Autism in Preschool-Aged Children: The Effects of COVID-19 Lockdown

While the COVID-19 lockdown was an unsettling period for most people, it was recognized to have been particularly hard for individuals with pre-existing learning, social, or behavioral difficulties, as is the case for neurodevelopmental disorders (Thorell et al., 2022; Guller et al., 2021). In an Italian study (Termine et al., 2021), the authors analysed more than 8000 questionnaires from school-aged children and adolescents with and without neurodevelopmental disorders and found that younger children suffered the most from the interruption of learning activities. This evidence was a prompt to focus on the impact of the pandemic on neurodevelopmental disorders in preschool-aged children. Indeed, preschool years represent a sensitive phase of development (Frankenhuis & Walasek, 2020; Woodar & Pollak, 2020), and integrated services in childcare (e.g., kindergartens, playrooms) are essential to providing affection, protection, and instruments through which to develop social, emotional, and cognitive abilities (Gordon et al., 2020; Mahoney et al., 2021). Discontinuation of rehabilitation therapies and closure of educational centers can block skill progress and lead to children with special needs missing the opportunity to advance their essential abilities (Lee, 2020).

Autism Spectrum Disorder (ASD) is one of the first neurodevelopmental disorders to emerge in children under 6 years of age (Persico, 2018); this may be due to the early manifestation of symptoms (e.g., motor stereotypies and eating and sleep issues) that impact families’ daily lives (Gabbay-Dizdar et al., 2022; Miller et al., 2021). Even if ASD is caused by genetic (Bailey et al., 1995; Muhle et al., 2004; Thapar & Rutter, 2021) and environmental factors (Emberti Gialloreti et al., 2019), stressful life events and routine/lifestyle changes are known to strongly contribute to the varied expression of autism-related traits (Fuld, 2018; Taylor & Gotham, 2016), and the pandemic certainly classifies as a stressful period. Some of the most relevant modifications in environment caused by this event were (1) the introduction of remote learning, (2) possible behavior changes due to routine alteration, (3) the shift from school to home activities, (4) the disruption of sleep habits, especially (5) night awakenings, (6) the potential interruption of physical activity due to home confinement, and (7) the increase in stress levels associated with an unknown virus. Such substantial changes in everyday life are added to the pathological traits that are characteristic of the ASD condition, such as inflexibility and trouble with variation in schedule (Reicher, 2020), with a potential regression of previously acquired socio-cognitive skills (Pearson et al., 2018). As for behavior, ASD-related aggressiveness and stereotypies were worsened by social deprivation (Feige et al., 2021) and changes in routine (Alhuzimi, 2021). Regarding home confinement, parents working from home, owing to a better work-life balance (Ipsen et al., 2021), could support ASD children in mitigating their symptoms (Tokatly Latzer et al., 2021; Zwaigenbaum et al., 2015); however, other studies suggest that the positive effect deteriorated when the lockdown lingered, which increased parents’ stress levels (Bozkus-Genc & Sani-Bozkurt, 2022). Sleep habits are also important to investigate, with insomnia and other sleep disturbances manifested by up to 80% of children with ASD between the ages of 4 and 6 years (Posar & Visconti, 2020). Indeed, other studies on the effect of the pandemic on sleep habits, which also included older children, reported that children with ASD had reduced sleep duration, anxiety at bedtime, and difficulties in falling asleep (Bruni et al., 2021). The interruption of physical activity hindered these positive benefits, such as, equilibrium between aggressive behaviors and self-control, tolerance of stress, improvement to executive functions, working memory, and communicative aspects (Bidzam-Bluma & Lipowska, 2018). Regarding virus-related stress, children with ASD may show limited interest in external events due to an avoidance attitude (O’Nions et al., 2018) or constant requests due to typically repetitive behavior.

Although other studies have investigated the effect of COVID-19 on these aspects, they usually focus on older ages, e.g., with regard to remote learning or home activities (Bozkus-Genc & Sani-Bozkurt, 2022), or they focus on single aspects such as sleep disturbance (Bruni et al., 2021) or behavior (Boterberg et al., 2022; Logrieco et al., 2022) rather than providing a comprehensive view, thus making it useful to further explore this topic. Lastly, the clinical implications of the COVID-19 lockdown should also be considered concerning the effect of the interruption of specific therapeutic plans, which probably caused at least the partial loss of the adaptive-relational expertise.

