Background
The potential impact of the COVID-19 pandemic on youth mental health was anticipated early in 2020 [
1] as communities around the world began to “lock down.” In the United States and other countries, schools were closed, recreational and extracurricular activities suspended, and even casual interactions were undermined by social distancing measures. Along with the disruption in learning and daily routines, the hardship of isolation was compounded by uncertainty, fear, and an economic and socio-emotional toll on the caregivers, teachers, and community organizations that often provide a buffer to children’s distress. The fact that youth, though less vulnerable in terms of their physical health, were at high risk with regard to their emotional and behavioral health was widely acknowledged [
1,
2].
In time, studies from around the globe have begun to document an impact of these pandemic-related disruptions on the mental health of children and adolescents. Consistent with general trends in the empirical literature, there have been fewer published studies of youth compared to adults [
3]. Nonetheless, emerging data in this age group have related the onset of the pandemic to increases in anxiety and depression [
4,
5] behavioral difficulties [
6], suicidality [
7,
8], and psychiatric emergency room visits [
9].
While such findings underscore the negative impact of the spring 2020 phase of pandemic on youth, data suggest that the emotional and behavioral sequalae of this period were not uniformly experienced. In spring 2020, for example, a survey of Canadian parents that addressed six common psychiatric dimensions [
10] found that 70% of school aged youth had experienced worsening of at least one domain; yet, approximately half of the responses for each domain indicated no change or improvement in functioning. Such results highlight the need to better understand the variability of responses to the pandemic.
Already, data from the spring and summer of 2020 suggest that children and adolescents with pre-existing neuropsychiatric concerns may be vulnerable to difficulties [
1,
10‐
14]. In a review of studies, Panchal and colleagues [
15] concluded that youth with prior mental health concerns may be at increased risk for anxiety. Additionally, based on parent reports in a United Kingdom survey [
13], youth with prior diagnoses of attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) both experienced greater pandemic-related emotional difficulties as well as greater “inattention/ hyperactivity” than youth without these diagnoses. Moreover, youth with diagnoses of ADHD were more likely to experience conduct problems, whereas youth with autism were more likely to have a decline in prosocial behavior. Thus, some domains of decline may relate to prior symptoms.
Yet, even among youth with mental health concerns, the emotional and behavioral response to the pandemic is not uniform. Indeed, a prospective survey of youth within a U.S. charter school found that mental health concerns measured prior to the pandemic predicted improved functioning during the late spring of 2020, presumably due to reduced academic and social stress when away from school. Given the data above in clinical populations, heterogeneity of response is likely, but not well understood. For example, in Cost et al.'s [
10] Canadian survey, parent reported prior psychiatric diagnoses predicted improvement as well decline on different traits such as depression and irritability.
Gaining a better understanding of the variability in emotional and behavioral reaction to the pandemic in youth clinical samples is critical to mobilizing appropriate supports for youth who may be at greatest risk for difficulties. The current study aimed to advance the literature on clinical populations in several ways. First, we investigated the mental health impact of the pandemic on a generalizable outpatient child psychiatry sample in the United States, which to our knowledge has not been represented in prior studies. Second, we characterized the functioning of youth during the middle of the school year following the spring 2020 lockdown, thereby representing a later time period than prior studies of clinical populations. Third, we leveraged a computational strategy that allowed us to combine measurement of a wide range of psychiatric symptom domains with a child-centered approach to provide a snapshot of global functioning. Specifically, we used latent profile analysis (LPA) of changes in parent-reported retrospective pre-pandemic and current symptoms to identify groups of youth with similar patterns of multivariate change. Finally, we looked at predictors of patterns of change (i.e. predictors of LPA groups), including some that had not been used in prior studies, such as clinician-rated neuropsychiatric diagnoses and cognitive variables. Together, these analyses aim to extend our understanding of the variability of the mental health burden of the pandemic on clinical samples and potential predictors and correlates of different profiles of change.
Discussion
Studies from across the globe are documenting the impact of the COVID-19 pandemic on youth mental health. We extend this literature in several ways. Specifically, we document increased severity of a wide range of psychiatric symptoms within a generalizable child psychiatry outpatient sample in the United States and at a time period well beyond the 2020 spring lockdown. Further, we idenitfied distinctive profiles of psychiatric symptom change among referred youth. Encouragingly, between the pre-pandemic period and the mid 2020–2021 school year, a large group showed minimal difference in symptom severity. Additionally, a small group was characterized by improved symptoms; however, two groups had differing profiles of worsening symptoms, with shared and unique predictors and correlates. Such data confirm that the impact of the ongoing pandemic on youth is not uniform, even within a clinical sample and even among those who are struggling. As such, supports will need to be tailored to the unique needs of those who remain in distress.
