Effects of the COVID-19 Pandemic on Early Childhood Development and Mental Health: A Systematic Review and Meta-Analysis of Comparative Studies Authors: Sasha Alcon, Sa Shen, Hong-nei Wong, Cynthia R. Rovnaghi, Leni Truong, Jordan K. H. Vedelli, Kanwaljeet J. S. Anand Published: 2024 ABSTRACT From 2020 to 2023, the coronavirus-19 (COVID-19) pandemic exposed children to a variety of adverse childhood events, including parental loss, abuse, and disruption in services, and it exacerbated societal inequities. Studies evaluating the mental health of older children and adolescents reported increases in depression and anxiety symptoms, but no reviews have addressed the effects of the pandemic on preschool children. This systematic review and meta-analysis is the first to explore these effects. The goal was to analyze and synthesize longitudinal cohort studies to determine impact of the COVID-19 pandemic on the development and mental health of young children. Searches of multiple databases were performed for studies published between 2018 and 2023 with pre- and post-pandemic evaluations of the mental health or development of preschool children (aged 0–6 years) using objective measures and according to Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. Cohen’s d effect sizes were calculated for each study that utilized the Strengths and Difficulties Questionnaire (SDQ), Ages and Stages Questionnaire (ASQ), or the Child Behavior Checklist (CBCL). Random-effects mixed models combined the estimates of effect sizes to calculate the overall mean effect size. The meta-analyses included 22,348 children from 16 countries. The analyses showed small increases in emotional symptoms and conduct problems, as well as increases in emotional reactivity, anxiety/depression, withdrawal symptoms, attention problems, and aggressive behaviors. A decrease in fine motor and personal–social skills was noted. Studies not included in these meta-analyses also showed negative effects on language and executive function. This systematic review characterizes the detrimental effects of the COVID-19 pandemic on the mental health and development of preschool children across the world. Our results suggest the vulnerability of early childhood to pandemic-related disruptions, although the heterogeneity in study design and child characteristics may limit some of these conclusions. Introduction Children experienced stressors and significant environmental disruptions during the pandemic. School closures disrupted mental health counseling, early intervention programs, access to peers, group activities, and safe spaces [1,2]. In the United States, approximately 56% of childcare centers temporarily closed or shut down indefinitely during the pandemic [3]. Preschool children aged 0–6 years are at greatest risk for child abuse and neglect [4,5], but they may not report these adverse experiences because of their innate behavioral inhibitions and limited verbal repertoire [6]. Because the youngest children are completely dependent on their parents, Adverse Childhood Experiences (ACEs) in early life are underreported and underestimated [4,5,7,8]. In addition to ACEs or other stressors [9], preschool children are exposed to frequent, low-grade Early Life Stress (ELS) due to parental absence [10] or incarceration [11,12], high parenting stress [13,14], parental ACEs [15], inconsistent and variable parenting, low parental self-esteem, and/or low resilience [16–18]. Increased screen time for children and parents has further reduced their opportunities for supportive parenting and positive and dyadic interactions [19–22], leading to the inadequate buffering of chronic stress in small children [23,24]. Preschools and daycare centers serve as critical resources in which children develop social skills through peer interactions and structured learning activities. Additionally, multiple studies have shown childhood social isolation is associated with poor school performance, unhealthy relationships, and criminal behavior as children grow up, at times lasting even into adulthood [3,25–30]. This loss of learning spaces and social interactions was further exacerbated by difficult home learning environments that were impacted by the pandemic. Many parents faced significant impacts on employment, including needing to take leave, reduce work hours, or even leave their jobs to care for their children [31]. This disproportionately affected families from low socioeconomic backgrounds [31]. Furthermore, increased stressors, such as financial stress or changes in living conditions, can negatively impact parental behavior and affect cognitively stimulating parent–child activities, leading to suboptimal home learning environments and long-term effects on child school readiness and school achievement [32,33]. Araujo et al. described the effects of previous pandemics such as H1N1 influenza, acquired immunodeficiency syndrome (AIDS), and Ebola on children, including increased risks of adverse childhood experiences (ACEs), toxic stress, mental/emotional problems, developmental delays, and health problems in adulthood [34]. Given the well-known effects of ACEs and other stressors on the mental health of children [35–39], it is crucial to characterize the effects of the pandemic on early childhood development. Although previous studies have focused on older children and adolescents, this is the first meta-analysis to examine the mental health