The aim of this study is to analyze the psychological and environmental impact of the COVID-19 pandemic on a pre-school population through an online questionnaire completed by parents. We hypothesized that both socio-demographic and clinical variables can influence the seven identified environmental factors that were modified by the pandemic. More specifically, our aim was to investigate whether the ASD condition accounts for certain differences in these factors compared to a matched control group. Moreover, we also inspected the trend of the core symptoms of the ASD group during the two-month complete lockdown period that occurred unexpectedly.

Children from 3 to 6 years of age participated in this study. In Italy, children between 3 and 6 usually attend three years of kindergarten (i.e., preschool classes), which prepares them for primary school by helping them acquire basic learning skills and relational abilities. All participants were residents of Varese, an Italian province with high territorial homogeneity, mainly composed of medium-sized cities. The sample comprised both children with ASD and typically developing children who did not present either ASD or any other neuropsychiatric diagnosis. We leveraged different recruitment sources for subjects affected by ASD to enrich our sample. Hence, parents were asked to fill out an online questionnaire by the Child Neuropsychiatry Unit of Filippo Del Ponte Hospital in Varese and the ANGSA association (National Association of Autistic Subjects’ Parents, Associazione Nazionale Genitori Soggetti Autistici). Controls were recruited through the websites of kindergartens of the Province of Varese, although children with ASD from this source were also included. Specifically, families were alerted about the availability of questionnaires to be completed through email by kindergarten teachers and could access the questionnaires through direct links. For both recruitment sources (hospital/association and schools), the questionnaire was proposed to parents of all children to avoid any selection bias.

Data were collected from April to June 2020. This time frame included both the lockdown period (March–May 2020), when people were allowed to leave home only for urgent reasons, and the period immediately afterwards, when people were allowed to go to work if remote working was not an option. During this second phase, schools and rehabilitation services remained closed, making it comparable to the lockdown phase from the children’s perspective. Due to the young age of the participants, the questionnaires were filled in by parents only. Parents provided consent electronically by answering a direct question, wherein they agreed to the anonymous and confidential participation in the study (Supplementary Material 1, question 1), as per the privacy policy established by the Ethics Committee of ASST dei Sette Laghi di Varese (n.82 of 2020). The questionnaire was completely anonymous as was all the data provided, which was sent and collected in an anonymized form. By completing the questionnaire, the parent declared to have understood the purposes of the study and agreed to participate by providing the anonymous data necessary for its implementation. The study was conducted in compliance with the Declaration of Helsinki.

An online questionnaire (Supplementary Material 1) was designed by a multidisciplinary team of psychologists and child neuropsychiatrists. The first section (questions 2–16) included questions that were useful for characterizing the sample by exploring the demographic and environmental features of the children and their families. First, children were stratified according to ASD or typical development condition. Given the heterogeneity of the recruitment and no possibility of a clinical investigation, we were not able to assign a severity degree to the children with ASD included in the study. For this reason, all children with ASD were treated as a single group, against the typically developing children. We further stratified subjects according to age and gender, as the gender is known to have an association with the ASD condition, with greater prevalence in the male population (Loomes et al., 2017), at least during childhood (Posserud et al., 2021). We then considered the environmental factors that could have influenced lockdown management, i.e. socio-economic status (Hollingshead index, H-SES) (Hollingshead, 1975); the presence of parents at home—considering either family status (child living with one or both parents) and whether they worked from home due to the lockdown— or the presence of siblings that shared home resources but might contribute to engaging in activities with the toddler. The presence of an outdoor space was also investigated as it is known to have several benefits, such as orienting the plays in a spontaneous and creative direction (Lee et al., 2021) and promoting attention restoration and memory while moderating externalising behaviors (McKormick, 2017). Finally, we collected variables that we hypothesized could model the effect of the pandemic-induced anxiety, such as exposure to COVID-19, both in the family or outside the family, or parents working as healthcare professionals. The encoding of these variables is presented in Table 1.

