Remote vs. In-person Schooling during the COVID-19 Pandemic and Internalizing Symptoms among Children on the Autism Spectrum

Abstract

Background:

Children on the autism spectrum encountered interruptions to their education due to the COVID-19 pandemic (White et al., 2021). This study examined the extent to which autistic children’s anxiety and depressive symptoms during the pandemic were associated with the school format they attended in October 2020 (remote, in-person/hybrid, homeschool), controlling for their pre-pandemic symptoms. Pre-pandemic peer victimization and autism symptom characteristics were tested as moderators.

Method:

Participants were 81 verbally-fluent autistic children (Mage = 14.71 years, 77.8% males), without an intellectual disability, and their mothers; families were part of an ongoing, longitudinal study that began before the pandemic.

Results:

School format did not significantly predict children’s anxiety or depressive symptoms. Results indicated that the only significant predictors of children’s anxiety and depressive symptoms during the pandemic were their pre-pandemic symptom levels.

Conclusions:

School format did not significantly contribute to variability in children’s anxiety and depressive symptoms during the pandemic, over and above their pre-pandemic symptoms. The results contribute quantitative findings to the growing body of research on pandemic-related effects on autistic children, and underscore the need to account for pre-pandemic child functioning when drawing conclusions about pandemic-level effects.

The COVID-19 pandemic disrupted the lives of families across the globe and resulted in numerous changes to children’s daily lives (Eshraghi et al., 2020; Neece et al., 2020; see also Kauhanen et al., 2022, Lee et al., 2021). In a review of longitudinal studies that assessed individuals before and during the pandemic, Robinson et al. (2022) reported increased mental health symptoms among adults and children during the initial stages of the pandemic. Similarly, in a systematic review of 21 studies with children and young adults that included repeated measures of mental health before and after the pandemic found that children experienced worsened mental health symptoms throughout the pandemic (Kauhanen et al., 2022). Research specifically with autistic children reported a similar pattern of worsened mental health during early phases of the pandemic (Castro-Kemp & Mahmud, 2022; Franz & Kelly, 2021; Mutluer et al., 2020; Oomen et al., 2021; Vasa et al., 2021). For example, in an online study of mothers during the first two months of the pandemic, Vasa et al. (2021) reported that 45% of mothers indicated that their autistic child’s mental health symptoms worsened (especially depressive symptoms), and 50% reported that their child developed new symptoms. Relatedly, in a study of 89 mothers of autistic children in Ireland, levels of emotional problems and total behavioral difficulties were significantly higher when compared to mental health data from an independent sample of children collected prior to the pandemic (Franz & Kelly, 2021).

Disruptions to daily routines may partially explain the worsening mental health symptoms among children on the autism spectrum (Eshraghi et al., 2020; O’Sullivan et al., 2021; Ozsivadjian et al., 2023). Indeed, children’s education was substantially disrupted during the early months of the pandemic as schools moved classes online to accommodate lockdown restrictions and help slow the spread of COVID-19. Moreover, given that schools often provide services such as speech, behavioral, or occupational therapy to help students meet education goals, many autistic children lost services due to these school closures (Nelson & Murakami, 2020; White et al., 2021). The purpose of the present study was to test the extent to which different school format types in the fall of 2020 were associated with autistic children’s internalizing symptoms after accounting for their pre-pandemic mental health.

Prior to the pandemic, research suggested that children on the autism spectrum faced difficulties with traditional educational settings (Fitzpatrick et al., 2020; Reicher, 2020). Research on associations between COVID-19-related school disruptions and children’s mental health is nascent and results have been mixed. Castro-Kemp and Mahmud (2022) surveyed 83 parents in England between October and December 2020; a qualitative analysis of silver linings reported by parents indicated that the pandemic lockdown removed the stress of their child going to school. In one of the few quantitative studies, Toseeb and Asbury (2022) longitudinally assessed families in the UK during and after the first COVID-19 lockdown (i.e., March and October 2020). They found that although anxiety levels decreased for children with special educational needs and disabilities as they transitioned back to in-person school after the lockdown, this was not the case for autistic children, whose anxiety levels remained stable. Qualitative data from the same UK-based sample suggested that the reduced pressure to attend school was likely a contributing factor to improved well-being among autistic children (Asbury & Toseeb, 2022). The intensive social nature of traditional classrooms often exacerbates anxiety among autistic children (Adams et al., 2019). Thus, remote school may have allowed autistic children a reprieve from the pressure of navigating peer interactions alongside academic material, and therefore, an opportunity to flourish in an environment more tailored to their needs (Fegert et al., 2020; Reicher, 2020). In fact, Mumbardó-Adam et al. (2021) found that in April 2020, a majority of students on the autism spectrum reported that they did not want to go back to school after the lockdown was over. Collectively, these studies suggest that autistic children may have actually benefitted from school closures and the move to remote learning.

