The psychological impact of the COVID-19 pandemic on adults with autism: a survey study across three countries

Background Previous studies have reported a negative psychological and mental health impact of the COVID-19 pandemic. This impact is likely to be stronger for people with autism as they are at heightened risk of mental health problems and because the pandemic directly affects social functioning and everyday routines. We therefore examined COVID-19 pandemic-related changes in mental health, the impact of the pandemic on their social life and routines, satisfaction with pandemic-related information and tips, and participants’ wishes for guidance. Methods We used a mixed-method approach, collecting quantitative and qualitative survey data from adults with and without autism across three European countries: Belgium, the Netherlands, and the UK (N = 1044). Results We found an increase in depression and anxiety symptoms in response to the pandemic for both the non-autism and the autism group, which was greater for adults with autism. Furthermore, adults with autism showed a greater increase in worries about their pets, work, getting medication and food, and their own safety/security. They felt more relieved from social stress, yet experienced the loss of social contact as difficult. Adults with autism also felt more stressed about the loss of routines. Pleasant changes noted by adults with autism were the increase in solidarity and reduced sensory and social overload. Adults with autism frequently reported problems with cancellation of guidance due to the pandemic and expressed their wish for (more) autism-specific information and advice. Limitations Our sample is likely to reflect some degree of selection bias, and longitudinal studies are needed to determine long-term effects. Conclusions Results highlight the psychological burden of the pandemic on adults with autism and shed light on how to support them during this COVID-19 pandemic, which is especially important now that the pandemic is likely to have a prolonged course. There is a need for accessible, affordable (continued) support from health services. Guidance may focus on the maintenance of a social network, and adjusting routines to the rapid ongoing changes. Finally, we may learn from the COVID-19 pandemic-related changes experienced as pleasant by adults with autism to build a more autism-friendly society post-pandemic.


Introduction
On March 11, 2020, the World Health Organization (WHO) officially declared the COVID-19 virus a global pandemic [1]. The global spread of the virus led to a massive and immediate public health campaign to reduce further spread. This campaign encouraged large-scale Oomen et al. Molecular Autism (2021) 12:21 behavior change, including social isolation and distancing, which has a massive impact on how we live and socialize on a daily basis. Added to these many daily life changes (e.g., studying/working from home, loss of activities, and real-life social contact) that we had to adapt to rapidly, this unprecedented pandemic comes with great uncertainty and heightened levels of threat (e.g., worry about loved ones or yourself contracting the virus; [2]). Both the pandemic itself and the encouraged behavior change can have an impact on our mental health as uncertainty is a cognitive and psychological stressor [3], while social isolation has consistently been associated with negative mental health outcomes and conversely, social support is considered an important buffer in adverse times (e.g., [4]). Hence, concerns have been raised regarding the impact of the pandemic and the associated containment measures on the mental health of individuals across society (e.g., [5]). Studies on previous epidemics as well as the current COVID-19 pandemic indeed highlight their negative mental health impact (e.g., Equine influenza: [6]; SARS: [7,8]; H1N1: [9]; COVID-19: [10][11][12][13][14][15][16]).
Importantly, although the COVID-19 pandemic has an influence on society as a whole, negative mental health effects due to the pandemic might not be evenly distributed. Individuals with autism spectrum disorder (ASD; henceforth 'autism' 1 ) may be considered one of the vulnerable groups [5] that are likely to be affected to a greater extent. Autism is a prevalent neurodevelopmental disorder with a global worldwide prevalence estimated as 0.62% or higher [17]. The impact of the pandemic may be stronger for individuals with autism for two main reasons: Firstly, autism is associated with a heightened risk of mental health problems, including a higher likelihood of comorbid mental health disorders such as mood and anxiety disorders [18,19]. Secondly, autism is characterized by two core symptom domains which are directly impacted by the pandemic. On the one hand, individuals with autism experience social interaction and communication difficulties, such as difficulties initiating or responding to social interactions, adjusting their behavior to suit various social contexts, and developing and maintaining relationships. On the other hand, individuals with autism show repetitive and restricted behavior, interests, and activities. The latter domain includes insistence on sameness and inflexible adherence to routines, which sometimes results in extreme distress in response to small (routine) changes and difficulties with transitions. These two core symptom domains of autism are directly affected by the pandemic, as it both impacts our social interactions and results in big, continuing changes to our everyday lives. This gives reason to believe that individuals with autism may experience the pandemic in a unique way. Given the fact that adults with autism are reported to have more difficulty coping with stressful events than neurotypicals do [20,21], and especially because levels of anxiety and depression are known to be generally high in the autism population [18,19], there is a need for research to ascertain the impact of the pandemic on the mental health of individuals with autism.
This need for research on COVID-19 and autism is also evident from the emerging articles that emphasize the concerns for individuals with autism during the COVID-19 pandemic [22][23][24][25][26][27][28][29]. Asbury et al. [23] asked parents of school-aged children with special educational needs and disabilities to describe the impact of the COVID-19 pandemic on the mental health of their child. The majority of the sample consisted of parents of a child with autism (82%). Children were reported to experience anxiety/worry and low mood. In the study of Amorim et al. [26], caregivers of children with autism also reported more anxiety during quarantine, both for themselves and their child, compared to children without autism and their caregivers. Furthermore, the majority of caregivers in the autism group reported that quarantine negatively impacted emotion management in their children, whereas the majority of parents without autism reported no impact or even a positive impact of quarantine on emotion management in their children. Studies on behavioral problems during the COVID-19 pandemic have found an increase in behavioral problems in children with autism since the pandemic [28], and that preexisting behavioral problems predict a higher risk of more intense and frequent behavioral problems during the pandemic [29]. The commentary of Ameis et al. [22] highlights the potential negative impact of changes to everyday routines and restrictions to regular services on children with autism. White et al. [24] surveyed caregivers of children with autism and indeed found services and therapies for individuals with autism to be disrupted. Thus far, there is only one survey that most likely reports data from adults (among younger responders): the National Autistic Society, a charity, published a report [27] which revealed that 9 out of 10 people with autism worried about their mental health during lockdown. However, they did not include a control group or sample characteristics, preventing any conclusions about how these mental health worries compare to the general population, or about how many of the respondents were adults. To date, no study has systematically investigated the psychological impact of the COVID-19 pandemic on adults with autism specifically. The current study therefore focuses on adults with autism, and includes a control group of adults without autism. Information given directly by adults with autism themselves will provide valuable insights into the influences of social distancing measures and abrupt changes to routines on the mental health and lives of adults with autism. These insights will help to form actions to mitigate the negative effects of the COVID-19 pandemic. Furthermore, insights may be relevant for future pandemics and/or public health crises.