Our first set of analyses focused on changes in individual psychiatric and psychosocial domains. Based on parent reports, over two thirds of the sample worsened on at least one of eight psychopathology domains. On average in the overall sample, seven domains showed a significant increase in severity, including: sad/depressed, worried/anxious, despair/ hopelessness, inattentive/ easily distracted, irritable/ gloomy, and lacking interest in social interactions. Consistent with adolescent self-reports in a UK survey [
30], which spanned the time prior to the pandemic and July 2020, hyperactive/impulsive symptoms failed to show a significant change on average within the whole sample, though other studies [
31] found increases in this domain when measured at an earlier time point in the pandemic.
Our US-based clinical sample is consistent with clinical cohorts from Canada [
10] and the Netherlands [
32] in showing change on a wide range of domains. Additionally, while other studies of clinical cohorts documented such symptoms in spring 2020, our results suggest that this increased symptom severity was present at even a later time, during the middle of the 2020–2021 school year. In the US, while the most extreme period of school closures and social isolation had abated by that point, significant psychosocial and educational disruption was still occurring [
33]. In contrast to evidence for a bounce-back effect for adults in China a month after the pandemic onset [
34], our results showed that psychiatric symptoms in clinically referred youth in the US were worse during a time period that was 10 months on average after the pandemic onset. Our finding that youth were impacted at this later date converges with longitudinal data from a population youth cohort in Germany [
35] showing that increases in individual symptom domains were observed in winter of the school year following the pandemic. While youth from a population cohort in Denmark [
36] did not identify significant depression symptoms in the overall group in fall of that year, they did note that youth with high pandemic related anxiety in the prior spring experienced greater depressive symptoms in the fall. This result is consistent with our findings of heterogeneity and of a relationship between functioning during the ongoing pandemic and prior symptoms.
Parent retrospective reports also indicated greater conflict between parents and children and, unsurprisingly, increased feelings of isolation in youth. Parents further reported a higher level of excessive screen time compared to the pre-pandemic period. Although our data are not longitudinal, the change from baseline into the 2020–2021 school year suggests the potential for prolonged exposure and that potential vulnerability to internet addiction and other negative effects of screen time should be followed up [
37]. We did not find worsening conflict with friends or increased use of alcohol or subtances during this time, likely due to the limited social opportunities outside of the home.
When we used LPA to characterize the heterogeneity of responses within our sample, participants segregated into four groups with statistically distinct profiles of symptom change. As noted, the majority of youth showed minimal change. We do not know whether members of this group had previously experienced greater distress at the height of the lockdown in spring 2020. Indeed, based on parents free-form verbal impressions of the impact of the pandemic, Asbury et al. [
38] posited that only a small subgroup of youth with ADHD and neurodevelopmental disorders had been unchanged/improved at that time. Nonetheless, it is encouraging that, even within a clinical sample, the majority of youth were relatively un-changed in their level of psychiatric symptoms during the following school year. This is not to say that the pandemic was not still challenging for these children nor that they didn't have psychiatric symptoms. Rather, our data speak to the fact that substantial increases in psychiatric symptoms during the 2020–2021 school year were not universal among referred youth.
Yet, the LPA also identified three subgroups whose psychiatric symptom profiles had changed compared to the pre-pandemic period. A small group of youth (6%) showed improved functioning. There were also two distinct groups with worsening symptoms. Both experienced increases in sadness/depression and worry/anxiety, consistent with studies from spring 2020 suggesting that these are some of the most common sequelae of the pandemic in youth [
20,
39]. Both groups also increased in their defiance/ oppositionality. Yet, members of one group exhibitied significant worsening of symptoms most commonly related to ADHD, particularly hyperactivity/ impulsivity but also inattention, and these youth showed minimal change in hopelessness/despair. In contrast, the other worse group showed notably increased symptoms linked to depression, including hopelessness/despair and social withdrawal with minimal change in hyperactivity/ impulsivity.