At the same time, however, studies also suggest that remote learning, especially during the pandemic when schools and teachers were not prepared to teach online, was stressful for families and associated with worse mental health difficulties for autistic children (e.g., Asbury et al., 2021). Parents of children with developmental disabilities reported higher levels of dissatisfaction with their school district’s solutions to socially distanced education (Baten et al., 2022). Moreover, approximately 40% of students on the autism spectrum who attended school remotely saw reductions in participation and achievement (Simpson & Adams, 2022). A potential reason for this may have been because remote learning was associated with a loss in daily routines that children had come to expect (Asbury et al., 2021) as well as loss of critical services for families (Nelson & Murakami, 2020; White et al., 2021). Research with neurotypically-developing children also reported negative effects of remote schooling on children’s mental health (Hawrilenko et al., 2021; Verlenden et al., 2021). For example, in a large study of neurotypically-developing children in December 2020, Hawrilenko et al. (2021) reported that older children who attended school remotely had more mental health difficulties compared to children attending school in-person; the negative impact of remote school on mental health was especially pronounced among children from families with lower socioeconomic status and among Hispanic and Black children.

In addition to remote school, some parents chose to homeschool their children during the pandemic. Prior to the pandemic, one study found that parents of children on the autism spectrum that implemented homeschooling reported an almost 60% decrease in problem behaviors and an approximate 50% decrease in child’s emotional stress (O’Hagan et al., 2021). Parents in the same study cited bullying, peer victimization, and teacher incompetency as reasons for homeschooling their children. However, in a study conducted in seven European countries, Thorell et al. (2021) reported that the majority of parents of neurotypically-developing children reported negative experiences with homeschooling after the onset of the COVID-19 pandemic. Many of these parents felt ill-equipped to support the educational needs of their children, especially those with mental health difficulties. Given the limited quantitative studies to date, however, the extent to which different school format types during the COVID-19 pandemic was associated with autistic children’s mental health remains unclear.

Moderators of School Effects on Mental health

Research conducted prior to the pandemic underscores that individual differences in how well children function in traditional, in-person vs. remote school may depend upon characteristics of the child and their environment (Baten et al., 2022). Thus, a secondary aim of this study was to test pre-COVID-19 characteristics as potential moderating factors of associations between school format and children’s mental health during the pandemic. Negative peer interactions, such as bullying, have been documented as frequent stressors for individuals on the autism spectrum (Scuitto et al., 2012). For example, Adams et al. (2016) reported that approximately 43% of children on the spectrum experienced verbal victimization at school and approximately 17% reported experiencing physical victimization within the last month; 61% of participants experienced peer victimization of any form within the last month. These negative interactions are strongly associated with low feelings of school enjoyment, belonging, and safety among children on the autism spectrum (Adams et al., 2016). Importantly, negative peer interactions between autistic children and their peers are directly associated with many psychopathological symptoms, including symptoms of anxiety and depression (Storch et al., 2012). Thus, among children who were experiencing higher levels of peer victimization, remote schooling might be associated with lower levels of mental health symptoms whereas in-person school may have been related to higher levels of mental health symptoms (Fegert et al., 2020).

Several studies indicate that autism symptom characteristics may also predict how an individual responded to and understood pandemic regulations (Cassidy et al., 2020; Eshraghi et al., 2020; Mutluer et al., 2020; Patel et al., 2020). Autism symptoms were associated with elevated levels of stress and disruption for individuals on the autism spectrum during the pandemic (Manning et al., 2021). One longitudinal study (conducted prior to the pandemic) examined autism symptom severity in children from ages 3 through 10 and reported that a majority of children on the autism spectrum (73%) plateaued in symptom improvement after entering elementary school at age 6 (Georgiades et al., 2021). Children with less symptom severity, however, followed a continuous positive trend in symptom improvement (Georgiades et al., 2021). These data indicated that children with more severe autism characteristics may have a more difficult time in traditional school settings as compared to children with less severe autism characteristics and supports the notion that the school context may contribute to mental health difficulties for some autistic children.

Present Study

Building on research to date, the purpose of this study was to examine the extent to which the type of school format that autistic children attended during the COVID-19 pandemic was associated with their internalizing symptoms, and to identify which children might be more vulnerable to experiencing higher levels of anxiety and depressive symptoms in the different school formats. Data was collected before and during the pandemic among families with a child on the autism spectrum who were part of an ongoing, longitudinal study. The first research aim tested the extent to which depressive and anxiety symptoms differed between children engaged in remote learning, children attending in-person school, and children who were homeschooled in October 2020 (which was the first opportunity to assess children after COVID-19 school restrictions were eased), controlling for pre-pandemic symptoms. We hypothesized that children on the autism spectrum attending school in-person would experience higher levels of anxiety and depressive symptoms than students attending remotely and students who were homeschooled. We did not have any a priori hypotheses about expected differences in mental health among children who attended school remotely vs. those who were homeschooled.

Second, we examined peer victimization and autism symptom characteristics prior to the pandemic as moderators of the association between school format type and children’s anxiety and depressive symptoms. With respect to autism symptom characteristics, we assessed two domains: social communication and interaction behavior (SCI) and restrictive, repetitive behavior and interests (RRBI). We hypothesized that in-person schooling would be associated with higher levels of anxiety and depressive symptoms, and remote schooling would be associated with lower levels of symptoms, among children with higher levels of pre-pandemic peer victimization as compared to lower levels of peer victimization. We also hypothesized that in-person schooling would be associated with higher levels of anxiety and depressive symptoms for children with higher levels of SCI difficulties or RBBIs as compared to those attending school remotely or who were homeschooled.