Current study
The current study aimed to ascertain the impact of the COVID-19 pandemic and the associated containment measures on the psychological well-being of adults with autism (in comparison with adults without autism), as well as their needs throughout this time. As the COVID-19 pandemic affects people globally, we sampled broadly and included participants from three European countries: Belgium, the Netherlands, and the UK. 2 The governments of each of these three countries released a statement to self-isolate or social distance halfway through March (i.e., Belgium: 18 March, the Netherlands: 15 March, UK: 23 March), and Belgium started to relax its containment measures in the beginning of May. The current study focused on the beginning of the pandemic and containment measures until the first relaxation of these measures. Data collection ran from April 3 to May 7, 2020.
The main focus was to investigate the impact of the pandemic and the associated containment measures on the mental health of adults with autism by assessing depression and anxiety symptoms and specific worries. We also assessed how the pandemic affected the two autism symptom domains by inquiring about the impact on their social life (domain I), and their everyday life and routines (domain II). In addition, we asked the participants how satisfied they are with the pandemicrelated information and tips available, and whether people received guidance or had a wish for guidance during this pandemic. We applied both a quantitative and qualitative approach. Qualitative data were gathered to gain a broader insight into individual needs, and to explain what pandemic-related changes were experienced as the most difficult, caused the most stress/anxiety, and made life more pleasant (if any).
To the best of our knowledge, this is the first systematic study on the psychological impact of the pandemic on adults with autism. For this reason, we had no strong specific hypotheses. We did however expect an overall negative mental-health impact across groups, in line with previous research in neurotypical populations (e.g., [11]). Furthermore, we expected the negative mental health impact of the pandemic overall to be greater for adults with autism than adults without autism.
Participants were recruited via online advertisement through various social media channels, organizations (e.g., discover network of Autistica) and via existing databases of individuals with autism that have previously given their consent to be contacted for autism-related research. Participants had to be 18 years or older, residents of Belgium, the Netherlands, or the UK and have sufficient knowledge of the English language. Following an online information letter and consent form, participants were asked to fill out a series of short questions assessing their personal situation, their pandemic experience, autism symptomatology, and the impact of the pandemic on their daily life and mental health. The survey took approximately 20 min to fill out. On every page of the survey participants were provided with links to resources for (mental health) support and pandemicrelated information in case they needed them. This study was approved by the local ethics committee of the Faculty of Psychology and Educational Sciences of Ghent University (EC/2020/46).

