Anxiety symptoms and disorders during the COVID-19 pandemic in children and adolescents: Systematic review and meta-analysis

Background and Objectives: The COVID-19 pandemic and its associated factors have been shown to affect anxiety levels of young people. We meta-analytically assessed the prevalence of anxiety symptoms and anxiety disorders in children and adolescents during the pandemic, and the predictors and moderating factors in ﬂ uencing anxiety. Methods: Multiple databases and registers were searched in this PRISMA and MOOSE-compliant systematic review and meta-analysis (PROSPERO:CRD42021266695) until 27/06/2021. We included individual studies evaluating the prevalence and characteristics of anxiety symptoms or anxiety disorders in children and adolescents (mean age ≤ 18 years), during the COVID-19 pandemic. Data extraction and quality assessment were carried out by independent authors. Random-effects meta-analyses of the prevalence of anxiety symptoms and anxiety disorders were conducted using Comprehensive Meta-Analysis (CMA) V3. Results: 74 articles (total participant sample=478,882) were included (mean age=13.4 years, 52.3% female). The pooled rate of children and adolescents ful ﬁ lling diagnostic criteria for anxiety disorders was 13.0% (95%CI=4.9 − 30.1); the pooled prevalence of anxiety symptoms was 26.5% (95%CI=20.3 − 33.9). Anxiety symptoms were signi ﬁ cantly more prevalent in females than males ( B = 0.103, p < .001), signi ﬁ cantly higher during the second wave of COVID-19, following July 2020, than during the ﬁ rst wave, prior to June 2020, ( Q = 8.136, p =.017), and during school closure ( Q = 8.100, p =.014). Quality of included studies was overall moderate. Conclusions: There is a high prevalence of anxiety symptoms in children and adolescents during the COVID-19 pandemic, especially amongst females. This study identi ﬁ es vulnerable groups, risk, and protective factors, which is crucial to developing clinical practice to prevent further mental health deterioration in young people.


Introduction
COVID-19 was declared a global pandemic on the 11th of March 2020 by the World Health Organization (WHO). 1 As of the 13th of April 2023, a total of 762,739,900 COVID-19 cases have been confirmed worldwide, with 6,896,778 deaths. 2 The COVID-19 pandemic is having detrimental effects on public mental health, 3 with a significant impact on mental health outcomes in children and adolescents, [4][5][6] especially older adolescents, girls, and young people with neurodiversities and/or chronic physical conditions. 7Anxiety disorders and anxiety symptoms account for the most prevalent psychiatric problems in those aged 6 to 18 years old, with a worldwide prevalence of 6.5%. 8Anxiety symptoms seem to have exacerbated in the general population during the COVID-19 pandemic. 3 previous meta-analysis assessed the prevalence of anxiety amongst medical students during COVID-19, 9 which was shown to significantly increase compared to pre-pandemic levels.Meta-analyses on the prevalence of anxiety amongst the general population 3 and college students 10 have also been conducted, showing female sex as the most significant risk factor.A narrative review and a systematic review on the psychological impact of COVID-19 thus far have shown a high prevalence of anxiety, amongst children 4 and adolescents, 5 respectively; but no meta-analyses were conducted.A recent meta-analysis of 29 studies assessed the prevalence of anxiety symptoms and depressive symptoms in children and adolescents, finding a 20.5% prevalence of clinically elevated anxiety, highlighting increased anxiety levels, and evaluated three moderating factors: age, sex, and pandemic data collection time point. 6Further to this, a recent metaanalysis of children and adolescents from Turkey and China found a 26% prevalence of anxiety, with adolescents and females reporting a higher prevalence of anxiety than children and males. 11his systematic review and meta-analysis evaluates and comprehensively characterizes anxiety symptoms and anxiety disorders in children and adolescents during the COVID-19 pandemic, evaluating a broad range of moderating factors.This review aims to meta-analyse the prevalence of anxiety symptoms and the proportion of anxiety disorder emergence present amongst children and adolescents during the COVID-19 pandemic.This systematic review also aims to provide comprehensive evidence about the influence of predictors and moderating factors on the anxiety symptoms within children and adolescents.