and development of preschool children. As the mental health crisis evolves, these analyses may help to support policies aimed at increasing capacity for younger children and designing social restrictions to limit their adverse effects on this vulnerable population. Materials and Methods Search Strategy A systematic review and meta-analyses were conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. This systematic review was registered with the International Prospective Register of Reviews (PROSPERO) (CRD42024536928) in April 2024, after the review was conducted. The review protocol may be viewed on the PROSPERO register. The Covidence® software platform (Melbourne, Australia; https://www.covidence.org/) was used to search for and evaluate relevant studies from Embase, PsycINFO, and PubMed. Both Embase and PubMed are extensive databases that provide over 30 million abstracts and papers and are well regarded as the top medical and healthcare research databases. PsycINFO was chosen given its focus on psychology-related research, which is pertinent to our query. For each database, a variety of MeSH terms were used, including preschool, toddler, infant, development, coronavirus, and mental health (see detailed search strategies in Appendix A). These keywords were entered utilizing Boolean search operators and all searches were conducted from June 2022 to June 2023. Selection Criteria All studies were reviewed by two reviewers for title and abstract screening. Studies that met the criteria from the initial screen were then reviewed independently by three reviewers for inclusion. Studies were included if they (1) were published in peer-reviewed journals, (2) included participants aged 0–6 years old, (3) evaluated participants at two time-points, from before (2018–2020) and after the specific country’s COVID-19 quarantine period (2020–2023), (4) contained objective assessments for one or more domains of child development, including physical, cognitive, behavioral, emotional, communicative, and social development, (5) were written in English, and (6) contained data representing 80% or more participants. Children aged 0–6 years of age were chosen as children under the age of 6 were most affected by the closure of preschools and daycare centers. We excluded studies only reporting qualitative data, unpublished preprints, abstracts, case series, and case reports. Data Extraction We evaluated the title and abstract of each citation for possible relevance. Full-text versions of those papers deemed possibly relevant were reviewed and included if the study inclusion criteria were met. Our systematic review contained 28 studies, with 20 reports eligible for quantitative meta-analyses based on the outcomes measured (Figure 1). Study Quality Quality assessment of the selected studies was performed based on National Institutes of Health (NIH) criteria for observational and cohort studies [40]. All eligible studies meeting our selection criteria, noted above, were scored 0 vs. 1 for each of the 14 quality criteria and summed for a total quality score [40]. We assigned ‘good’ quality for studies scoring 9 or higher, ‘fair’ quality for those with scores of 5 to 8, and ‘poor’ quality for studies scoring 0 to 4. Only studies with ‘good’ or ‘fair’ study quality were included in our meta-analyses. Statistical Data Analyses Data on the population demographics, study characteristics, and child outcomes were extracted from each study, databased, and reviewed twice to minimize errors. Given the limited number of studies, data were extracted by a single reviewer, and all data were reviewed by three additional team members independently. If data were missing, authors were contacted via email to obtain additional data. For each study, differences between the mean pre- and post-pandemic scores were calculated to estimate between-group effect sizes. Under the assumption of independence, Cohen’s d effect sizes were derived from the available descriptive statistics (means, standard deviations, correlation coefficients) or the test statistics (T statistic, F statistic) reported in each study. Effect size and its 95% confidence intervals were estimated for each outcome of interest. Random-effects mixed models were used to combine the estimates of effect sizes to calculate the overall mean effect size. To evaluate whether the true effect sizes were exchangeable across studies, study ID was specified as a random-effects term in each model. Both the estimated mean effect size and random-effects variance were estimated. Forest plots were used to summarize the effect sizes of individual studies. Each study and combined effect sizes were depicted as point estimates bounded by confidence intervals. Homogeneity tests were performed to examine the variation in study outcomes across studies. The inconsistency of studies’ effect sizes was measured by Cochran’s Q and I2 statistics. Cochran’s Q is the weighted sum of squared differences between individual study effect sizes and the combined effect size across studies, whereas the I² statistic describes the percentage of variation across studies that is due to heterogeneity rather than chance [41]. Publication bias was examined graphically using funnel plots, which plot study size on the vertical axis and effect size on the horizontal axis. Funnel plot asymmetry was examined and tested using four different methods: Egger weighted linear regression [42], Begg