Survey
See "Appendix A" for the full questionnaire. Before going online, the survey was checked by an adult with autism who provided feedback on (the clarity of ) the questions.

Demographics
We inquired about the following demographics: age, biological sex, country of residence, years of education (starting from year 1, i.e. learning to read), own employment status, and (if applicable) their partner's employment status.

Personal situation
We inquired about participants' personal situation by assessing their living situation (e.g. assisted living facility, living with flat mates), and the number of adults, children, and pets living with them.

Personal situation: COVID-19
We inquired about the COVID-19 specific personal situation of participants by assessing the following topics: Whether they have (had) COVID-19; time spent actively looking up/reading about COVID-19; time spent going over COVID-19-related information in their minds; their self-perceived COVID-19 knowledge level; how strictly they think they are following the recommendations from national authorities; how the pandemic influenced their daily routines (e.g., living as normal, self-isolation, social distancing); usual amount spent socially interacting; and the number of days they: self-isolated, went outside, met someone outside, had face-to-face contact, had a phone or video call.

Autism spectrum disorder
We inquired whether participants had a formal clinical diagnosis of autism or not. To ensure answers were reliable, questions were added inquiring about the specific diagnosis, and the age of diagnosis. Additionally, the survey also included a question inquiring about other formal clinical diagnoses (ADHD, depressive or anxiety disorder, or other). Autism symptomatology was assessed with the Autism spectrum quotient-short (AQ-short; [30]) in all participants. The AQ-short contains 28 items comprising two higher-order factors: 'social behavioral difficulties' and 'fascination for numbers/patterns' . The former factor can further be divided into four lower-order factors: 'social skills' , 'routine' , 'switching' and 'imagination' . Items were rated on a four-point scale ranging from 1 ('definitely agree') to 4 ('definitely disagree'; total score ranging from 28 to 112). Reverse items are included in such a way that agreement suggests the presence of an autistic-like trait in half of the items. Hoekstra et al. [30] suggest a cut-off of > 65. The internal consistencies of the AQ-short in the current sample were excellent (Total: α = 0.94; Social behavior: α = 0.94; Social skills: α = 0.91) and good (Routine: α = 0.82; Switching: α = 0.82; Imagination: α = 0.84; Numbers and patterns: α = 0.81).

Depression and anxiety
To asses COVID-19 pandemic-related changes in depression symptoms we adapted the Patient Health Questionnaire-9 (PHQ-9; [31]). The PHQ-9 contains 9 items designed to correspond to the nine DSM-IV diagnostic criteria for major depressive disorder, and an impact question (i.e., "How difficult have these problems made it for you to do your work, take care of things at home, or get along with other people?"). To assess COVID-19 pandemic-related anxiety-symptom changes we adapted the Table 1 Demographic data per group Autism group: n = 613, non-autism group: n = 431 a 3 cases that indicated intersex (autism group n = 3; non-autism group n = 0) and 13 cases that indicated 'prefer not to say' (autism group n = 12; non-autism group n = 1) are not included in the percentages of sex, due to an insufficient number of cases b Following the DSM-5 (e.g. ADHD includes ADD). Examples of 'other' category: trauma and stress-related, obsessive-compulsive, psychotic, eating, personality, learning disorders  Generalized Anxiety Disorder scale (GAD-7; [32]). The GAD-7 contains 7 items designed in part on the basis of the DSM-IV criteria for GAD, and an impact question. All items of both screeners were included, but to capture COVID-19 pandemic-related changes in depression and anxiety symptomatology we adjusted the rating scale: 1 ('significantly less than usual'), 2 ('slightly less than usual'), 3 ('not more than usual'), 4 ('slightly more than usual'), 5 ('significantly more than usual'). We left the four-point scale of the impact questions unchanged, ranging from 1 ('not difficult at all') to 4 ('extremely difficult'). In this paper, we will refer to the adapted PHQ-9 and GAD-7. The internal consistency of the adapted PHQ-9 was good, α = 0.88. The internal consistency of the adapted PHQ-9 was excellent, α = 0.91.