Methods
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (eTable 1 & eTable 2) and Meta-analyses Of Observational Studies in Epidemiology (MOOSE) guidelines 12 (eTable 3).We pre-registered the study protocol in PROSPERO (CRD42021266695).
To identify relevant grey, non-published, literature, we looked at medRxiv, psyArXiv, and bioRxiv pre-print databases.A manual search of the references of the included primary studies and reviews related to this topic was conducted.Articles identified were screened as abstracts by two independent researchers (UP, IC; MSc level researchers, supervised by a senior academic researcher).After excluding those that clearly did not meet our inclusion criteria, the full texts of the remaining articles were then assessed for eligibility and decisionsweremaderegardingtheirfinal inclusion in the review.
The inclusion criteria were as follows: 1) individual studies with original data, including grey literature, 2) conducted in children and adolescents with a mean age of <18y, in line with previous studies, 13 3) assessing exposure to COVID-19, 4) reporting the percentage of anxiety and/or depression symptoms/disorder in this population, 5) in English.The exclusion criteria were as follows: 1) conference proceedings, case reports, narrative reviews, or qualitative studies, 2) studies including adults with a mean age >19-years-old, 3) studies in which children and adolescents were not exposed to COVID-19, 4) studies focusing on physical health outcomes or mental health outcomes other than anxiety.

Data extraction
Data extraction was led by the main researcher (UP) and also independently extracted by members of the research team who extracted data in pairs and attained consensus via group discussion (HAG, KP, BK; MD/Doctorate level researchers, supervised by a senior academic researcher).Any discrepancies arising were resolved through consensus, consulting another senior academic researcher (GSP) if an agreement was not attained.The variables extracted included: first author, year of publication, country(ies), wave of COVID-19 during study duration, whether there was school closure, the percent attending school, whether there was a lockdown (as operationalised in each study -such as, home confinement, school closure, quarantine etc.), months since March 2020 -when COVID-19 was declared a global pandemic, study design (cross-sectional, longitudinal), sample size, sex (% females), age (mean §SD, range), tools to assess anxiety symptoms (see eTable 4), mental health outcome (anxiety measurement instrument used, threshold of the anxiety measurement instrument used), quality appraisal (see below) and key findings.

Strategy for data synthesis
A meta-analysis for the prevalence of anxiety symptoms was conducted as our primary outcome.A secondary analysis was conducted on the prevalence of anxiety disorders according to DSM/ICD criteria.Random effects model metaanalyses were conducted pooling the percentage (95%CI) expressing the rate of the symptoms/disorders.We also tested using sub-analyses the prevalence of anxiety symptoms by continent (Europe, Asia, North America, South America), wave of the pandemic (first wave − up to June 2020 −, second/subsequent waves − from July 2020), school closure (yes, no) lockdown status (yes, no), design (cross sectional, longitudinal).
Hedges' g was used to estimate parameters and effect size in the meta-analysis.Publication biases were assessed with the metafunnel function visually by examining the funnel plot and quantitatively by the Egger'st e s t . 14The trim and fill method was used to correct the estimates in the case of publication biases.The presence of significant heterogeneity amongst study point estimates was assessed using Q statistics.The proportion of the total variability due to true heterogeneity was evaluated with the I 2 index.We conducted metaregressions including age, sex, sample size, percent attending schools and longitudinal progression (i.e., time passed from March 2020) as moderating factors.The meta-analyses were conducted using Comprehensive Meta-Analysis (CMA) V3. 15 Results of the systematic review were synthesised narratively and summarised in tables.Results were stratified by clinical at-risk populations and risk factors, followed by protective factors.

Quality appraisal
For study appraisal, this systematic review used the Newcastle-Ottawa Scale (NOS).We used the appropriate version of the NOS according to the design of the studies (cross-sectional studies (see appendix) or cohort studies 16 ).The quality assessment process was led by the main researcher, and also independently assessed by other co-authors (HAG, KP, BK).Any disagreements that arose were discussed with the corresponding author and resolved.