Methods

Participants

Participants for the current study were 81 children (77.8% male) on the autism spectrum and their mothers drawn from a larger multi-site study of 119 families who were part of a longitudinal study about family relationships and co-occurring internalizing symptoms in the Southern USA. Inclusion criteria for the study were that children were between the ages of 10 and 17, had an autism diagnosis from a licensed professional (verified with documentation provided by parents), the parents were married or living together for at least one year, the child lived with the parents for at least 50% of the time, and all participating family members were able to complete measures in English. Given the tasks involved in the larger study, additional inclusion criteria were that children were verbally fluent, and did not have a co-occurring diagnosis of an intellectual disability or severe mental illness (e.g., psychosis, schizophrenia). On average, the children were 14.71 years old (SD = 2.13 years) at the time of the October 2020 survey; 76.5% of the child sample was White, 9.9% was Hispanic/Latine, 6.2% was Asian or Pacific Islander, 1.2% was Black/African American, and 6.2% identified as more than one race. The median yearly family household income for participating families (reported at the beginning of the larger study) was between USD$80,001 and $100,000 (14.8%); 12.3% of families earned less than US$50,000, 17.3% reported between US$50,001 and $80,000, 28.4% reported between US$100,001 and $150,000, and 27.2% of families earned more than US$150,000 a year. One mother did not answer the question about yearly family household income.

Procedures

All procedures were approved by the Southern Methodist University Institutional Review Board. Families were recruited from the community, through online advertisements, flyers distributed to local service provides, and through the Simons Powering Autism Research (SPARK) research match, which is a national registry of families with autistic individuals interested in participating in research. Families interested in the study completed a phone screen to confirm some of the eligibility criteria. If the family and child met initial criteria, the primary caregiver and child were scheduled for an initial lab visit at one of two campus sites to corroborate the child’s autism diagnosis. During this visit, the primary caregiver completed the Social Responsiveness Scale-2nd edition (SRS-2; Constantino & Gruber, 2012; described below) and the child completed the Autism Diagnostic Observation Scale, Second Edition, Module 3 (ADOS-2; Medda et al., 2018). The ADOS-2 is a 45–60 minute semi-structed observational assessment of autism symptoms, including child’s communication, social interaction, and restricted/repetitive behaviors or interests. The PIs of the study or trained graduate students administered and scored the ADOS-2. Twenty percent of tapes (n = 24), equally divided per site, were selected and coded by a second ADOS-trained researcher on the study team to calculate percent agreement. Average percent agreement on all codes was 86.5% and the average percent agreement on algorithm items was 83%. Given other tasks children were asked to complete during the original study lab visit (not included in the present study), we also assessed whether children were verbally fluent and did not have a co-occurring diagnosis of an intellectual disability with the Peabody Picture Vocabulary Test (PPVT version 4; Dunn & Dunn, 2007) and the Differential Abilities Scale-II (DAS-II; Elliott, 2007), respectively. Children were eligible if they had a PPVT-4 standard score ≥ 70 and a DAS-II Global Conceptual Ability score ≥ 75.

If eligible, both parents and children were scheduled to participate in a 3-hour lab session (Time 1; T1). Parents provided written informed consent and children provided assent. During this session, parents independently completed questionnaires while a research assistant helped the child complete their questionnaires. Parents and children also completed other tasks (e.g., observational assessment of interparental conflict) that are not part of the current study. Families returned for a follow-up visit approximately 6-months later (Time 2; T2; N = 104 families) and completed the same protocol as their T1 visit. All T1 visits were completed between June 23, 2018 and March 14, 2020. The majority of the T2 follow-up visits occurred before the pandemic; however, 22 families completed their T2 visit during the pandemic remotely via Zoom. Families were compensated at each study visit ($85 at the eligibility visit, $85 at T1, $110 at T2; remote families received $135 at T2).

During the pandemic, we sent additional online questionnaires to parents three times to assess families’ well-being. These surveys were sent in April 2020, July 2020, and October 2020. These time points were chosen to capture family and child functioning close to the onset of the pandemic when schools transitioned to remote learning (April), during the summer (July), and around the time that schools in the State of Texas resumed options for in-person classes (October). Approximately 75% of mothers (n = 89) completed the April survey; 63% (n = 75) completed the July survey; and 69% (n = 82) completed the October survey. Participants earned a $20 (April) or $25 (July, October) Amazon gift-card for completing these supplemental online surveys. The present study used data from the mothers’ October 2020 survey (collected between October 25 and November 18) because it contained data regarding schooling, as well as the child’s depressive and anxiety symptoms at that time; one child’s data was excluded from the current study because they were no longer in school.

Pre-pandemic levels of depressive and anxiety symptoms and peer victimization measures were used from the last assessment families conducted prior to the pandemic; we used T2 data from 55 families (67.9%) and used T1 data for 26 families (32.1%). Thus, all pre-pandemic data was collected on or before March 14, 2020. The average amount of time between the pre-pandemic assessment used in analyses and the October 2020 survey was 1.23 years (SD = 0.44). Autism symptom data was taken from the initial eligibility lab session at the beginning of the study.