Worries
We inquired whether in the last two weeks, participants experienced more or less worries than usual about their marriage or other romantic relationship(s), pet(s), access to medication, food and internet, safety/security, future plans, finances, work (even if they feel their job is safe), losing their job, and boredom. Items were rated on a fivepoint scale: 1 ('significantly less than usual'), 2 ('slightly less than usual'), 3 ('not more than usual/ not applicable'), 4 ('slightly more than usual'), 5 ('significantly more than usual'). Furthermore, we inquired whether participants experienced the following COVID-19 pandemic specific worries: following the recommendations to prevent spread of COVID-19 correctly, catching COVID-19, and friends or family catching COVID-19. Items were rated on a five-point scale ranging from 1 ('not at all') to 5 ('very much so').

Social life and routines
We assessed the impact of the pandemic on the two symptom domains of autism through the following three questions on their social life and routines.

Guidance
We asked whether participants received any help/guidance from a qualified medical specialist (e.g., psychiatrist, psychologist, counsellor) in the last 6 months, and if they answered 'no, but I would like to' , whether their wish for help/guidance was related or unrelated to the pandemic. Furthermore, we assessed how satisfied they were with the tips and tools that are available during the COVID-19 pandemic. If someone indicated to have a formal diagnosis of autism, we adjusted this question to include 'for individuals with autism' . Lastly, participants with autism were also asked how important it was to them that individuals with autism are consulted for the development of these autism-specific tips and tools. These guidance questions were not mandatory.

Qualitative questions
Four (

Quantitative data
Means with standard deviations or, where applicable, frequencies with percentages are reported for all demographics per group. We used linear mixed-effect models (LMMs) to detect group differences (autism versus nonautism group) on our outcome variables: COVID-19 pandemic-related changes in depression (adapted PHQ-9 total score) and anxiety symptoms (adapted GAD-7 total score), as well as on the additional question to assess the impact of the depression and anxiety symptoms on everyday life; 14 topics of worry (11 general ones, and 3 COVID-19 pandemic specific worries); the 6 routine (domain I) and social contact (domain II) questions (3 for each domain). In our models we included group as a fixed factor to detect group differences on our outcome variables. To control for variation coming from the three different countries, we included country as a (nested) random (slope and intercept) factor. LMMs allow us to take advantage of the full sample (power) and yet account for the correlation between data (i.e., non-independence) coming from the three different countries. This way we acquired results on the impact of the pandemic on adults that generalize across the three countries. We used the lme4 package to test the LMMs [33]. Lastly, for the questions on guidance, we reported frequencies with percentages.
In total we ran 24 LMMs. As a Bonferroni correction was regarded as too conservative, we made use of the False Discovery Rate (FDR) approach to correct for multiple testing [34]. In short, q values (FDR adjusted p-values) control the number of type I errors in the tests that result in a significant result, when determining adjusted p-values for each test. In the current study, we rejected 15 (62.5%) null hypotheses with q* = 0.05. Simply put, we estimate that 95% of these 15 rejections are true discoveries (following a Bayesian interpretation of the q-values; see [35]). Our interpretations will be based solely on the q-values, but for transparency we also report p-values. Note that Bonferroni correction for p-values would test each individual hypothesis at α = 0.05/24 = 0.002.

Qualitative data
In total, 82% of participants provided a response to at least one of the open questions (85% of the autism group, 78% of the non-autism group). Thematic analysis [36] was conducted by DO and AN. Both researchers independently familiarized themselves with the data by reading through the answers of each question of the first 200 participants, and noting down initial codes that described the key messages conveyed with the answers. We used an inductive approach for each open question, allowing the data to determine the codes. Thereafter the researchers met to discuss reoccurring codes to create preliminary codebooks-a separate codebook for each question-and to align the usage of code labels. After this, the four questions for all participants were independently coded by the two researchers separately. While coding, the researchers met regularly when the codebooks needed adjustment (e.g., breaking down codes into separate codes to obtain more specificity, inserting new codes that are not covered by an existing code). When all data were coded, the authors collaborated in identifying the most important codes per question (i.e., the ones that occurred the most often). We then generated themes that overarched the most important codes that related to each other. Codes were omitted that did not occur frequently or that did not significantly add to the developing themes. Lastly, we looked at how the themes fit together to describe the data in a coherent fashion, and whether they were autonomous, or subthemes. Themes, and where applicable subthemes, are presented together with example quotes to add in-depth meaning to the (sub)themes. We mostly selected quotes written in English, so as to avoid having them lost in translation, but quotes represent answers of the whole autism group. Quotes that we did translate from Dutch are indicated with "[translated]". Results are reported per question, although some themes occurred under multiple questions, and were later assigned under the most appropriate question by the researchers to avoid too much repetition.

