 2020). Thirteen studies targeted primary school populations (<12 years) (Alonso-Martínez et al., 2021; Chen, Chen, et al., 2020; de Avila et al., 2020; Dumas, Ellis, & Litt, 2020; Fontenelle-Tereshchuk, 2020; A. Gassman-Pines et al., 2020; Glynn, Davis, Luby, Baram, & Sandman, 2021; Lee, Ward, Chang, & Downing, 2021; Liu, Liu, & Liu, 2020; Romero, López-Romero, Domínguez-álvarez, Villar, & Gómez-Fraguela, 2020; Waller et al., 2021; Xie et al., 2020; Xue et al., 2021), 27 reported on middle and high school populations (aged 12–18) (Alamrawy et al., 2021; Breaux et al., 2021; Buzzi et al., 2020; Cao et al., 2021; Chi et al., 2021; Ellis, Dumas, & Forbes, 2020; Fish et al., 2020; Giannopoulou, Efstathiou, Triantafyllou, Korkoliakou, & Douzenis, 2021; Janssen et al., 2020; Kılınçel, Kılınçel, Muratdağı, Aydın, & Usta, 2020; Li et al., 2021; Liebana-Presa et al., 2020; Lu et al., 2020; Luthar, Ebbert, & Kumar, 2020; Magson et al., 2021; Masuyama, Shinkawa, & Kubo, 2020; Murata et al., 2021; Oosterhoff, Palmer, Wilson, & Shook, 2020; Pons et al., 2020; Qi, Liu, et al., 2020; Qi, Zhou, et al., 2020; Rauschenberg et al., 2020; Rogers et al., 2021; Scott et al., 2021; Secer & Ulas, 2020; Zhang, Ye, et al., 2020; Zhou, Yuan, et al., 2020; Zhou, Wang, et al., 2020), while 16 studies did not report the range or grades of their child and adolescent population (Abawi et al., 2020; Alivernini et al., 2021; Amorim et al., 2020; Amran, 2020; Colizzi et al., 2020; Ezpeleta et al., 2020; Graziola et al., 2020; O’Sullivan et al., 2021; Patra et al., 2020; Paulauskaite et al., 2021; Sama et al., 2021; Storch et al., 2021; Theis et al., 2021; Tromans et al., 2020; Yue, Zang, Le, & An, 2020; Zorcec, Jakovska, Micevska, Boskovska, & Cholakovska, 2020). The remaining 60 articles targeted a combination of children and adolescents in primary, middle, and/or high school.

Factors and outcomes

The following factors and outcomes emerged in our review and were used to categorize findings: mental health-related outcomes (depressive symptoms; self-harm, suicidal ideation, and suicide; anxiety symptoms; COVID-19-related fear, concern, and worry; general mental health; mental health service utilization; other mental health outcomes); demographic factors (age, gender); contextual factors (pandemic control measures; social connection; family relationships; technology use and media consumption; and population factors (neurodiverse children and adolescents; chronic physical conditions; additional resilience and protective factors)). We used frequency of reporting and strength of findings to organize the mental health-related outcomes section.

Mental health-related outcomes

COVID-19-related fear, concern, and worry. Thirty-five articles discussed COVID-19-related fear, concern, or stress in children and adolescents (Abawi et al., 2020; Abdulah, Abdulla, & Liamputtong, 2020; Adibelli & Sumen, 2020; Al Omari et al., 2020; Alshahrani et al., 2020; Anbarasu & Bhuvaneswari, 2020; Asbury et al., 2020; Banati et al., 2020; Buzzi et al., 2020; Commodari & La Rosa, 2020; Dumas et al., 2020; Ellis et al., 2020; Evans et al., 2020; Guo et al., 2020; Idoiaga, Berasategi, Eiguren, & Picaza, 2020; Jefsen et al., 2020; Jiao et al., 2020; Kılınçel et al., 2020; Liebana-Presa et al., 2020; Lu et al., 2020; Masuyama et al., 2020; Matovu et al., 2021; McElroy et al., 2020; O’Sullivan et al., 2021; Qi, Liu, et al., 2020; Rauschenberg et al., 2020; Saurabh & Ranjan, 2020; Sciberras et al., 2020; Scott et al., 2021; Secer & Ulas, 2020; Shah, Kaul, Shah, & Maddipoti, 2021; Xie et al., 2020; Xue et al., 2021; Zhang, Zhang, et al., 2020; Zhou, Yuan, et al., 2020). The most reported COVID-19-related fear was fear of infection of either themselves or, more commonly, fear of infecting a vulnerable loved one (Abawi et al., 2020; Abdulah et al., 2020; Adibelli & Sumen, 2020; Alshahrani et al., 2020; Anbarasu & Bhuvaneswari, 2020; Asbury et al., 2020; Banati et al., 2020; Buzzi et al., 2020; Commodari & La Rosa, 2020; Dumas et al., 2020; Ellis et al., 2020; Evans et al., 2020; Guo et al., 2020; Idoiaga et al., 2020; Jefsen et al., 2020; Jiao et al., 2020; Kılınçel et al., 2020; Liebana-Presa et al., 2020; Lu et al., 2020; Masuyama et al., 2020; Matovu et al., 2021; McElroy et al., 2020; O’Sullivan et al., 2021; Qi, Zhou, et al., 2020; Rauschenberg et al., 2020; Saurabh & Ranjan, 2020; Sciberras et al., 2020; Scott et al., 2021; Secer & Ulas, 2020; Shah et al., 2021; Xie et al., 2020; Xue et al., 2021; Zhang, Zhang, et al., 2020; Zhou, Yuan, et al., 2020). An Indian study noted that 23% of children kept a daily count of deaths and hospitalizations from COVID-19 (Shah et al., 2021).