Search results
Our electronic search identified 16,082 publications; a further 6 publications were identified via backward searching of key papers.1043 and 166 publications went through abstract screening and full-text screening, respectively; 92 publications were excluded (see eTable 7), and 74 publications, including one study from pre-print databases, 17 met the criteria for inclusion.After accounting for overlap in study outcomes, 35 samples reported meta-analysable results for anxiety.Results of the search follow in the PRISMA 2020 flow diagram (Fig. 1).

Meta-regressions between the outcome of anxiety and other predictors
Increased number of months since March 2020 (B = 0.150, p<.001) and increased % of females (B = 0.103, p<.001) were associated with a higher prevalence of anxiety symptoms (see eTable 6).Meta-regressions for age, % attending school, and sample size did not result statistical significance (all p>0.05) (eTable 6).
Publication bias (small study effect) assessment Some asymmetry was found within the funnel plot (Fig. 5) and Egger's test quantitatively suggested publication bias (t-value=3.695;p=.00079).However, trim and fill value did not adjust the results and the plot did not indicate smalleffect bias (Fig. 5).Thus, publication bias was not found.

Discussion
This systematic review and meta-analysis provides a large and comprehensive review on the presentation of anxiety symptoms and disorders in children and adolescents.The pooled estimated prevalence of anxiety disorders during the COVID-19 pandemic was 13.0%, and that of anxiety symptoms was 26.5%.This meta-analysis found the prevalence of anxiety disorders during the COVID-19 pandemic to be twice as high as the worldwide prevalence of anxiety disorders before the COVID-19 pandemic. 8Meta-regressions found anxiety symptoms to be significantly more frequent in females.Furthermore, our results suggest that children and adolescents' anxiety symptoms significantly increased since the onset of the pandemic, in March 2020.Anxiety symptoms were more prevalent in the second wave of COVID-19 than the first wave of COVID-19, more prevalent in young people whose school had closed, and more prevalent in North America and Europe than in Asia and South America.Previous meta-analyses of the COVID-19 period have found the prevalence of anxiety symptoms to be between 28.0% 9 -31.0% 10 in college students, which is comparable, if not higher, to the figures that this meta-analysis found, suggesting that the increase in anxiety symptoms extends from childhood and adolescence to early adulthood. 55The 26.5% prevalence of anxiety symptoms found in this meta-analysis was higher than that found in a previous meta-analysis (20.5%), 6 suggesting that anxiety symptoms kept increasing, up to 29.3%, during the research period from the start of the pandemic in December 2019 until June 2021.
Resilience is our ability to adapt well to adversity and is shaped by individual, familial, and systemic factors. 56,57ypically, young people show high levels of resilience. 58owever, resilience in young people may have a limit, with anxiety symptoms increasing from wave 1 to wave 2 of COVID-19.The first wave of the pandemic may have been associated with the fear of the unknown as the developing COVID-19 situation is unpredictable and novel 59,60 ; furthermore, young people may have been exposed to consistently increasing morbidity and infection rates 2 alongside anxietyprovoking media. 18With repeated activation of acute stress mechanisms, subsequent waves may be associated with chronic stress, 61 post-traumatic stress 62,63 and hypochondriasis. 59Children seem to cope better with shorter lockdowns 64 and face more mental health difficulties with longer lockdowns, 65,66 showing further that resilience in children may have a limit.Furthermore, the mental health difficulties seen with longer lockdowns emphasise that they are only appropriate in emergency cases, due to the negative impacts of social restrictions and their associated stressors.Consequently, children and adolescents should have access to resilience-based interventions, 67,68 including the provision of academic skills at the individual level, support for family cohesion, and supportive teacher interactions. 69urther longitudinal research is required on the impact of the COVID-19 pandemic on anxiety in young people to assess if further waves do lead to chronic stress or if young people habituate 70 to the unprecedented global situation which may become their new normal. 71chool closures resulted in more anxiety symptoms in children and adolescents.One potential hypothesis is that this could be due to parental burnout and limited support from teachers during remote learning, 52 alongside difficulty accessing school-based psychological services. 725][76] One in seven children were unable to access the internet at home, the rate of which is two times higher in low-income communities 77 ; which may exacerbate social comparison. 78The compounded social and economic difficulties may complexify the transition to remote learning. 79While school closures are effective in controlling COVID-19, the psychological harm for young people could be profound, 80 as children face disruptions in school routines. 72Families with low socioeconomic status should be supported with financial safety nets, with schools ensuring all students have access to resources required to get online, to ensure equity.Furthermore, school closures, which are needed for protection, occur during periods of high transmissibility, and may result in more anxiety in young people due to the underlying emphasis on high infection rates.Schools should aim to not overwork students and be mindful of the mental health difficulties vulnerable students are facing.