Measures

Pre-Pandemic Child Mental Health

Mothers reported on their child’s symptoms on the Children’s Depression Inventory-2 (CDI-2; Kovacs & MHS Staff, 2011) and the Screen for Child Anxiety Related Disorders (SCARED; Hale et al., 2011). Mothers rated how often their child displayed 17 symptoms (e.g., “My child looks sad”) during the last two weeks using a 4-point Likert scale ranging from 0 (Not at all) to 3 (Much or Most of the time). The SCARED included 41 statements about their child’s anxious behaviors during the last three months (e.g., “My child is nervous”). Mothers rated how true each statement was about their child on a 3-point Likert scale, ranging from 0 (Not true or hardly ever true) to 2 (Very true or often true). Both the CDI-2 and SCARED have shown strong construct validity and reliability in autistic samples (Carruthers et al., 2020; Ozsivadjian et al., 2017; Stern et al., 2014). In the present study, Cronbach’s alphas were 0.83 and 0.93, respectively, for the mother-reported CDI-2 and SCARED. Based on a raw cut-off score of 17 on the parent-report CDI-2 (Kovacs & MHS staff, 2011) and a clinical cut-off score of 25 on the SCARED (Birmaher et al., 1999), 34 children (42%) had potentially clinically-significant levels of depressive symptoms and 19 children (23.5%) had potentially clinically-significant levels of anxiety prior to the pandemic.

Pre-Pandemic Child Peer Victimization

Children completed the Revised Peer Experiences Questionnaire (RPEQ; Prinstein et al., 2001), which included 9 statements based on their experiences over the last six months. Children rated the frequency of negative peer interactions using a 5-point Likert scale with answer choices ranging from 1 (Never) to 5 (A few times per week). The RPEQ assesses both overt peer victimization (e.g., “a child/teen hit, kicked, or pushed me in a mean way”) and relational victimization (e.g., “a child/teen did not invite me to a party or other social event even though he or she knew I wanted to go”). The RPEQ has been shown to be reliable for use in ASD populations (Ung et al., 2016). In the current study, Cronbach’s alphas were 0.74 for the overt peer victimization subscale and 0.74 for relational victimization subscale.

Pre-Pandemic Child Autism Symptom Characteristics

Children’s autism symptom characteristics were assessed with the Social Responsiveness Scale, 2nd edition (SRS-2; Constantino & Gruber, 2012) at the eligibility visit at the beginning of the study. The SRS-2 is a 65-item questionnaire, which assesses children’s behavior over the past 6 months in areas regarding social awareness, social communication, social cognition, social motivation, and rigid and repetitive mannerisms. This scale was completed by the parent identifying as the child’s primary caregiver (84.2% mothers, 10.5% fathers, 4.8% both parents completed the SRS-2 together). Items were rated on a 4-point Likert scale, ranging from 1 (Not true) to 3 (Almost always true). Cronbach’s alpha in the current sample was .91 for the social communication and interaction subscale and .80 for the restrictive and repetitive mannerisms subscale.

Type of Schooling during Pandemic

Mothers reported on their child’s schooling format before the pandemic and at the time of the survey (October 2020). The survey item regarding schooling format included the following answer choices: fully in-person, hybrid-method (attend school partially in-person and partially remote), fully remote/e-learning, educational pod/small group, homeschooling, or other. Three groups were created: (1) in-person, which included fully in-person or hybrid methods, (2) remote, and (3) homeschooled. We included hybrid methods in the in-person school category because it included an in-person component and because there were not enough students in hybrid methods to warrant its own category.

Child Mental Health During the Pandemic

Mothers completed the Children’s Depression Inventory-2 (CDI-2; Kovacs & MHS Staff, 2011). Cronbach’s alpha of the CDI-2 in the current sample was 0.81. Mothers also completed the Parent Proxy Anxiety questionnaire from the NIH PROMIS toolbox (Irwin et al., 2010), which included eight statements about their child’s anxious behaviors during the last seven days (e.g., “my child felt nervous”). Mothers rated how often their child displayed anxious behaviors using a 5-point Likert scale, ranging from 1 (Never) to 5 (Almost always). Ratings across items were summed to create a total score, with higher scores reflecting higher levels of anxiety symptoms. Cronbach’s alpha in the current sample was 0.92. Based on a raw clinical cut-off score of 17 on the parent-report CDI-2 (Kovacs & MHS Staff, 2011), 17 children (21%) had potentially clinically-significant levels of depressive symptoms in October 2020. Based on T-score ≥70 on the anxiety questionnaire, 5 children (6.2%) had potentially clinically-significant levels of anxiety; 16 children (19.8%) had a T-score between 60 and 70, indicating borderline-clinical levels of anxiety.

Analysis Plan

Our analysis plan was pre-registered on the Open Science Framework (https://osf.io/63jng). In preliminary analyses, we tested for differences between families who did and did not complete the October 2020 survey, reported descriptive information about children’s school format type before and during the early months of the pandemic, and tested bivariate correlations between study variables. We also considered child age, sex, and study site as potential covariates by testing their associations with the key study variables.

