Participants included parents who had at least one child aged between 5 and 18 years with a formal ASC diagnosis (i.e., autism spectrum disorder, autistic disorder, Asperger's disorder, and PDD-NOS) or with typical development. All families resided in Malaysia during the COVID-19 pandemic. In total, 392 parents started the survey, 133 participants were excluded due to incomplete surveys. A further 95 participants were excluded as their children were out of the age range. The final sample consisted of 134 participants, with 30 participants excluded as their children were not “typically developing” (i.e., parent-reported a child with an intellectual disability or other developmental difficulties).

The final sample comprised parent-child dyads for 72 children on the autism spectrum (54 mother-child dyads, 18 father-child dyads) and parent-child dyads for 62 TD children (53 mother-child dyads, nine father-child dyads). Demographic information for parents and children of both groups are presented in Tables 1, 2. More than half of the parents obtained at least a bachelor's degree. Almost 70% of parents are working parents. Among the working parents, almost 70% of them worked from home and around 7% of them stopped working during the MCO. During the CMCO, almost half of the working parents went back to work regularly.

The age of the children ranged from 5 years 4 months to 17 years 0 month. All diagnoses were performed by a registered mental health professional or developmental pediatrician per parent report. The Social Responsive Scale-2 [SRS-2, (32)] was utilized to help confirm the ASC diagnosis, with T-scores ≥ 60 used as the inclusion criteria for the ASC group and exclusion criteria in the TD group.

Eleven of the children in the ASC group had comorbid ADHD, one child on the autism spectrum had epilepsy, one child on the autism spectrum had both ADHD and epilepsy, and three children on the autism spectrum had ADHD and learning difficulties based on the parent reports. Nineteen parents of children on the autism spectrum reported their children having intellectual disability. More than half of all the children (ASC and TD) study in government schools. More than 80% of children on the autism spectrum and half of the typically developing children attend morning schools while the other half of the typically developing children attend both morning and afternoon schools every school day. Most of all the children (ASC and TD) attended online classes during the national lockdown (MCO and CMCO). Before the lockdown, more than half of the children attended additional activities (such as tuition classes or therapy sessions) but canceled these additional activities during the lockdown.

No significant difference was found in parental age between the ASC and TD groups, t₍₁₃₀₎ = 0.46, p = 0.650. Moreover, there was no significant difference in child's age between the ASC and TD groups, t₍₁₃₁₎ = 0.732, p = 0.465. On the contrary, there was a significant difference in child's gender between the ASC and TD groups, χ(1)2 = 4.332, p = 0.037.

Ethical approval for the present study was obtained from the Monash University Human Research Ethics committee (MUHREC; Project 24673). The researchers reached out to ASC-related organizations and schools in Malaysia through initiating contact via email and phone calls to center administrators (e.g., Genius Kurnia). Several organizations responded and provided written permission, agreeing to post online advertisements. On top of that, digital posters were posted in several online parent support groups (e.g., Autisme Malaysia and Autism Parents Support Group Malaysia). Parents of children on the autism spectrum also helped with recruitment and referred parents of both groups. Interested parents then clicked on the link attached in the digital posters and completed the survey online.

Due to the COVID-19 pandemic, participants (parents) were provided an online written informed consent and asked to complete an online survey through the Monash University Qualtrics site. Both the English and the Malay translated versions of the questionnaires were included in the online survey. The data collection period started from late July 2020 to late October 2020. Most of the participants completed the online survey in August 2020.

The participants were asked to complete their own as well as their child's demographics. For parents of children on the autism spectrum, they answered extra demographic questions regarding their children's diagnosis. Then, participants completed a list of questions regarding the COVID-19 situation. Next, they were asked to fill the survey asking about their child's behaviors and psychological distress, as well as their own psychological distress and well-being. Importantly, they responded to these questions relative to their lives during the lockdown (18th March to 9th June 2020), and retrospectively before the lockdown.