Rank Correlation [43], Funnel Plot Regression [44], and Trim and Fill methods [45]. The meta-analyses were performed using SAS v9.4 statistical software (SAS Institute, Cary NC, USA), and a SAS macro (PUB_BIAS) was implemented [46,47]. The critical threshold for α-error was set at 0.05. Results Study Characteristics The studies eligible for the systemic review ranged in sample size from 40 to 200,000 participants and 22,348 subjects were included in the quantitative meta-analysis. Participants were followed for periods from 3 weeks to over a year. Eligible studies came from a variety of locations around the globe. Twelve European studies were found, including five from Italy, two from Germany, one that combined children from Austria and Italy, and one each from Spain, Switzerland, the United Kingdom, and Denmark. Five studies came from North America, with three from the US and two from Canada. Three studies were conducted in China, and the remainder were from Turkey (two), Israel (two), Japan (two), Nepal (one), and Brazil (one). The study characteristics are presented in Table 1. Additional study characteristics can be found in Table A1. Only one study included children with known risk factors for neurodevelopmental delay [48]. Of the studies analyzed, 12 studies used the Strengths and Difficulties Questionnaire (SDQ) to assess 5622 pre-pandemic and 5042 post-pandemic children (n = 10,664), five studies used the Child Behavior Checklist (CBCL) to evaluate 3262 pre-pandemic and 3899 post-pandemic children (n = 7161), whereas three studies used the Ages and Stages Questionnaire (ASQ-3) to assess 3033 pre-pandemic and 989 post-pandemic subjects (n = 4022). The remaining studies used a variety of other measures including the Oxford Communicative Development Inventory (O-CDI), the Spence Preschool Anxiety Scale (PAS), Griffiths Scale for Childhood Development (GCSD), the Child Faces Task, and new tools for measuring anxiety or depression. Changes in Strengths and Difficulties Questionnaire (SDQ) The forest plot for the effect size of pre- and post-pandemic changes in SDQ (Figure 2) revealed minor decreases in the total score (d = −0.088, 95% CI −0.247 to 0.07), prosocial behaviors (d = −0.145, 95% CI −0.474 to 0.185), hyperactivity (d = −0.012, 95% CI −0.124 to 0.1), and peer problems (d = −0.112, 95% CI −0.275 to 0.501), while emotional symptoms (d = 0.07, 95% CI −0.044 to 0.184) and conduct problems showed a negligible increase (d = 0.081, 95% CI −0.111 to 0.273) (Table 2). No publication bias was evident from the funnel plot (Table A2, Figure A1, Appendix B). Moderate or high inconsistency in the study results occurred across the SDQ subscales and total SDQ scores (I2 range: 54.17% to 93.57%). Cochran’s homogeneity Q tests suggested significant heterogeneity across the studies and in their outcomes (Table A3, Appendix B), although the between-study variance was small across the total SDQ and subscale scores. Changes in Ages and Stages Questionnaire (ASQ-3) The effect sizes of the changes in ASQ-3 scores pre- vs. post-pandemic (Figure 3) showed negligible-to-small improvements in communication (d = 0.274, 95% CI −0.08 to 0.628), gross motor (d = 0.111, 95% CI −0.127 to 0.349), problem-solving (d = 0.065, 95% CI −0.041 to 0.171), and overall scores (d = 0.069, 95% CI −0.032 to 0.17), coupled with minimal decreases in fine motor (d = −0.036, 95% CI −0.384 to 0.312) and personal–social skills (d = −0.036, 95% CI −0.215 to 0.143) (Table 2). The funnel plot showed no evidence of systematic publication bias (Table A4, Figure A2, Appendix B). The χ2-test for homogeneity was significant for communication, gross motor, fine motor, and personal–social scores, suggesting significant heterogeneity across the studies (I2 values > 80%). The total ASQ and problem-solving skills indicated no significant heterogeneity across the studies. The between-study variance was small for the total ASQ score and all the subscales (Table A5, Appendix B). Changes in the Child Behavior Checklist (CBCL) The effect sizes for the multiple subscales of CBCL (Figure 4) showed evidence for small increases in emotional reactivity (d = 0.344, 95% CI 0.087 to 0.6), anxiety/depression (d = 0.464, 95% CI 0.263 to 0.722), withdrawal symptoms (d = 0.42, 95% CI 0.163 to 0.678), attention problems (d = 0.28, 95% CI 0.024 to 0.536), and aggressive behaviors (d = 0.277, 95% CI −0.016 to 0.571, p = 0.206), associated with minimal decreases in somatic complaints (d = −0.075, 95% CI −0.428 to 0.277) and sleep problems (d = −0.085, 95% CI −0.340 to 0.169) (Table 2). Internalizing problems increased significantly (d = 0.156, 95% confidence intervals: 0.108, 0.203), while other changes were not significant. Publication bias was not significant (Table A6, Figure A3, Appendix B). Only the aggressive behavior subscale showed significant heterogeneity based on the χ2 homogeneity test (p = 0.025, I2 = 45.65%). The between-study variance was small for the total CBCL score and all the subscales (Table A7, Appendix B). Individual Study Findings (Not Eligible for Quantitative Meta-Analyses) Eight papers were not included in the meta-analysis based on the unique outcomes measured and inability to pool their data with other studies. Davies et al. reported the language skills of children aged 8–36 months from 189 English-speaking families in the UK. The results showed that receptive vocabulary growth (O-CDI) was greater