Demographics
Demographic information on participants per group can be found in Table 1. The majority of adults with autism in our sample were female (68%), which is in disproportion with diagnostic rates [37], but common in large online studies (e.g., [38][39][40]). The larger part of the participants were residents of Belgium. Years of education between groups was similar, but relatively more adults without autism in our sample were employed (61% versus 45%; including self-employment: 70% versus 50%), in line with general findings of reduced employment in adults with autism [41,42]. AQ-short scores within the autism group were comparable to previous samples [30,39], as were the scores of the non-autism group [30]. Over half of the adults with autism in our sample had one or more comorbid psychiatric diagnoses (57%), whereas 18% of the adults without autism reported to be diagnosed with a mental health problem. We collected more information on pandemic-specific characteristics of our sample, as can be found in "Appendix B" Table 10. Not only did we inquire about participants' COVID-19 pandemic-related behavior, we also asked extra questions regarding their specific living situation during the start of the pandemic.

Pandemic-related changes in worries
Means, standard deviations, and LMM statistics of each of the worry topics are given in Table 3. On average, both groups indicated to worry more than usual about most topics of worry (mean scores > 3). Moreover, this increase in worries was significantly greater in the autism group for the following worry topics: Their pets, work (even when they felt their job was safe), losing their job, getting medication and food, and their own safety/security. The autism group did not worry more than the non-autism group about any of the COVID-19 pandemic specific worries.
The non-autism group more than the autism group felt they had to change their daily routines due to the pandemic, β = 0.32, χ 2 (1) = 16.18, p < 0.001, q < 0.001. Nonetheless the autism group felt more stressed about the changes in daily routines than the non-autism group, β = − 0.58, χ 2 (1) = 13.91, p < 0.001, q = 0.001. The groups did not differ in the extent they enjoyed the freedom to deviate from society's expectations when adjusting their routines to the new situation, β = − 0.20, χ 2 (1) = 2.75, p = 0.097, q = 0.129.

Impact of the pandemic on guidance and satisfaction with offered tips and tools
Of the participants from both groups who received guidance pre-pandemic from a qualified medical specialist (e.g. psychiatrist, psychologist, counsellor, thus entailing a broad range of services, practitioners and therapies),

Table 2 COVID-19 pandemic-related change in depression and anxiety symptoms
Autism group: n = 613, non-autism group: n = 431; Questions were rated on a five-point scale going from 1 ('significantly less than usual') to 5 ('significantly more than usual'), except for the additional impact questions that were rated on a 4 point rating scale going from 1 ('not difficult at all') to 4 ('extremely difficult') a Only includes participants that indicated 'slightly/significantly more than usual' to any of the above items. Adapted PHQ-9: autism group n = 524, non-autism group n = 340; adapted GAD-7: autism group n = 504, non-autism group n = 330 Autism M (SD)  Table 5). Across groups, for the majority of participants who had a wish for guidance, the wish was unrelated to the pandemic (58%). However, in the autism group, it was indicated more often that the wish for guidance was 'both unrelated and related' to the pandemic (42% versus 16%). Figure 1a shows that the majority of adults with autism in our sample find it important that adults with autism are being consulted for the development of COVID-19 pandemic tips and tools. Furthermore, more adults with autism indicated to be unsatisfied (answering options 1 or 2: n = 219) than satisfied (answering options 4 or 5: n = 86) with the available tips and tools for individuals with autism during the COVID-19 pandemic (Fig. 1b). Moreover, less adults without autism indicated to be unsatisfied (answering options 1 or 2: n = 73) than satisfied (answering option 4 or 5: n = 176).