Our main study hypotheses were tested using multiple linear regressions, in which we predicted children’s anxiety or depressive symptoms as a function of school type. We collapsed the three school type variables (in-person, remote, home-schooled) into two dummy coded variables for regression analyses, with in-person school as the reference category. We controlled for pre-pandemic levels of anxiety or depressive symptoms and the number of days between when the pre-pandemic data was collected and when the October survey data was collected.

Because the COVID-19 surveys were not planned as part of our original study, and we therefore had a smaller sample size than our original study, we conducted sensitivity analyses to determine the minimum detectable effect size based on our October 2020 sample size using the G*Power v. 3.1.9.6 function for a linear multiple regression (Faul et al., 2007). We set power to 80% and our Type I error rate to .05. We also used Cohen’s (1988) guidelines, in which an f² ≥ 0.02 indicates a small effect, f² ≥ 0.15 indicates a medium effect, and f² ≥ 0.35 indicates a large effect. Results from this analysis indicated that we were powered to detect a significant effect of at least one predictor (e.g., detect a difference between at least two school format groups ) in a multiple regression analysis with four total predictors that was between small to medium in size (f² = 0.10).

Results

Preliminary and Descriptive Analyses

We compared families who completed the October 2020 survey (n = 82) to those that did not (n = 37) to test whether there were any systematic differences on their pre-pandemic variables. There were no significant differences based on child age, t(117) = 0.29, p = .78, or sex, χ²(1) = 1.04, p = .31, between families whose mother completed the October 2020 survey and those that did not. We also did not find any significant differences in mother-reported child anxiety or depressive symptoms, child-reported overt or relational victimization, or child ASD symptom severity pre-pandemic between families whose mother completed the October 2020 survey and those that did not (all p > .05; Supplemental Table 1). Thus, there did not appear to be any systematic bias in which mothers completed the October 2020 survey.

In October 2020, 44.4% (n = 36) of children attended school in-person; 4.9% (n = 4) attended school in a hybrid model (range: 2–20 hours per week in-person), 33.3% (n = 27) attended school fully remote, and 16% (n = 13) were homeschooled. Prior to March 2020, 85.2% (n = 69) of participants attended school in-person, 13.6% (n = 11) were homeschooled, and one child was completing an online public education program. Nine children were homeschooled both before and during COVID-19; four children that were homeschooled during the pandemic attended school in-person before the pandemic. Twenty-four adolescents transitioned from in-person schooling prior to the pandemic to remote learning during the next school year (i.e., fall of 2020). Thus, approximately 44% of children (n = 36) experienced a change in their school format in October 2020, whereas 55.6% (n = 45) remained in the same school format type.

Descriptive statistics and bivariate correlations for all study variables are present in Table 1. Anxiety levels before the pandemic and anxiety during the pandemic were moderately, positively correlated, r = .40, p < .001. Anxiety levels before the pandemic were also positively correlated with depressive symptoms, r = .30, p = .007. Depressive symptoms prior to the pandemic were positively correlated with anxiety, r = .25, p = .024, and depressive, r = .63, p = < .001, symptoms during the pandemic. Child age was not associated with either their anxiety or depressive symptoms levels. Children’s SCI difficulties were positively correlated with anxiety and depressive symptoms both prior to and during the pandemic, and children’s RRBIs were positively associated with anxiety symptoms during the pandemic, r = .24, p = .028.

Table 1.

Descriptive Statistics and Bivariate Correlation of Study Variables

	1.	2.	3.	4.	5.	6.	7.	8.
1. Child Sex	--
2. Child Agea	.06	--
3. Site	−.19†	.10	--
4. Time between pre-Covid and Octsurveys	−.05	.09	.01	--
Pre-Pandemic Variables
5. Child Anxiety	.15	.04	.17	.10	--
6. Child Depressive Symptoms	.07	.01	.03	.03	.30**	--
7. Peer Overt Victimization	−.06	−.17	−.07	−.07	−.09	.14	--
8. Peer Relational Victimization	.11	−.05	−.02	.04	−.08	.13	.70***	--
9. SRS- SCI	−.13	−.08	−.01	.08	.49***	.30**	.02	.003
10. SRS- RRBI	−.24*	−.12	−.18	.16	.16	.15	.10	−.002
October 2020 Variables
11. Change in school format (yes = 1)	.06	.09	−.06	−.04	.03	−.03	−.24*	−.21†
12. School: In-person vs not in-person	−.01	.08	.04	−.08	−.01	−.08	.17	.20†
13. School: Remote vs. not remote	.06	−.04	−.04	−.07	.10	.04	−.20†	−.23*
14. School: Homeschool vs. not homeschool	−.07	−.05	−.01	.20†	−.11	.06	.03	.03
15. Child Anxiety	.21†	−.16	.08	−.01	.40***	.25*	.18	.13
16. Child Depressive Symptoms	.14	−.15	.07	.06	.21†	.63***	.04	.07
M	--	14.71	--	449.35	18.58	15.69	5.81	8.61
SD	--	2.13	--	162.09	13.76	6.61	2.84	4.17
	9.	10.	11	12.	13.	14.	15.	16.
1. Child Sex
2. Child Agea
3. Site
4. Time between pre-covid and Oct surveys
Pre-Pandemic Variables
5. Child Anxiety
6. Child Depressive Symptoms
7. Peer Overt Victimization
8. Peer Relational Victimization
9. SRS- SCI	--
10. SRS- RRBI	.74***	--
October 2020 Variables
11. Change in school format (yes = 1)	.06	.02	--
12. School: In-person vs not in-person	−.09	−.14	−.66***	--
13. School: Remote vs. not remote	.13	.06	.79***	−.72***	--
14. School: Homeschool vs. not homeschool	−.04	.11	−.12	−.44***	−.31**	--
15. Child Anxiety	.38***	.24*	−.03	.02	−.01	−.02	--
16. Child Depressive Symptoms	.26*	.13	.04	−.22†	.17	.07	.41***	--
M	81.85	20.46	--	--	--	--	15.95	12.37
SD	19.60	6.03	--	--	--	--	6.46	6.95