Other reported reasons for fear and concern during the COVID-19 pandemic included fear of not being able to cope with academic workload and worries about the impact of COVID-19 on the school year and future plans (Anbarasu & Bhuvaneswari, 2020; Ellis et al., 2020; Kılınçel et al., 2020; O’Sullivan et al., 2021; Scott et al., 2021). For example, a study in Canada showed that 73% of adolescents surveyed in April 2020 were ‘very concerned’ about the impact of COVID-19 on their school year (Ellis et al., 2020). Young people in two studies also reported fear of becoming socially isolated from their friends (Kılınçel et al., 2020; Scott et al., 2021).

Increased feelings of fear and concern regarding COVID-19, measured via the ‘Fear of COVID-19 Scale’ and other COVID-19-specific questionnaires, were found to be cross-sectionally associated with higher levels of depressive and anxious symptoms, measured via separate questionnaires used to screen for those disorders (Chi et al., 2021; Lu et al., 2020), post-traumatic stress (Guo et al., 2020), lower emotional well-being (Adibelli & Sumen, 2020; Commodari & La Rosa, 2020), and insomnia (Chi et al., 2021). COVID-19-related cognitive preoccupation, worries, and anxiety were cross-sectionally associated with current use of mental health-related apps (Rauschenberg et al., 2020). In contrast, Qi, Liu, et al., (2020) observed that adolescents who were more concerned about COVID-19 experienced fewer anxious symptoms. The authors speculate that this relationship may be explained by a positive correlation between concern and knowledge of COVID-19, a factor which could protect against more severe fear and anxiety regarding the pandemic. In line with this hypothesis, Zhou, Zhang, et al., (2020) observed that those with more knowledge of the local COVID-19 epidemic and associated control measures reported fewer anxious symptoms.

Depressive symptoms

A majority of studies measured depressive symptoms as a mental health outcome. Of these, twenty-five articles identified a higher prevalence of depressive symptoms in children and adolescents during the pandemic compared to the prepandemic period (Abdulah et al., 2020; Breaux et al., 2021; Chen, Chen, et al., 2020; Crescentini et al., 2020; Duan et al., 2020; Ellis et al., 2020; Giannopoulou et al., 2021; Glynn et al., 2021; Gotlib et al., 2020; Janssen et al., 2020; Lee et al., 2021; Lu et al., 2020; Magson et al., 2021; Matovu et al., 2021; Oosterhoff et al., 2020; O’Sullivan et al., 2021; Rogers et al., 2021; Sama et al., 2021; Scott et al., 2021; Secer & Ulas, 2020; Xie et al., 2020; Zhang, Ye, et al., 2020; Zhang, Zhang, et al., 2020; Zhou, Yuan, et al., 2020; Zhou, Zhang, et al., 2020). In contrast, a small minority of studies detected a decrease in depressive symptoms (Ezpeleta et al., 2020; Tso et al., 2020; Xiang, Zhang, & Kuwahara, 2020). Ezpeleta et al., (2020) suggested that differences between parent and child self-report may have accounted for the decrease they observed among adolescents in Barcelona. On the other hand, Xiang, Zhang, et al., (2020) proposed that high ownership of digital devices, rapid implementation of remote learning, and a reduction in stresses experienced by some students at school may explain the decline in depressive symptoms they noted in a sample of 6- to 17-year-old students in Shanghai. Studies conducted at the height of the epidemic in China (January–May 2020) reported depression prevalence estimates ranging from 11% to 45% (Duan et al., 2020; Lu et al., 2020; Murata et al., 2021; Sama et al., 2021; Xie et al., 2020; Zhou, Zhang, et al., 2020), with one study of adolescents after extended lockdown reporting a prevalence of 64% (Zhou, Zhang, et al., 2020), compared with prepandemic estimates in similar populations ranging from 13% to 17% (Stewart & Sun, 2007; Xu et al., 2020).

Anxiety symptoms

The majority of studies measured anxiety symptoms in young people; among these, 17 found increased levels of anxiety symptoms in comparison with prepandemic estimates (Amorim et al., 2020; Asbury et al., 2020; Breaux et al., 2021; Chen, Chen, et al., 2020; Conti et al., 2020; Duan et al., 2020; Giannopoulou et al., 2021; Lee et al., 2021; Lu et al., 2020; Magson et al., 2021; Meherali et al., 2021; O’Sullivan et al., 2021; Ravens-Sieberer et al.,