After the characterization section, we proposed questions related to the different areas of interest (questions 17–30), as presented in Table 2. Remote learning was aimed at modelling whether kindergarten activities continued to be performed during the lockdown. Behavior changes described potential degeneration due to home confinement. Home activities was aimed at describing whether parents proposed alternative activities to substitute the lack of kindergarten. Physical activity inquired about the possibility of maintaining a healthy lifestyle during the lockdown. Night awakening and sleep habits investigated potential problems at night time. Info virus was mainly related to the fear that COVID-19 could have infected toddlers. A last section (questions 31–50) was targeted at participants affected by ASD only. Some of the questions were specifically aimed at investigating the ASD condition: Irritability, Aggressivity, Hyperactivity, Attention, Social interaction, Non-verbal communication, Stereotypies, Restricted interests and Sensory reactivity (Fig. 1). Whilst Attention, Tolerance of frustration, Social interaction and Non-verbal communication were considered negative outcomes if they showed a decrease, the other clinical features were considered negative outcomes if they showed an increase. Other questions were aimed at investigating drug therapy and non-pharmacologic support to children before the pandemic (Fig. 2).

We conducted a cross-sectional study. Given the small sample of children with ASD compared to the typically developing ones and to provide a meaningful sample, we reduced the number of typically developing children. Subsampling was performed to achieve a sample of Controls that was matched by age, gender, and socio-economic status. The proportion of ASD children was set to 10% and kept constant in the three variables used for matching the sample. As for the other variables that characterized the sample, we performed Fisher’s exact tests for categorical variables or Mann–Whitney U tests for ordinal variables (dependent variables) to describe any difference in the distribution between ASD and Controls (ASD condition was set as the independent variable).

Statistical analysis was conducted using R (version 4.0.1), with 0.05 considered as the significance threshold. Answers that characterized the population (Table 1) were included as independent variables in a multivariate model. Dependent variables were encoded assigning a positive (1) or negative (0) meaning, according to the latent trait modelled in each of the seven dependent variables (Table 2). When a single answer built the dependent variable (Home activities; Physical activity; Sleep; Night awakenings; Information about virus), it was kept as binary. Otherwise, when more than one answer built the dependent variable (Remote Learning; Behavior), the latent trait was achieved by applying a 2-parameter Item Response Theory model (IRT) (Hambleton & Swaminathan, 2013). To keep a meaning of low-medium–high value in the latent trait, whenever a higher granularity resulted from IRT, we chose the 33rd and the 66th percentiles to re-code it. To understand the effect of health, family, and environment (independent variables) on the dependent variables, we applied binary logistic regression or ordinal logistic regression, according to the granularity of the endpoint. After checking the assumptions of absence of multicollinearity between independent variables (variance inflation factors below 10) (Hair et al., 1998) and proportional odds for the ordinal dependent variables (Brant test) (Brant, 1990), we leveraged the Akaike Information Criterion to keep only the most significant covariates, according to both a forward and backward procedure. Then, only the significant independent variables were used in the final logistic regression models. In case the condition of being affected by ASD was eliminated during the Brant test analysis, two separate models (one for ASD cases and one for Controls) were built. To describe the clinical characteristics of the ASD sample, we applied frequency analysis on the remaining questions.

Results of the logistic regression are summarized in Table 4. The algebraic sign of the value represents the direction of the effect over the dependent variable, whereas the absolute value represents its importance. As shown, the presence of ASD had a strong negative effect on the Remote learning variable, suggesting that the presence of ASD made it more difficult to carry out remote education activities than in the control group. Male gender, high SES, and parents working as healthcare professionals are factors that negatively impacted on Remote learning. A significant negative effect of the ASD condition was also reported for the Info-virus variable, suggesting that children with ASD did not properly ask for information about the virus (asking too little or too much) compared to the control group. This was noted in subjects who were exposed to COVID-19 inside their families and those who could not play outside as well. As in the Info-virus variable, we considered both asking too frequently and never asking to be negative, and we further investigated whether the result was imbalanced towards one of the two extremes. A Fisher’s exact test revealed that a significant difference occurred between ASD and Controls on the three answers (never asking, sometimes asking, frequently asking, p < 0.001): 24 children with ASD never asked about the virus (68.6%, in contrast with 13.7% of the Controls, p < 0.001 in the Bonferroni-corrected post hoc), 2 ASD children frequently asked about the virus, but there was not a significant difference with Controls (ASD: 5.7%, Controls: 23.2%, p = 0.304), 9 ASD children sometimes asked about the virus (25.7%, in contrast with 63.2% of Controls, p < 0.001). Data show that the majority of children in both groups had fun activities in their families (334 positive answers), even if none of the variables had a significant impact on it; in both groups there were variations in the sleep–wake rhythm, with differences in sleeping and waking times compared to the period before the lockdown (it occurred for 262 children), and in night awakenings that decreased or remained stable in both groups (301 positive answers), but the situation was improved by the presence of an outdoor space to play in and in families where parents could not work from home. With regard to Physical activity, 302 of the children from both groups interrupted or reduced the frequency of activities. Finally, the Behavior variable differed so widely between the ASD and control group that it was necessary to build separate models. Having an outdoor space positively influenced the Behavior variable for the Controls: those who had an outdoor space available showed better behavioral modulation.