Note. N = 81.

^aChild age in October 2020; SRS = Social Responsiveness Scale-2; SCI = social communication and interaction difficulties score; RRBI = restrictive and repetitive behaviors and interest score. Sex coded 1 = male, 2 = female;

^†p < .10,

^*p < .05,

^**p <.01,

^***p <.001

Independent sample t-tests indicated that boys reported significantly more RRBIs as compared to girls, t(79) = 2.24, p = .028. Children who changed school formats in October 2020 (coded yes/no), compared to how they attended school prior to the pandemic, reported significantly higher levels of pre-COVID overt victimization, t(77) = 2.13, p = .036; there were no other significant mean differences on the other variables. We also tested for any differences between the three school types based on children’s pre-COVID mental health, peer experiences, and autism behaviors (Supplemental Table 2). There were no significant differences in children’s pre-COVID levels of depressive or anxiety symptoms or their autism behaviors between the three school type groups. We also did not find any significant mean differences between the three school type groups in pre-COVID levels of overt or relational victimization.

School Type as a Predictor of Mental Health

Overall, the regression model predicting mother-reported child anxiety symptoms was statistically significant, F(4, 76) = 3.86, p = .007, and accounted for 17% of the variance in anxiety symptoms (Table 2, Model 1). The results indicated that the only significant predictor of child anxiety symptoms during the pandemic were pre-pandemic levels of anxiety symptoms, b = 0.20, SE = 0.05, p <.001, β = 0.42. Contrary to hypotheses, however, there was no significant difference in children’s level of anxiety symptoms based on school type.

Table 2.

Results from Multiple Regressions testing School Type as a Predictor of Children’s Anxiety and Depressive Symptoms

	Model 1: Anxiety Symptoms				Model 2: Depressive Symptoms
	b (SE)	p	B	CI	b (SE)	p	B	CI
Intercept	13.47 (2.19)	<.001	--	[9.11, 17.83]	0.43 (2.29)	.85	--	[−4.13, 4.99]
Days between pre-pandemic assessment and Oct 2020 survey	−0.002 (0.004)	.60	−0.06	[−0.01, 0.006]	0.001 (0.004)	.73	.03	[−0.006, 0.009]
Pre-pandemic Symptoms	0.20 (0.05)	<.001	0.42	[0.096, 0.30]	0.65 (.09)	<.001	0.62	[0.47, 0.84]
School 1 (In-person vs. Remote)	−0.64 (1.50)	.67	−0.05	[−3.63, 2.35]	2.55 (1.34)	.06	0.17	[−0.112, 5.21]
School 2 (In-person vs. Homeschool)	0.36 (1.97)	.86	0.021	[−3.57, 4.29]	1.63 (1.75)	.35	0.09	[−1.85, 5.11]
R 2	.17				.43

Note. N = 81. School 1 and School 2 are dummy-codes representing school type, with in-person as the reference category.

The overall regression model for mother-reported child depressive symptoms was also statistically significant, F(4, 76) = 14.43, p < .001, and accounted for 43% of the variance in depressive symptoms (Table 2, Model 2). Pre-pandemic depressive symptoms significantly predicted pandemic levels of depressive symptoms, b = 0.65, SE = 0.09, p <.001, β = 0.62. There was a marginal association between school type (in-person vs. remote) and children’s depressive symptoms, b = 2.55, SE = 1.34, p = .06, β = 0.17, suggesting that children attending school in-person (M = 10.90, SD = 7.18) had marginally lower levels of depressive symptoms as compared to children attending school remotely (M = 14.04, SD = 6.82). However, this effect did not reach statistical significance.

Post-hoc, Exploratory Analyses

Given the small sample size for children who were home-schooled, we ran a post-hoc, exploratory analysis (i.e., not pre-registered), in which we collapsed the remote and homeschool children into one group. Thus, analyses compared children who attended school in-person (n = 41) to those who engaged in schooling from home (n = 40). School format did not significantly predict children’s anxiety or depressive symptoms; however, pre-pandemic levels of symptoms emerged as the only significant predictors (Supplemental Table 3). As an additional post-hoc analysis, we also tested whether a change in school format in October 2020 (relative to their schooling prior to the pandemic) predicted children’s anxiety or depressive symptoms. Experiencing a change in school format, however, did not predict children’s anxiety, b = −0.52, SE = 1.35, p = .70, nor their depressive symptoms, b = 0.82, SE = 1.22, p = .51; pre-pandemic symptoms again were the only significant predictors in these models.