Table 3 Mean scores per group regarding eleven general, and three pandemic-related topics of worry
Autism group: n = 613, non-autism group: n = 431; Questions were rated on a five-point scale going from 1 ('significantly less than usual') to 5 ('significantly more than usual'), except for the COVID-19 pandemic specific worries that were rated on a five-point scale going from 1 ('not at all') to 5 ('very much so'). Participants were asked to indicate 'not more than usual' when not applicable a Analysis only includes participants who indicated to have an pet, autism group n = 349, non-autism group n = 209 b Analysis only includes participants who indicated to be self-, part-time, or full-time employed, autism group n = 306, non-autism group n = 300

Measures that may help manage mental health during the pandemic
Themes identified from responses of adults with autism to the open question inquiring about their needs during the pandemic can be found in Table 6. Adults with autism indicated a need for (continued) professional support (e.g. medical, psychological, household support) that is both affordable and accessible. Many adults with autism that did not need regular support before, reported to need it during the pandemic but did not know how to access relevant services. Furthermore, many adults with autism lost some, if not all, of the support they received before the pandemic. The type of support that was lost varied widely, ranging from medical and psychological support to household support. Adults with autism that did report to receive support often indicated to prefer face-to-face support as soon as this was once again possible. Regarding online support, some adults with autism noted to experience anxiety on voice or video calls due to difficulties with the back-and-forth flow of social communication during calls, and would rather communicate via chat (i.e., text-based messaging). Nevertheless, there were some adults with autism that commended the accessibility of remote support. The open question further revealed that the need for help was often related to structuring their days (i.e., creating new routines that fit into pandemic life). Both adults with and without autism expressed the need for clearer information and rules. However, whereas adults without autism mainly described the lack of clarity as an annoyance, adults with autism made explicit statements of distress. This lack of clarity and the accompanying distress led adults with autism to avoid activities that were still allowed, as they were afraid to break protocols (e.g., not going for walks alone or with members of their household at all, as it was not specified how far from their house they were allowed to go). Adults with  autism further wished for (more) autism-specific information and advice that was specifically tailored to adults with autism, as many noted that existing information was mostly directed to parents of children with autism. The answers also indicated that the sources at that time were not yet promoted widely enough, as some were not aware of them.

COVID-19 pandemic-related changes that are the most difficult
The two themes identified from responses of adults with autism to the open question inquiring about COVID-19 pandemic-related difficulties can be found in Table 7.
One theme related to a loss of routine. Many adults with autism indicated to have a clear daily or weekly routine they stuck to pre-pandemic, which they lost because of the pandemic affecting them negatively. They emphasized the importance of structure in their daily life to maintain their mental health, and the experienced difficulty with having to create a new routine by themselves. Some adults with autism also noted they missed social contact in their daily/weekly routines, overlapping with the second theme, namely, a loss of social contact. Both adults with and without autism mentioned an negative impact of the pandemic on their social life. However, whereas adults without autism mainly expressed the burden of missing their pre-pandemic social life (seeing others face-to-face), adults with autism rather described the burden of losing their social support network, which for some people led to full social isolation in an already stressful time. Some adults with autism also expressed the need for social contact to keep up their social skills. Furthermore, answers to this open question revealed that adults with autism experienced new social interaction difficulties directly related to the pandemic, such as not being able to read faces because of face masks, and having difficulty with the back-and-forth flow of social interaction during video calls. Other changes that both groups found difficult were not being able to spend time alone due to living with others, balancing work-life while working from home, balancing work with taking care of children (including home-schooling), and the uncertainty of how long the pandemic and the associated changes will last.

COVID-19 pandemic-related changes that cause the most stress/anxiety
Themes identified from responses of adults with autism to the open question inquiring about COVID-19 pandemic-related stress/anxiety can be found in Table 8. By