Our meta-regression analysis showed female sex to be associated with higher levels of anxiety symptoms 28-31,33,35- 37,42-45,49,50,53,54,81-86 and chances of developing anxiety disorders 87 during the COVID pandemic.This seems to be more pronounced during adolescence, coinciding with puberty and a shift in hormones 88 ; females in general are more prone to experiencing anxiety than males. 89Females reportedly face more interpersonal stressors, 90 including family, peer, and intimate relationships, 91 and were more likely than males to experience increases in interpersonal violence during lockdowns. 92Females also ruminate more, 93 which may adversely impact their mental health and could be targeted for early intervention support.Age was not found to be a predictor for anxiety symptoms or disorders in our metaregression analyses.
Further specific vulnerable groups identified by this systemic review include children and adolescents with neurodevelopmental disorders or Special Educational Needs or Difficulties (SEND), 20,23,25 haematological or oncological diseases, 22 neurological disorders 23,24 or hearing difficulties, 21 amongst others.Vulnerable young people have lost both specialised educators 52 and a structured learning environment due to the pandemic. 77Children with SEND may experience a delay in the developmental progress of essential skills resulting from a suspension of social groups and therapy sessions. 94Those with chronic illnesses are reported to have higher levels of anxiety during pandemics. 95,96Physical health conditions may affect one's immune system and result in more COVID-19 related anxiety due to the higher risk of vulnerability-based complications during COVID-19. 97urthermore, as those with chronic physical health conditions already face more disability and non-disability challenges, 98 they may experience diminished resilience when facing the additional stressor of COVID-19.Young people with disabilities may also face difficulties in accessing the routine medical care they rely on as healthcare systems are at risk of being overwhelmed by COVID-19, leading to an increase in anxiety. 99Services should attempt to check-in with those known to be medically vulnerable to mediate negative mental health effects because of the chronic and acute difficulties they face.
Children with behavioural problems, pre-existing mental health difficulties, and children who already had a poor relationship with their parents prior to the pandemic are especially vulnerable to an exacerbation of conflicts at home 100 which may result in exacerbation of anxiety symptoms. 33,46,101Home is not a safe space for all children 99 and schools being shut means children are less able to access safeguarding networks that would be available to them if schools were open. 102Furthermore, studies included in this systematic review found parental stress and anxiety to be risk factors for anxiety symptomatology within children and adolescents. 26,27,103Parents experiencing parental burnout during COVID-19 are more likely to engage in child abuse and neglect, 104 posing a short-and-long-term risk for children. 105,106Some parents, for instance, healthcare workers, may be under particular pressure which may result in an exacerbation of anxiety symptoms in children. 30,38,107his is related to long working hours, highly emotive work, and increased potential exposure to COVID-19. 108,109his systematic review found that pre-pandemic anxiety symptoms were predictive of anxiety symptoms severity during the pandemic. 46Having previously accessed psychiatric help is a risk factor for the exacerbation of anxiety symptoms. 25Those known to have previous psychiatric disorders should receive early intervention to help them to cope with expected difficulties and prevent further mental health consequences.Telemental health services have shown to be effective 72 and should be adapted to reach vulnerable groups in an extended scope.