in children who continued to attend early childhood education and care (ECEC) over a 6-month period, with stronger positive effects for children from less advantaged backgrounds. Cognitive/executive functions (Early Executive Functions Questionnaire) were boosted by ECEC attendance, regardless of socioeconomic background [52]. Ferrari et al. used Griffiths Scales of Child Development (GSCD) to assess the global development of infants aged 6 and 12 months. The global development scores decreased during the periods of COVID restriction in Italy (pre-COVID n = 34, median 98, IQR 97–103 vs. post-COVID n = 70, median 94, IQR 90–100; p < 0.001), related to the severity of the restrictions [69]. Wermelinger et al. studied child development, specifically evaluating changes in child facial recognition and emotion labeling. No pandemic-related differences occurred in the children’s emotion labeling, suggesting that they received enough input from masked faces to support the normal development of emotion recognition [72]. Fitzpatrick et al. evaluated children for changes in their screen time and temperament utilizing the Children’s Behavior Questionnaire from Spring 2020 (n = 316) to Spring 2021 (n = 265). Lower parental education contributed to more child screen time at the ages of 3.5 (ß = 1.37, p < 0.001) and 4.5 years (ß = 1.76, p < 0.001). Based on the cross-lagged panel model, child screen time at age 3.5 significantly contributed to decreased effortful control scores at age 4.5 (ß = −0.10, SE = 0.042; p = 0.023), whereas effortful control at age 3.5 did not contribute to child screen time at age 4.5 (ß = 0.016, SE = 0.046; p = 0.729) [54,75]. Kolcakoglu et al. utilized the Spence Preschool Anxiety Scale (SPAS) to assess 405 children over a 2-month period during the first lockdown in Turkey and found increases in separation anxiety and physical injury anxiety, associated with increased tantrums (p = 0.010), crying attacks (p = 0.010), and aggression (p = 0.010) in these children during COVID as compared to pre-COVID [56]. Kostev et al. also reported similar trends with increased new anxiety and depression diagnoses from outpatient pediatric practices in Germany during the pandemic, as compared to 2019–2020 [73]. Alonso-Martinez et al. used the Child Self-Regulation and Behavior Questionnaire (CSBQ) to evaluate child self-regulation, as well as internalizing and externalizing problems, in Spain. Over a 6-month period, the children showed significant decreases in physical activity (−43.3 min/day, 95% CI −68.1 to −18.5) and sleep efficiency (−2.09%, 95% CI −4.12 to −0.05), and increases in sedentary time (+50.2 min/day, 95% CI 17.1 to 83.3), associated with greater internalizing (0.17, 95% CI 0.06 to 0.28) and externalizing (0.33, 95% CI 0.23 to 0.44) problems [60]. Hanno et al. reported similar results, showing that internalizing, externalizing, and dysregulated behaviors increased in US children while their adaptive behaviors declined after the pandemic lockdown [74]. Discussion To our knowledge, this is the first meta-analysis to characterize the effects of the COVID-19 pandemic on the mental health and developmental outcomes of children aged 0–6 years. Over 3000 papers were screened, yielding 28 studies for systematic review and 20 studies for meta-analyses, which included 22,348 participants from around the world. Like the meta-analyses performed on older children and adolescents [76–78], our findings suggest negative trends in mental health, cognitive, behavioral, and other developmental domains during the COVID-19 pandemic, as compared to pre-pandemic periods. However, many of these effects were not significant. The child behaviors included difficulties in emotional control (SDQ/CBCL data, others [55]), behavioral conduct, and socializing behaviors (SDQ, ASQ-3, CBCL data). Cognitive or executive functions [51,53,71] and overall development also showed some decline (SDQ data, others [68]), and were associated with increases in anxiety, depression, or withdrawal (CBCL data, others [55,72]) and internalizing or externalizing behaviors (CBCL data, others [59,73]) during the pandemic. Because of the significant heterogeneity and small numbers of studies using the same instrument, most of the findings from our meta-analyses were not significant, except for the internalizing problems from the CBCL data. Despite this, the patterns of similar findings across all the instruments raise concern for early childhood development during the pandemic. The SDQ assessments showed negligible negative effects on emotional symptoms, conduct problems, and prosocial behaviors, whereas the CBCL data showed a small negative effect on emotional reactivity, anxiety/depression, and withdrawal. Small negative effects were also seen with attention problems, aggressive behaviors, and internalizing and externalizing problems, all of which are consistent with the current literature on the effects of stress in children [79–81]. These results can be explained by studies that show that prolonged stress can disrupt brain circuitry and stress regulatory systems that can affect physiology and behavior [82–84]. In addition to behavioral and mental health, our meta-analysis also analyzed changes in child development on the ASQ-3. Overall, the effect sizes for all ASQ-3 domains were small, suggesting minimal negative effects of the pandemic on fine motor skills and personal–social skills, whereas