As a token of appreciation, a RM30 ($7 USD) e-shopping voucher or e-wallet top-up was given to the participants upon completion of the online survey. After the survey completion, participants were also offered the opportunity to participate in a qualitative semi-structured interview as this quantitative study was part of a broader mixed-methods longitudinal project examining the impact of COVID-19 on children on the autism spectrum in Malaysia.

The data was analyzed using the IBM SPSS version 25 for descriptive and inferential analysis. The data was examined for univariate outliers, normality, multicollinearity, and homoscedasticity prior to the main analysis. All univariate outliers with SD > ± 3 above the mean in each group were winsorised using the formula, X = (3 × standard deviation) + mean (40). Multivariate outliers were not removed as there was no substantive difference between the analyses with and without the multivariate outliers.

Multiple two-way mixed-model analysis of variances (MM-ANOVAs), between group [diagnosis (ASC, TD)] X within group [lockdown (before, during)], were utilized to compare the mean average of the child's behaviors (as measured by Conners'-3 Inattention subscale, Hyperactivity/Impulsivity subscale, and Conners'-3 Global Index), child's psychological distress (as measured by Conners'-3 Anxiety and Depression subscales as well as PSSC), and parent's psychological distress (as measured by PSSA and the Depression and Anxiety subscales from DASS-21) and well-being (as measured by SPANE). Bonferroni corrections applied to p-values were used to reduce the risk of type I error. Hence, the adjusted alpha value was set at 0.005 (10 comparisons: alpha = 0.05/10). Levene's test statistic was used to test the assumption of homogeneity of variance. The assumption for homogeneity of variance was violated for some of the MM-ANOVAs (p < 0.001). However, the analysis of variance was robust to violations of this assumption as the size of the groups was reasonably similar [i.e., ASC group/TD group = 1.16; (41)].

Pearson's correlation coefficients were performed to investigate the relationship of child and parent variables mid-lockdown. Prior to the correlation analyses, statistical assumptions such as normality, linearity, and homoscedasticity were assessed. Pearson's correlations were used to understand any relations among the mid-lockdown child and parent variables in both ASC and TD groups. Bonferroni corrections applied to p-values were used to reduce the risk of type I error. Hence, the adjusted alpha value for the ASC group was set at 0.0045 (11 comparisons; alpha = 0.05/11) and the adjusted alpha value for the TD group was set at 0.005 (10 comparisons; alpha = 0.05/10).

Results of COVID-19 related questions are reported in Table 3. COVID-19 positivity was not reported among the parents and their children. 11.3% of the parents reported to know someone who has been hospitalized due to COVID-19 and 2.3% of the parents reported to experience people who they know passed away due to the virus. Almost all the parents reported making changes to their daily lifestyle due to the pandemic, with more than half of them (57.25%) self-quarantining most of the time. Around 5% of all the parents experienced job loss due to the pandemic. Twenty-five percent of the parents with children on the autism spectrum and 13% of parents with typically developing children reported to have lost income due to the pandemic. Around a quarter of parents (24.7%) described their household money situation as “have to cut back” since the pandemic.

Parents from both groups were equally concerned about the pandemic. However, although not statistically significant, parents of children on the autism spectrum were more concerned that their child might be infected by the virus than parents of TD children. Since hearing about the pandemic, parents in the ASC group reported to have marginally more elevated household conflict than the parents in the TD group, t₍₁₃₂₎ = −1.983, p = 0.049.

Based on the parent reports, all TD children knew about the pandemic while only 69.4% of children on the autism spectrum knew about the pandemic. Moreover, TD children were significantly more concerned about the pandemic than children on the autism spectrum, t₍₁₃₂₎ = −2.224, p = 0.024. The lockdown was equally disruptive to the children's daily routine. However, children on the autism spectrum experienced significantly more difficulties in following the modified routines than the TD children, t₍₁₃₂₎ = 3.973, p < 0.001.