This meta-analysis found young people from western continents, including Europe and North America, to report more anxiety symptoms than young people from Asia and South America.Young people from Asia face a range of more severe adverse risks before COVID-19. 110Children and adolescents from high-risk backgrounds are found to show more resilience. 111Furthermore, young people from cities are reported to face more mental health difficulties than those from rural areas. 112There is a lower prevalence of people who use the internet or own a smartphone in Africa, Asia, and South America, in comparison to people from Canada, Australia, or the Netherlands. 113Excessive exposure to COVID-19 related media 18,25,34,35 is a risk factor for anxiety symptom exacerbation.Previous research has supported this by showing increased smartphone usage is associated with anxiety levels 114 which is more prevalent in western continents.Limiting exposure to media may result in decreased anxiety levels. 115Another hypothesis would be that fear relating to COVID-19 is a risk factor for anxiety symptoms. 30- 32,36-41,101,116Since the start of the pandemic, western Europe and North America have had higher rates of COVID-19 community infections, with a higher death rate per million inhabitants, compared to eastern continents and eastern Europe. 117Low-income countries have an estimated infection-fatality rate of 0.23% compared to 1.15% in highincome countries. 118The rate of COVID-19 in the community is associated with more COVID-19 fear and anxiety 36 and a lower rate of community COVID-19 cases in low-income countries could explain why there are lower subsequent levels of anxiety.In addition, children from western countries may be more open to report anxiety and communicate their feelings compared to children from eastern countries; this may come from the cultural model of individualism in western cultures more easily allowing individuals to express internal personal characteristics and emotions. 119Furthermore, research has found stressful life events to be perceived as less stressful than breaking cultural taboos 120 ; therefore, there are cultural differences in what is perceived as stressful or anxiety inducing.
More government support is needed to promote recovery.Some studies report disasters to require global and governmental efforts to maintain and restore resilience within society. 121,122This should involve both a rise in funding and the provision of more trained staff to Child and Adolescent Mental Health Services (CAMHS).As COVID-19 is exacerbating young people's mental health and resulting in the emergence of anxiety, the demand on CAMHS, an already stretched service, will grow. 74To prevent further adversity to young people's mental health, we need to provide an adaptable, multi-disciplinary support system 74 to protect children and adolescents' mental health.][125] The COVID-19 pandemic has resulted in a shift in mental health care for young people. 126To ensure continuity of care, mental health services have adopted a telehealth support format. 127Remote interventions have shown good effectivity in anxiety treatment. 128Research has found minimal therapist contact therapies to be efficacious for a range of anxiety diagnoses, 129 showing that face-to-face contact may not be essential for therapeutic efficacy. 129elehealth interventions are cost-effective 130 and reduce barriers to care by improving access. 131Telehealth-based psychotherapy has a range of uses within paediatric care 132 however, can be limited by one's psychological resistance The European Journal of Psychiatry 37 (2023) 100218 to novel treatment methods, cultural background, alongside their technology literacy. 133Most reviews looking at the efficacy of remote psychotherapy interventions for anxiety occurred pre-COVID-19.More research including randomised control trials and meta-analyses are needed to evaluate the effectiveness of remote psychotherapy during COVID-19 for children and adolescents so that the best care can be provided.

Limitations
The findings of this review should be interpreted in the context of various limitations.First, 52.7% of included studies either did not provide clear outcome data or were too heterogeneous or not rigorous enough to be meta-analysed (e. g., the thresholds of the instruments were not in line with validation data).Only four studies provided meta-analysable data regarding anxiety disorders.Second, 87.9% of included studies were from the first wave of COVID-19.More studies on the 2nd and 3rd wave of COVID-19 are needed to evaluate the long-term impact that COVID-19 has on the mental health of young people.Third, 86.5% of included studies were cross-sectional, which limits causal inference.Future longitudinal research should follow those presenting anxiety symptoms during the pandemic to assess if the symptoms presented reduce, continue, or develop into anxiety disorders either as the pandemic continues or post-COVID-19.

Conclusion
This meta-analysis found a high prevalence of anxiety symptoms and the emergence of anxiety disorders in children and adolescents during the COVID-19 pandemic.Specific vulnerable groups include young people in North America and Europe, and females.Identifying those at risk is the first step in developing more appropriate community-based preventative clinical interventions.

Table 1
Test for heterogeneity.