As shown in Fig. 2, before the lockdown, most subjects (94%) carried out rehabilitation therapies such as occupational and speech therapy, cognitive-behavioral therapy, and psychoeducational therapy. Slightly more than half (52%) maintained alternative therapies (remote therapy) during the lockdown. Approximately 30% of the children with ASD in the sample were placed in residential/semi-residential facilities before the lockdown, most of whom (70%) continued to receive treatment remotely. In addition, before the pandemic, 10 out of 35 children had a home educational intervention; 60% of these children maintained this intervention remotely. With regard to drug therapy, the analysis recorded a single case undergoing drug therapy before the lockdown. The therapy consisted of the intake of a neuroleptic, but the dosage was not changed during the lockdown. As shown in Fig. 1, a total of 43% of parents of children with ASD judged their child’s general symptoms to be worsened; another 43% of parents said they were stable; in contrast, a small percentage (14%) judged their child’s symptoms as improved during the lockdown. In terms of clinical symptoms, our data show a worsening, particularly of repetitive movements or speech patterns and restricted interests, in addition to a worsening of aggressive behavior among children who already had it before the pandemic. The alterations of clinical features in subjects with ASD are shown in Fig. 1. Of the sample, 57% of showed a regression of the skills previously acquired after the interruption of rehabilitation therapies during the lockdown. Finally, 37% of the children required a medical certification to justify a short exit from home to reduce anxiety and hyperactivity caused by persistent confinement at home and to prevent deregulatory behavior.

These results suggest that the presence of autism had a strong negative effect on learning: children with ASD had more difficulties with kindergarten distance learning. This was characterized by watching videos or listening to recorded audios, participating in video calls with educators, and classmates and carrying out certain educational home activities proposed by teachers. For many educators, as well as for tutors, the pandemic was the very first chance to propose remote activities to children, most of whom were not specifically trained to apply these methods. This could have contributed to amplifying the children’s difficulties, especially in those with ASD, who require individualized educational plans that remote learning could not completely provide. Moreover, the sudden necessity of remote programs could have contributed to logistic-related problems, such as the absence of a good internet connection or proper tools, given that a preschool child does not typically engage in online activities and families with young children could also be less prone to do so. Our findings are supported by studies that show difficulties in attending remote activities by children with special needs, who face barriers due to the absence of accessible materials and of the proper support to follow online programs (United Nations, 2020).

During the first wave of COVID-19, there were few confirmed cases of infection in toddlers or children. Even amongst children who contracted the disease, its course was mild to moderate, especially for those under the age of 5 (Bhuiyan et al., 2021), and the mortality rate was considerably low (Spaull, 2020). However, considering the overall implications of the pandemic, a strong impact on the mental health of young children was noticeable. Behavioral differences with typically developing children was so marked that it was necessary to build separate models for the two populations. The upset of usual activities indeed led children with ASD to a loss of balance, since one of the core domains of the spectrum is the inflexible adherence to routine (American Psychiatric Association, 2013). The presence of ASD also affected children’s behavior and emotions, as evidenced by the reports of parents: tolerance of frustration, aggressiveness, and hyperactivity worsened significantly. Numerous studies have found that prolonged isolation is a risk factor for functional decline and tends to fuel negative feelings such as loneliness, intolerance, irritability, anger, and problematic behaviors (Jacques et al., 2022). These exacerbations could have been increased by new routines practiced during home confinement, such as varying domestic activities, spending a great deal of time using computers, watching television or playing video games, which altered the usual dietary habits and sleep–wake rhythm (SIDiN, 2020). In addition, the alterations in sleep–wake rhythm could contribute to enhancing behavioral signs, as well as in worsening daytime functioning (Mazurek et al., 2019), given also the fact that most ASD children between the ages of 4 to 6 years suffer from insomnia (Posar & Visconti, 2020) and other sleep disturbances that are often associated with aggressive behaviors (Wang et al., 2021). According to evidence about the lockdown, parents reported mental health difficulties and increases in child behavioral problems since the onset of the pandemic (Yoshikawa et al., 2020); children with ASD were particularly vulnerable, considering (i) the specific support they needed in this period to preserve their psychological well-being (Colizzi et al., 2020) and (ii) the atypical emotion regulation that characterizes children with autism at all ages (Mazefsky, 2015). As a future development, it could be of interest to investigate to what extent poor non-verbal communication abilities led to frustration and thus affected the other variables of the study. Indeed, non-verbal communication is moderately correlated with Behavior (Spearman’s rho 0.58, p < 0.001), and its effect may transcend COVID-related modifications.