Multiple two-way MM-ANOVAs were conducted to compare the effect of diagnosis and lockdown on the change in time spent on electronic devices, time spent with parents, and time spent on physical activities. Based on the MM-ANOVA, there is a significant lockdown effect on the time children spent on electronic devices, F_(1,124) = 135.14, p < 0.001, ηp2 = 0.52. However, there's no group difference in time spent on electronic devices, F_(1,124) = 0.03, p = 0.856, ηp2 = 0.00. On top of that, the interaction effect between lockdown and diagnosis group was not significant, F_(1,124) = 0.02, p = 0.882, ηp2 = 0.00.

Similarly, the MM-ANOVA reveals a significant lockdown effect on children's time spent with their parents, F_(1,120) = 114.32, p < 0.001, ηp2 = 0.49. However, there is no significant group difference in time spent with parents. The interaction effect between lockdown and diagnosis group was significant, F_(1,120) = 6.61, p = 0.011, ηp2 = 0.05.

The final MM-ANOVA shows no significant lockdown [F_(1,122) = 0.18, p = 0.67, ηp2 = 0.001] and interaction [F_(1,122) = 0.29, p = 0.591, ηp2 = 0.002] effects on children's time spent on physical activities. However, there is a significant diagnosis group effect in time spent on physical activities, F_(1,122) = 8.32, p = 0.005, ηp2 = 0.064.

There were no significant gender differences (boy/girls) in all the child subscales (p > 0.05). Among the ASC group, paired samples t-tests were conducted to compare the mean scores of SRS-2 and its subscales pre- and mid-lockdown. The paired samples t-tests revealed that there was no significant difference in the mean scores of the SRS-2 and its subscales pre- and mid-lockdown (all p > 0.05; see Table 4).

Multiple two-way MM-ANOVAs were conducted to compare the effect of diagnosis and lockdown on child's behaviors and psychological distress as well as parent's psychological distress and well-being. As shown in Table 5, the interaction effect between lockdown and diagnosis was not significant for all dependent variables (p > 0.005).

On the other hand, significant main effects for diagnosis group were identified for child's behavior [CPRS-3 Inattention [F_(1,132) = 79.78, p < 0.001, ηp2 = 0.38]; CPRS-3 Hyperactivity/Impulsivity [F_(1,132) = 72.50, p < 0.001, ηp2 = 0.36]; CPRS-3 Global Index [F_(1,132) = 93.48, p < 0.001, ηp2 = 0.42]], child's psychological distress [CPRS-3 Anxiety [F_(1,132) = 81.67, p < 0.001, ηp2 = 0.38]; CPRS Depression [F_(1,132) = 65.18, p < 0.001, ηp2 = 0.33]; and PSSC [F_(1,132) = 68.32, p < 0.001, ηp2 = 0.34]] as well as parent's psychological distress [DASS Depression [F_(1,132) = 15.75, p < 0.001, ηp2 = 0.11]; DASS Anxiety [F_(1,132) = 13.90, p < 0.001, ηp2 = 0.10]], and parent's well-being [SPANE [F_(1,132) = 15.14, p < 0.001, ηp2 = 0.10]]. On the other hand, significant main effect for diagnosis group was not identified for PSSA, F_(1,132) = 5.28, p = 0.023, ηp2 = 0.04.

Significant main effects for lockdown (pre- vs. mid-lockdown) were found for child's behavior [CPRS-3 Inattention [F_(1,132) = 14.17, p < 0.001, ηp2 = 0.10]; CPRS-3 Hyperactivity/Impulsivity [F_(1,132) = 8.75, p = 0.004, ηp2 = 0.06]; CPRS-3 Global Index [F_(1,132) = 15.28, p < 0.001, ηp2 = 0.10]], child's anxiety [CPRS-3 Anxiety [F_(1,132) = 15.96, p < 0.001, ηp2 = 0.11]], and parent's psychological distress and well-being [PSSA [F_(1,132) = 9.48, p = 0.003, ηp2 = 0.06], DASS Depression [F_(1,132) = 9.14, p = 0.003, ηp2 = 0.07]; SPANE [F_(1,132) = 10.09, p = 0.002, ηp2 = 0.07]]. On the other hand, significant main effects for lockdown were marginally non-significant for CPRS-3 Depression [F_(1,132) = 7.64, p = 0.007, ηp2 = 0.06], PSSC [F_(1,132) = 7.11, p = 0.009, ηp2 = 0.05] and DASS Anxiety [F_(1,132) = 7.90, p = 0.006, ηp2 = 0.06; see Table 5].