Although two prior studies [
40,
41] identified heterogeneous groups (based on profiles of behaviors and pandemic-related life changes, respectively) in non-clinical samples, our identification of distinct patterns of psychiatric symptom change within a clinical sample of youth is novel and helps to integrate discrepancies across prior studies showing worsening [
8] and improving symptoms or the worsening and improving of individual traits. By using overall child functioning rather than individual traits as our outcome, we found that different children were exhibiting globally distinctive patterns of stability, improvement and worsening by the mid 2020–2021 school year.
We also found that specific types of problematic symptoms prior to the pandemic associated with change profiles relevant to those symptoms. Prior hopelessness was associated with membership in the worsening group that stood out for its depressive symptoms, while prior conduct problems predicted the group with worsening hyperactivity/impulsivity with no increase in hopelessness. Such findings echo prior studies [
13,
41,
42] that suggest a worsening of prior symptoms.
We examined variables that had not been examined in prior pandemic related studies, including (1) clinician-rated lifetime diagnoses of ADHD and autism (rather than parental reported diagnoses) and (2) cognitive measures reflecting general ability, working memory, and processing speed. None of these variables were associated with change profiles. The lack of association with cognition was somewhat surprising given that general ability and executive functions are known to relate to successful problem solving and psychosocial adjustment [
43]. However, results are consistent with growing evidence for the separability of cognitive functioning and psychopathology per se [
44] and do not speak to the important question of whether cognition may relate to pandemic related adjustment within academic domains specifically. Additionally, having previously been on a special educational plan (504 or IEP) did not significantly associate with group membership; however, it is notable that the group whose functioning improved, presumably because they were experiencing less academic and social stress during the pandemic, showed the largest effect size for this variable.
We also found that certain change variables related to the groups with worsening symptoms. Increased screen time and increased isolation were significantly associated with increased risk of being in each of the two worse groups compared to the no change group. Increases in arguing/conflict with parents increased the risk for being in the hyperactive group. Because these variables changed over the same time period as the symptoms did, we cannot presume causality and must consider them correlates. Nonetheless, they do underscore the need for studies to determine whether reducing social isolation, excessive screen time and conflict could provide relief to youth who experienced worsening symptoms.
The current results represent an important step towards understanding the impact of the pandemic on clinical samples. Certainly, the toll on youth mental health generally is clear, and the impact of the ongoing disruption of social and educational structures has yet to be fully characterized. Nonetheless, our data provide arguably the clearest empirical evidence to date of the variability within a child and adolescent clinical sample, including evidence that subsamples of youth experienced significantly increased and distinguishable psychiatric symptoms that parents attributed to the pandemic as recently as the 2020–2021 school year. Given provider shortages in child psychiatry that existed even prior to the pandemic [
28] compounded with the continued burden of educational disruption, identifying the unique needs of youth at highest risk may allow for more targeted and effective care. This is not to say that programs targeting general wellness are not of value. Indeed, Copeland et al. [
45] found a small but significant benefit to the mental health of college students enrolled in a neuroscience based wellness program during the pandemic. Nonetheless, in clinical samples, youth who experienced notably increased hopelessness during the last school year in addition to other symptoms may require a different response than youth whose symptoms improved during the period of reduced interactions in academic and social settings.
Our findings should be considered in light of their limitations. Our assessment of the pre-pandemic period was based on parental retrospective reports and prospective data would have improved the accuracy of reports about the pre-pandemic period. Additionally, youth were assessed at different times during a six month period and, even though we controlled for the time of assessment, we cannot rule out the possibility that these differences contributed some variability to our sample. Importantly, the groups showing changes in symptoms were small, and the effect sizes for some variables (e.g. prior 504/IEP plan, irritability, increasing job insecurity) suggest that there may be additional predictors and correlates of group membership that we were not able to identify due to Type II error. Additionally, we acknowledge that the time frame of our assessment does not speak to functioning during the current 2021–2022 school year, and futher data on the current time period are needed. Finally, the fact that our sample is predominantly White creates uncertainty with regard to the generalizability of our findings to youth from other racial and ethnic groups who may have experienced additional or different stressors during the pandemic. For example, there is growing evidence for structural inequities that may place burdens on BiPOC (Black, Indigenous, and People of Color) youth that were compounded by the pandemic [
46]. Examination of clinical cohorts with greater racial and ethnic representation is needed to determine the relevance of our results to referred youth from these historically under-represented groups.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.