Table 6 Themes identified from responses of adults with autism to the following open question: What measures could be taken to help you manage your mental health during the current COVID-19 pandemic?
Theme Subtheme Example quote (Continued) professional support "I need to talk to my psychologist from time to time, but this is impossible right now as it is 'not essential'" p. 38 "Would be nice if a professional checked up, helped with implementing tips and tools and some measure of reassurance was possible" p. 148 "Help with nuanced issues arising from the pandemic-changes to social codes, how to behave around others" p. 171 Affordable and accessible "More easily accessible and affordable psychological/therapeutic help (via Skype, chat …)" p. 318 Availability of various communication mediums "Something that would help me and many others would be if access to primary healthcare was available by some means other than making a phone call-e. Autism-tailored information and advice "ASD Societies contacting individuals who are registered with them to advise them of the situation" p. 8 "The only tips and tricks that are easily available for people with ASD are catered towards children or people with ASD and learning disabilities" p. 224 "I was not aware that there were tips and tools for people with ASD specifically. Being made aware of these things might help. " p. 415 far the most often reported anxiety-provoking theme for adults with autism was related to shopping (grocery, but sometimes also for medication). The mentioned reasons for this were twofold. First, shopping routines were disrupted. This is due to, among other things, the unavailability of specific food and non-food products, the introduction of limits on time spent in the shop and on the number of products one may buy, and having to line up to enter the shop. Second, shopping was described as more stressful due to the social distancing rules. On the one hand, some adults with autism found it difficult to adhere to the rules themselves, on the other hand, being in an environment where other people do not always adhere to the rules induces anxiety and frustration. Another recurrent anxiety-provoking theme was related to the prospect of having to go back to 'normal' after the pandemic. Adults with autism indicated to feel anxious about the prospect of having to change their routine yet again, possibly repeatedly, to adjust to the gradual lifting of lockdown and other pandemic-related measures. Furthermore, they feel anxious about the prospect of a sudden return of external stimuli (e.g. traffic, noise, more people being outside) paired with a loss of coping skills to deal with them. That is, many adults with autism protected themselves from sensory overload pre-pandemic by means of coping skills, which they often needed less of during lockdown, causing them to fear a loss of capability to cope with sensory overstimulation. Similarly, they fear to have lost their social skills, which makes the foreseen pressure to attend social events and return to a work/school environment with new social norms stress-inducing.
Other recurring topics for both adults with and without autism were worries about employment and finances, as well as worries about one's own health and that of close others.

COVID-19 pandemic-related changes that made life more pleasant
Themes identified from responses of adults with autism to the open question inquiring about COVID-19 pandemic-related changes that made life more pleasant can be found in Table 9. One overarching theme was identified, as nearly all answers were related to changes that made the world a little more autism-friendly. Two subthemes further break down why. First, many adults with autism described a feeling of solidarity, now that their lives did not differ as much anymore from those of the people around them. For example, some adults with autism explained that, for once, the whole population had to socially isolate themselves for their own goodsomething that many adults with autism already have experience with, as they sometimes have to socially isolate themselves to reduce sensory overload. Second, lockdown reduced sensory and social overload. The outside world is quieter (i.e., less sensory stimulating), and many adults with autism described a feeling of relief from certain social stressors such as obligatory parties and appointments, spontaneous visits from others, or strangers getting too close.
Recurring topics that both adults with and without autism mentioned were: having more time for themselves (adults with autism particularly seemed to enjoy the extra time they could spend on their special interests) and for family, having complete control over their daily structure (e.g., when to take a break from work, when to relax), working from home (although people also acknowledged that this would not work for months on end), and the overall pressure/ social stressors of everyday life being diminished.

Theme Subtheme Example quote
Groceries Disrupted shopping routines "The unavailability of my specific groceries, for weeks now. Getting to the shop and seeing empty shelves, yet again, bring me to tears. Because it's something I have absolutely no control over. And having to change what I eat on top of everything else is just too much" p. 611 "Unable to do groceries due to meltdowns/ or panic attacks not sure which of these is the culprit. Sensory overload is entangled with supermarkets on a regular day, with all the change it's become impossible to do on my own" p. 440 "shopping-not being able to get my routine food" p. 86 Social distancing rules "Not being able to go grocery shopping without a lot of anxiety. I only go once a week, whereas I was used to go almost daily. I am very scared about people at the shops not adhering to the rules" p. 432 "Grocery shopping is awful now! It was always awful to begin with, but with all the extra regulations it's even more stressful now. Constantly having to keep my distance and reminding myself is an extra, unwanted step in this already exhausting process of having to do everything right, having to think of everything. I now come home upset and sweaty every time. " p. 442 Going back to normal "The sudden change in my routine and then the sudden change in going back to my previous routine when this has ended. My threshold for external stimulus is non-existence now so I am fearful that, when I do return to work, I am going to be exhausted due to sensory overload on my commute to/from work and the social interaction is going to also be exhausting. I am not "masking" anymore and I am concerned I will have forgotten how to do that and how to hide my stimming. The reality is it is probably muscle memory at this point in my life but the exhaustion from external stimuli is worrying me" p. 977 "The relief of not having to mask anymore, but simultaneously the worry I will not be able to join back the rat race as quickly as the others. When things will open back up, expectations of enjoying the social get together will be super high and I will be enormously stressed and out of practice in acting normal" p. 579 "After a few weeks I have now gotten used to this calm pace, these new social rules slowly seem clear to me, so now it gives me a little anxiety that afterwards I will have to get used to the hustle and bustle and chaos again, and having to find my own way in this. " [Translated] p. 317 Social cohesion/ solidarity "As the entire nation is on lockdown, and everyone is self-isolating, I feel like I am finally at one with society. My way of living life is the norm. I no longer have to feel alone. I have a sense of belonging that everyone is in the same situation and we're all in this together" p. 629 "I notice that I find the contact with friends and my parents more pleasant now: I feel like my own and their daily worlds have more of a common denominator now. There's a difficult situation now, that others also intuitively experience as being difficult. (…) Furthermore, interactions are more frequent but shorter, maybe that's why I feel less like I have to dig into my reserves in order to lead the social life that I want. " [Translated] p. 668 Reduced sensory and social overload "These times are in many respects more autism-friendly. There's less car traffic and noise in my neighbourhood. There's less traffic jam frustration and fewer exhaust gases. Working from home is a nice experiment that the crisis made possible. It is less stressful, because I am not disturbed and distracted by colleagues. And because I can take breaks whenever I need it, without worrying about social control. And the stress from socially obligated small talk is no longer there. " [Translated] p. 505 "No more handshaking, no more people getting way too close and not feeling weird for staying at home all the time" p. 233 Lastly, we would like to acknowledge two additional points that this open question revealed. First, although not specific to the COVID-19 pandemic, many adults with autism revealed that they received crucial support from within the autism community as they share their experiences with one another. Second, despite it being a small minority, it should be noted that there were some adults with autism for whom living conditions under the COVID-19 pandemic had actually improved overall, with the main reason for improvement being the reduced sensory and social overload.