Pearson's correlations among variables mid-lockdown were conducted for the ASC and TD groups separately (see Tables 6, 7). Pearson's correlations among the ASC group revealed significant correlations (p < 0.0045) among all child behavior indicators (CPRS-3 Inattention, CPRS-3 Hyperactivity/Impulsivity and CPRS Global Index) and psychological distress indicators (CPRS-3 Anxiety, CPRS-3 Depression, and PSSC). PSSA was significantly correlated with CRPS-3 Inattention [r₍₇₂₎ = 0.396, p = 0.001], CPRS-3 Hyperactivity/Impulsivity [r₍₇₂₎ = 0.408, p < 0.001], CPRS-3 Global Index [r₍₇₂₎ = 0.448, p < 0.001], SRS-2 [r₍₇₂₎ = 0.333, p = 0.004], and PSSC [r₍₇₂₎ = 0.503, p < 0.001]. Moreover, PSSA was marginally correlated with CPRS-3 Hyperactivity/Impulsivity [r₍₇₂₎ = 0.325, p = 0.005], CPRS-3 Global Index [r₍₇₂₎ = 0.318, p = 0.006], and SRS-2 [r₍₇₂₎ = 0.323, p = 0.006]. DASS Depression was significantly correlated with PSSC [r₍₇₂₎ = 0.359, p = 0.002], PSSA [r₍₇₂₎ = 0.663, p < 0.001], and DASS Anxiety [r₍₇₂₎ = 0.770, p < 0.001]. Furthermore, DASS Depression was marginally correlated with CPRS-3 Global Index [r₍₇₂₎ = 0.320, p = 0.006]. DASS Anxiety was significantly correlated with PSSC [r₍₇₂₎ = 0.351, p = 0.003] and PSSA [r₍₇₂₎ = 0.611, p < 0.001]. SPANE was significantly correlated with CPRS-3 Inattention [r₍₇₂₎ = −0.399, p = 0.001], CPRS-3 Hyperactivity/Impulsivity [r₍₇₂₎ = −0.381, p = 0.001], CPRS-3 Global Index [r₍₇₂₎ = −0.475, p < 0.001], SRS-2 [r₍₇₂₎ = −0.357, p = 0.002], PSSC [r₍₇₂₎ = −0.458, p < 0.001], PSSA [r₍₇₂₎ = −0.770, p < 0.001], DASS Anxiety [r₍₇₂₎ = −0.672, p < 0.001], and DASS Depression [r₍₇₂₎ = −0.775, p < 0.001]. Moreover, SPANE was marginally correlated with CPRS-3 Depression [r₍₇₂₎ = −0.326, p = 0.005].

Similarly, Pearson's correlations among the TD group revealed significant correlations (p < 0.005) among all child behavior indicators (CPRS-3 Inattention, CPRS-3 Hyperactivity/Impulsivity and CPRS Global Index) and psychological distress indicators (CPRS-3 Anxiety, CPRS-3 Depression, and PSSC). On the other hand, none of the parent variables (PSSA, DASS Depression, DASS Anxiety, and SPANE) was significantly correlated with the child variables (p > 0.005).

In examining the first hypothesis, no significant difference was found in the severity of ASC symptoms of children on the autism spectrum pre- and mid-lockdown. This finding differs from what has been reported in other COVID-19 studies [e.g., (13, 14, 19)], where parents have been frequently reporting their children on the autism spectrum displaying more ASC symptoms during the lockdown. Based on the preliminary interviews in the broader project, some children on the autism spectrum actually enjoyed staying at home during the first 2 months of the lockdown. This may help explain why the ASC severity did not elevate during the lockdown, as the survey was distributed relatively at the early stage of the lockdown.