Peer Victimization and ASD Symptoms Characteristics as Moderators

Two sets of multiple linear regressions were performed to examine (a) overt and relational peer victimization and (b) SCI difficulties and RRBIs as moderators of the association between school type and pandemic levels of mental health symptoms. We controlled for pre-pandemic levels of anxiety or depressive symptoms and the number of days between when the pre-pandemic data was collected and when the October 2020 survey data was collected. The results for these models are presented in Supplemental Tables 4–7. Neither peer victimization nor RRBIs emerged as a significant moderator in any of the models. One significant interaction emerged when examining SCI difficulties, b = −0.15, SE = 0.07, p = .027. Simple slope analyses showed that for children with lower SCI difficulties, mothers’ reported lower levels of depressive symptoms among children who were attending in-person school as compared to those in remote schooling (p = .006). Among those with higher levels of SCI difficulties, there was no difference in levels of depressive symptoms among those attending in-person vs remote schooling (p = .84). This interaction, however, did not remain significant once accounting for the number of interactions tested using Benjamini-Hochberg’s (Benjamini & Hochberg, 1995) false discovery rate correction. As an additional post-hoc, exploratory analysis, we re-tested children’s SCI difficulties and RRBIs as moderators of the association between school type and pandemic levels of mental health symptoms using the ADOS-2 social affect and RRBI scores (Supplemental Tables 8 and 9); none of the interactions, however, were significant.

Discussion

A number of qualitative and quantitative studies have shown that the pandemic had a negative impact on the mental health of autistic children (Castro-Kemp & Mahmud, 2022; Mutluer et al., 2020; Oomen et al., 2021; Vasa et al., 2021) and disruptions to children’s school setting has been cited as a particular challenge during the pandemic (Asbury et al., 2021; Ozsivadjian et al., 2023). The current study tested the extent to which children’s anxiety and depressive symptom levels during the pandemic were associated with how they attended school (in-person, remote, homeschool) in October 2020. Building on a growing body of research examining the effects of the pandemic (Lee et al., 2021; Shorey et al., 2021), and educational settings specifically (e.g., Baten et al., 2022; Simpson & Adams, 2022; Ozsivadjian et al., 2023), on autistic children, a novel feature of our study was the longitudinal design which allowed us to account for children’s pre-pandemic mental health symptoms. Contrary to our hypotheses, we did not find evidence to support differences in levels of anxiety and depressive symptoms between children who attended school in-person as compared to those who attended school remotely or were homeschooled. Rather, pre-pandemic levels of anxiety and depressive symptoms consistently predicted children’s level of symptoms in October 2020. Thus, our findings suggest that autistic children who experienced mental health difficulties during the pandemic were likely struggling before the pandemic.

Given our sample only included children who were verbally fluent and did not have an intellectual impairment, it is possible that they had a better understanding of the COVID-19 pandemic and rationale for local closures. For example, emerging research suggests that those on the autism spectrum with more limited cognitive functioning may have struggled to understand COVID-19 regulations (Cassidy et al., 2020; Eshraghi et al., 2020; Mutluer et al., 2020; Patel et al., 2020). Thus, children in our sample may have had the cognitive capabilities to understand the reasoning behind COVID-19 school protocols, and were therefore less affected by different schooling decisions. Children attending school remotely also may have understood that this was a temporary solution that was necessary at that particular time during the pandemic, and they would soon return to in-person schooling.

Although not statistically significant, mothers reported marginally lower levels of depressive symptoms when their child attended school in-person rather than remotely. This finding was counter to our hypothesis as well as recent studies which suggested remote schooling might reduce mental health difficulties among students (Fegert et al., 2020; Hansen et al., 2022; Mumbardó-Adam et al., 2021; Reicher, 2020). One potential explanation for this marginal finding is that in-person school allowed students to return to familiar routines and, for some children, resume in-school services (Reicher, 2020). A qualitative study found that both autistic and non-autistic youth similarly cited a lack of social connection as a challenge of remote schooling (Ozsivadjian et al., 2023); thus, in-person schooling may have also facilitated greater social connection thereby improving mental health. Alternatively, mothers whose children attended school in-person may not had as much opportunity to observe their child’s mental health symptoms compared to parents whose children were at home. Although we tested school type as a predictor of children’s mental health, another possibility is that families whose child had lower levels of depressive symptoms may have been more likely to choose in-person schooling. Given this finding was not statistically significant and counter to hypotheses, caution should be used interpreting this finding.