communication, gross motor skills, problem solving, and total scores showed small improvements. Interestingly, Hessami et al. reported contrasting findings from a systematic review of infant neurodevelopment, with increased risks for communication delay among infants exposed to COVID-19 pandemic lockdowns in the first year of life [85]. Infancy and early childhood are sensitive periods for cognitive and language development, in which negative experiences can undermine a child’s typical developmental trajectory [86]. Davies et al. also found lower language skills among children not receiving early educational enrichment during the pandemic [53]. Our meta-analysis for ASQ-3 data included only three studies, which could not capture the full impact of the pandemic on early neurodevelopment. Our study showed small effects on various child mental and developmental health domains in a limited sample size. Further research is needed and may find, with a larger sample size, that these issues are more extensive than our study has been able to capture. These small changes can greatly affect a child’s future, and continuing to spurn these issues based on lack of statistical significance may further contribute to the growing mental health crisis across the globe. Our findings, and those of other studies analyzing mental health in children, bring attention to the need for resources and interventions to address the effects of the pandemic [38]. These interventions, whether community-based or primary-care driven, should focus on strengthening family social supports and positive parenting techniques, and teach parents how to facilitate social, emotional, and language skills [31–33,86–88]. In addition to improving access to programs that support families, research evaluating these programs and their efficacy is also needed [88]. Finally, more widespread screening for development and behavioral issues should be implemented to better characterize the full extent of the effect of the COVID-19 pandemic on early childhood development [72]. There are limitations to the current study. The findings from the quantitative meta-analyses did not reach significance because of the heterogeneity of the study designs, children’s characteristics, timing of assessments, small sample sizes, variable durations between pre- and post-pandemic evaluations, and inconsistent/incomplete application of the chosen tools. In addition, assessment methods such as the SDQ, ASQ-3, and CBCL are mainly designed as screening tools rather than diagnostic instruments. Therefore, they were not sensitive enough to reveal subtle changes in the children’s development. Some studies using the SDQ, CBCL, and ASQ-3 presented their results in a way that did not allow the calculation of pre- vs. post-pandemic effect sizes. Some studies also did not use all the available subscales in these screening methods, thus limiting the data available from the subscales. Given the limited data, only one reviewer performed data extraction. However, these data and the articles were then subsequently reviewed by three additional reviewers to ensure the accuracy of the extraction. To be eligible, all the studies were required to evaluate patients at two time points. However, the interval time between the two evaluations ranged from 3 weeks to 12 months. This discrepancy across studies may have contributed to the vast differences in the study results. Additionally, there was a lack of diversity in patient representation; specifically, few studies looked at patients with neurodevelopmental or psychiatric diagnoses. Additionally, language bias may have been present, as only articles available in English were able to be included. A large majority of the studies utilized parent evaluations and reports of the symptoms of the child, which may have led to reporting bias and higher reporting of problem behaviors. Additionally, this meta-analysis and systematic review did not evaluate the intrinsic familial protective factors or assistance programs instituted in various countries during the pandemic. Various countries instituted assistance programs throughout the pandemic to support families during the pandemic and quarantine periods, which may have also contributed to the small effects seen in this study. Conclusions This systematic review and meta-analysis of longitudinal cohorts suggests that pandemic-related stressors and social restriction mandates were disruptive for the development and behavioral health of young children across the world. While the effect sizes in our study were small, it is important to consider that even these small effects can make a significant difference in a child’s life course trajectory by further escalating pre-existing developmental concerns or mental health conditions, as well as adding to the cohorts of children having such conditions. These results identify a strong need for standardized early childhood assessment and intervention via child-supportive services, community-based resources, and family-resilience-building programs to address the risk of mental health problems among school children, adolescents, and youth. This evidence provides a strong impetus for policymakers’ consideration of measures to safeguard proactive childhood policies. It is critical that we act now to address these issues and develop a more robust infrastructure to support families in the event of another global pandemic. Supplementary Material