The second hypothesis, which states that parents of children on the autism spectrum and TD children would report more behavioral problems and higher psychological distress mid-lockdown than pre-lockdown, with children on the autism spectrum displaying more behavioral problems and psychological distress than TD children, was partially supported. Children of both diagnosis groups displayed significantly more inattention, hyperactive, and global behavioral problems as well as higher levels of anxiety during the lockdown. Parallel with our results, other COVID-19 papers also indicate elevated behavioral problems among children [including children on the autism spectrum; (13, 14, 16, 19, 44–46)]. The impact of the COVID-19 lockdown appeared to be disrupting the children's routine, which in turn may be contributing to the elevated inattentive and hyperactive behaviors as well as higher levels of psychological distress. The increase in inattention and hyperactivity symptoms might be also contributed by the impatience in performing the suggested hygiene procedures (13) or struggling to stay focused throughout the online classes at home. Moreover, these negative impacts might also be associated with the more screen time children were getting during the home confinement. Previous studies have found that children who engaged in more screen time tend to display more behavioral problems such as inattention and hyperactivity (47–49) as well as higher psychological distress (50, 69).

On the other hand, children of both groups did not display significantly more perceived stress and depressive symptoms. These findings align with papers that reported marginal improvement in emotional moods and psychopathological dimensions in individuals on the autism spectrum (68) and children with ADHD (51) during the pandemic. One of the explanations to the non-significant change in stress and depression levels might be that some children, especially some children on the autism spectrum are more comfortable at home without the academic and social requirements in school during home confinement (17, 64).

Moreover, parents of children on the autism spectrum reported their child displaying more inattention, hyperactivity, and global behavioral problems as well as higher levels of depression, anxiety, and perceived stress than the TD children, both pre- and mid-lockdown. This is in line with the literature where children on the autism spectrum have been commonly reported to display more inattentive and hyperactive behaviors (52) and higher levels of psychological distress (21, 53) than their TD peers even before the pandemic.

In addition, the third hypothesis that parents of both groups were expected to report a higher level of psychological distress and lower well-being mid-lockdown than pre-lockdown, with parents of children on the autism spectrum reporting higher psychological distress and lower well-being than TD parents, was also partially supported. Parents of children of both diagnosis groups reported experiencing marginally higher levels of perceived stress and depression as well as lower level of well-being mid-lockdown. This is supported by research that has found that caregivers of children with NDDs (including children on the autism spectrum) reported worsening of mental health as well as a heightened level of stress due to the pandemic (13, 28, 68). Based on the limited qualitative and mixed method research conducted among children on the autism spectrum and their families during the pandemic, including in the project's preliminary qualitative interview analysis, the inability to leave the house and the lack of personal space (e.g., “me time”) were some of the common themes that have been mentioned by majority of the caregivers interviewed (54, 55, 67). However, the level of anxiety did not elevate significantly for these parents during the lockdown. As the survey was distributed relatively at the early stage of the lockdowns, the parents might be still adapting to the new lockdown or even enjoying the “break time” of remote working. Based on a qualitative interview research conducted in Turkey, parents reported to appreciate the quality time they spent with their families and children during the lockdown (54). Therefore, this might help explain why the anxiety level increased but did not elevate significantly during the initial stage of the lockdown.

Furthermore, parents of children on the autism spectrum reported significantly higher levels of depression and anxiety as well as lower well-being than parents of TD children pre- and mid-lockdown. On the other hand, the perceived stress level was similar for parents of both groups. Under normal time, parents of children on the autism spectrum frequently report having higher levels of anxiety and depression than parents of TD children as well as children with other neurodevelopmental conditions (56–58, 66). The pandemic and lockdown situations are notably stressful for parents of children on the autism spectrum. With the lack of therapy and social supports during the lockdown, parents of children on the autism spectrum have to manage without external help from school and therapists.