Overall, we did find evidence for moderation by peer victimization or children’s RRBIs. One significant interaction emerged with children’s SCI difficulties; however, this interaction was no longer significant once accounting for multiple moderation tests. Our findings are inconsistent with prior research which has shown associations between peer victimization and increased emotional distress among children on the autism spectrum (Adams et al., 2016; Storch et al., 2012), although notably these studies were conducted prior to the pandemic and in traditional educational settings. Additionally, in a pre-pandemic study, Georgiades et al. (2021) reported that children with more severe autism symptoms typically found school much more difficult than children with less severe autism symptoms. Notably, children’s reports of peer victimization in our sample were positively skewed, suggesting low levels of peer victimization experiences. Additionally, given our inclusion criteria for the larger study, we excluded children on the spectrum with higher support needs. Thus, the general lack of moderation effects may be due to a restriction in range. Although we were sufficiently powered for our main analyses testing school type as a predictor of children’s internalizing symptoms, we were not sufficiently powered to test for interaction effects. Thus, results from the moderation analyses should be interpreted with caution, and replication is needed.

Limitations of our study provide directions for future research. First, our sample was comprised primarily of White/European American, high socio-economic status, two-parent families. Consistent with other studies, the child sample was also predominately male. Thus, our findings may not generalize to children from historically minoritized groups, females, or those living in single parent households. Notably, in a population-based study, Hawrilenko and colleagues (2021) reported that remote schooling during the pandemic was most detrimental to mental health among children who were Black or Latine and from lower-income households. Thus, the demographics of our sample, including higher-resourced families, may have affected our ability to detect the effects of different school formats on children’s mental health.

Second, as noted, our sample excluded children with higher support needs, who may have had different experiences with COVID-19-related school formats. Moreover, Ozsivadjian et al.’s (2023) qualitative study highlighted that the effect of schooling on autistic children’s emotional well-being is nuanced; whereas some children benefited from reduced sensory demands afforded by remote learning and reported a decrease in anxiety, other youth reported that the lack of social connection during remote learning was challenging. Thus, research with more diverse samples is needed to provide a more comprehensive investigation of the extent to which different school format types adopted during the pandemic were associated with autistic children’s mental health, as well as research into the potential mechanisms by which different school formats were associated with worsened or improved mental health outcomes.

Third, we relied on mothers’ reports of their child’s symptoms of anxiety and depressive symptoms prior to and during the pandemic. Informant discrepancies between mothers and children when reporting on the child’s internalizing symptoms are common in the literature (see De Los Reyes & Kazdin, 2005). In a sample of autism children and their parents, Ooi et al., (2016) found greater parent-child discrepancies when reporting on less observable symptoms. Moreover, research suggest that mothers’ own mental health and stress, which was also likely affected by the pandemic, may influence their reports of their child’s mental health (Briggs-Gowan et al., 1996; Ooi et al., 2016; Truetler & Epkins, 2003). Qualitative research also indicated that remote schooling may have been particularly frustrating for parents to navigate (Asbury et al., 2021; Baten et al., 2022), which may have spilled over to negatively affect their reporting of their child’s symptoms. Thus, it is possible that mothers may have under- or over-reported their child’s symptoms. Relatedly, we used a different measure of children’s anxiety symptoms during the pandemic, thus, precluding a direct comparison of levels of anxiety pre- and during the pandemic. However, even with two different measures, we did find significant associations in levels of anxiety symptoms across the two assessments.

Fourth, our data was collected in October 2020, shortly after in-person school options were available to families in our study. Thus, there might have been greater differences in emotional distress between children in different school formats later in the school year, as participants became more acclimated to, or fatigued of, one school type or another. Additionally, school decisions during the Fall of 2020 were controversial in certain parts of the U.S., with protests surrounding the lack of in-person school options as well as resistance to masks in the classroom. Therefore, it is possible our findings are specific to the local region of this study and may not necessarily generalize across the US.

Our findings indicate that school format type did not significantly predict symptoms of anxiety or depression among children on the autism spectrum during the early months of the COVID-19 pandemic. The most robust predictor of children’s symptoms were their pre-pandemic levels of anxiety or depression. Considering educators’ and parents’ concerns regarding the impacts of COVID-19-schooling on children, these findings should provide some relief. Messages of the harmful effects of remote schooling were highlighted by the media (e.g., Wood & Mascarenhas, 2021), and research with neurotypically-developing children provided some preliminary support for that view (Hawrilenko et al., 2021; Verlenden et al., 2021). A limitation of those studies, however, was that they did not control for children’s mental health prior to the pandemic, and in general, there is limited longitudinal research on the impact of the pandemic for autistic individuals as compared to other populations (Kunzler et al., 2023). A more recent study using national data of 12–18 year olds before and during the pandemic found that risk for teen suicide dropped during periods of remote schooling during the pandemic, and increased during the return to in-person schooling (Hansen et al., 2022). Our findings suggest that, at least for children on the autism spectrum, remote schooling was not significantly associated with higher levels of anxiety and depressive symptoms compared to other schooling types. Future research should consider and examine additional potentially confounding factors associated with the pandemic and school modalities (e.g., level of stress within the family, loss of family members/bereavement, children’s social connectedness, schools’ pandemic safety precautions, preparedness of teacher for remote delivery of class content) before we can draw more cogent conclusions regarding potential effects of in-person vs remote schooling for children on the autism spectrum during the pandemic.

Highlights.

• Remote vs. in-person school format during pandemic did not predict children’s symptoms

• Pre-pandemic internalizing symptoms positively predicted symptoms during pandemic

• Results underscore need to account for pre-pandemic child functioning when drawing conclusions about pandemic-level effects