The presence of ASD also negatively influenced the proper requests for information about the virus. In this study, assessing whether children were asking for information about COVID-19 was intended to indirectly measure the children’s concerns and stress levels caused by the virus and by their state of health. Children with ASD did not ask for information about the disease; the Controls, however, showed a proper level of concern and anxiety about it. This could be elucidated by the marked difficulties for children with ASD concerning social reciprocity and communication (American Psychiatric Association, 2013) and also by their low proneness to understanding the hygiene-related guidelines needed during the pandemic (Mutluer et al., 2020). On the other hand, it should be acknowledged that children in the ASD group were seldom exposed to the virus (only 1 case out of 35), which justifies the lack of interest in the topic.

The dependent variables that changed in all subjects irrespective of ASD condition, were influenced by the following main factors (i.e. independent variables). Having an outdoor space at home had a positive effect on sleeping, behavior, and virus-related anxiety in all children: fewer night awakenings, behavioral difficulties and concerns about the virus occurred in those who had a terrace or a garden. These results can straightforwardly be explained considering that outdoor activities have a positive impact on sleep and sedentary behavior in children under 5 years of age (World Health Organization, WHO 2019); outdoor activities, by reducing sedentariness, ensure a good quality of sleep in young children and thus can improve school performance as well (Janssen and Leblanc, 2010). Likewise, outdoor spaces in kindergarten are dominant for children’s activities: a natural environment captures their attention and thus has a positive influence on their psycho-physiological status (Lan & Guo, 2021; Biørgen, 2016).

As for the descriptive analysis of clinical data of children with ASD, results showed that telematic rehabilitation therapies were ensured in half of the cases during the first lockdown. Furthermore, the need for training of previously learned skills in children with ASD is clear-cut, as evidenced by the fact that almost half the parents of those who kept the telematic rehabilitation described a waning of acquired abilities as well as a worsening of their children’s general symptoms. With regard to symptoms, stereotyped movements and speech, restricted interests, and externalizing behaviors (hyperactivity, aggressivity) displayed an increase, probably as a consequence of the lack of scheduled activities and the frustration induced by home confinement. Concerning social and neurocognitive decline, the regression of previously acquired skills was arguably magnified by the disruption of clinical and educational routine, considering the ubiquitous limitation of usual pleasant activities, as well as of relationships with therapists, teachers, and tutors (Neece et al., 2020). Children with ASD in particular can experience a waning of acquired skills, a phenomenon known as late regression that can simulate global delay (Persico, 2018) and that is strongly influenced by environmental stimuli, i.e., by the possibility of constant training. Evidence about telehealth in toddlers with ASD underlines its strength in the chance to actually observe the home environments and, as a consequence, give families strategies that they could easily implement in daily activities. Indeed, even though families receiving in-person services report a slightly higher perceived improvement than those receiving tele-consults only, no significant differences were experienced between hybrid and in-person service models (Corona et al., 2021). Researchers offer two important statements concerning the use of mixed tools (i.e. technology-based and in-person tools) for managing ASD support: first, technology-based interventions in ASD subjects show a good effect if applied for a proper duration (Grynszpan et al., 2014); second, naturalistic developmental behavioral interventions proved to be effective in improving social engagement and cognitive abilities (Tiede & Walton, 2019). Finally, only one autistic subject in our sample was undergoing drug therapy and the dosage did not change during the lockdown. This is in line with the absence of specific drugs for treating the nuclear symptoms of autism (Sanchack & Thomas, 2016) and emphasizes once more the importance of environmental support in managing ASD.