The third objective of the current study was to explore the relationships among the children's behavior and psychological distress as well as parents' psychological distress and well-being during the COVID-19 lockdown in Malaysia among both ASC and TD groups. The results of mid-lockdown correlation analyses revealed that child behavioral indicators were significantly correlated with child psychological distress indicators among both ASC and TD groups. This indicates children's psychological distress might exacerbate children's inattentive and hyperactive behaviors during the COVID-19 pandemic lockdown. Hence, it is important to also take care of the children's psychological well-being in order to address their behavioral issues.

On the other hand, child behavioral and psychological indicators were only significantly correlated with parent psychological distress and well-being among the ASC group, but not in the TD group. This implies that the transactional parent-child relationship may be stronger in the autistic parent-child dyads than the typically developing parent-child dyads. The result contradicted with studies on transactional parent-child relationship that showed similar parent-child relationships between ASC and TD groups [e.g., (59)]. This might be because during stressful events such as the COVID-19 pandemic, where children's daily routines and therapies are significantly disrupted, children on the autism spectrum experience heightened stress. Hence, they might be extra reliant on their parents. The result is in line with the previous findings where the anxiety levels of parents were significantly correlated with children's ASC symptoms and behavioral problems during the pandemic (13, 60).

There are a number of limitations that should be taken into account in interpreting the findings of the study. First, the data was collected solely online. Hence, the results may not be representative of families who do not have access to the internet or electronical devices. Second, the online survey was distributed when Malaysia was in the Recovery MCO (RMCO), where most businesses were allowed to operate, and students were allowed to go to school. Therefore, some parents might be busy adapting to the new routines and had no opportunity to complete the survey. Parents' willingness to participate in the study is notable and greatly appreciated, although some of the parents had difficulties completing as it was time consuming, leading to a relatively high number of incomplete surveys. In view in these possible selection biases, our findings may underestimate the difficulties experienced by these families during the pandemic, as those experiencing more difficulties may not have the privilege to complete the survey. Third, the participants were instructed to complete the survey based on their experiences at two time periods: during the lockdown and before the lockdown in retrospective. Though we made it really clear about the timepoints in the survey, the participants' evaluation of their previous experience may have been influenced from current experience, and such recall bias can be a limitation of this study. Lastly, future studies are recommended to analyse in more detail on the children on the autism spectrum based on comorbidities such as intellectual disability and ADHD.

Despite the limitations, the present study still provides valuable information for future research. This is the first study that provides data on the behavioral and psychological distress changes due to the COVID-19 lockdown among children on the autism spectrum in Malaysia. Some practical implications could be derived from the results. For instance, the obtained data could help governments to take the needs of children with developmental conditions, such as children on the autism spectrum, and their families into consideration when deciding the confinement rules to preserve their mental health and well-being. Moreover, professionals should be alert to the more common behavioral and psychological responses of children and parents to detect the need for intervention as early as possible. Particularly, vulnerable children, such as children on the autism spectrum should receive special attention as they have higher risk factors of displaying behavioral and psychological symptoms. Lastly, knowing the specific impact of the multistage lockdowns on the behaviors and mental health of children on the autism spectrum and their families may help professionals, such as teachers, therapists, psychologists, pediatricians, psychiatrists, public health workers, or those who work with these populations to provide tailored support. For example, considering the associations between parents' mental health and children's behaviors and mental health, these relationships imply that treatment and public health policy needs to address both child and parent concerns (61).

In conclusion, it is evident that the pandemic has created a lot of stressors which impacted on the behavior and psychological well-being of children, as well as the psychological distress and well-being of their parents. After the study has concluded, the lockdown scenario has changed multiple times, from loosening the restriction to reintroducing the lockdown and eventually loosening the restrictions again. Overtime, we follow the evolution of these participants through similar surveys at different timepoints, hoping to draw conclusions on the long-term impact of these measures on the participants' clinical and non-clinical outcomes overtime.