Discussion
This study investigated the psychological impact of the COVID-19 pandemic and the associated containment measures on adults with autism (as compared to adults without autism) across three countries. Specifically, we examined COVID-19 pandemic-related changes in mental health, issues related to the two autism symptom domains, satisfaction with pandemic-related information and tips, and participants' wishes for guidance. For this, we used a mixed-method approach, combining quantitative and qualitative data. Our results highlight the burden of the pandemic on the mental health and daily life of adults with autism, and give insight into how they can be supported.

Mental health
We found an increase in depression and anxiety symptoms in response to the pandemic for both the nonautism and the autism group. However, this increase in symptoms and the impact of these symptoms on everyday life was greater in the autism group. Moreover, an overstretched healthcare system paired with social distancing rules meant that many adults with autism lost some, if not all of the support they received before the pandemic, as evident from both our quantitative and qualitative results. These results are particularly worrisome considering that adults with autism ordinarily have higher rates of anxiety and depression than neurotypicals [18,19], as was also reported by participants in our sample, and struggle to access mental health services [43,44]. These results are in line with a recent report from the National Autism Society [27], where 9 out of 10 people with autism indicated to worry about their mental health during lockdown. We further found that the autism group worried significantly more than the non-autism group about their own safety/security, pets, work (even when they felt their job was safe), losing their job, and getting medication and food. The increase in worry concerning their pets in adults with autism is interesting as a previous pre-pandemic study by Dachez and Ndobo [45] found that some adults with autism seek support from pets as a coping strategy. The increase in work-related worries is concerning but not surprising, as autism was already associated with poor employment outcomes pre-pandemic [46]. Our quantitative results on getting medication and food were supported by our qualitative data. Shopping for groceries was by far the most often reported anxiety-provoking topic for adults with autism, as revealed by the qualitative findings, with some also particularly highlighting their worries about the reduced accessibility of medication. Shopping was described as more stressful due to the social distancing rules and disrupted shopping routines. To our knowledge, no study as of yet has specifically investigated everyday life topics of worries in autism, so it remains unknown whether adults with autism for instance also worried more about access to food or medication pre-pandemic. Research efforts aimed to learn about everyday life topics of worries in adults with autism would not only be worthwhile in light of our current results to determine if post-pandemic worries are different from those that increased during the pandemic, but may also add to the mood and anxiety literature on autism.

Social life
Further, we found that the non-autism group, more than the autism group, felt the pandemic changed their social life, and consequently felt more socially isolated. These results may be explained by previous research showing that adults with autism were already more likely than neurotypicals to face loneliness and social isolation before the pandemic (e.g. [47,48]), making the change to their social life caused by the pandemic less pronounced for them. Still, the difficulty most often reported by both groups was the loss of social contact. With autistic individuals already more likely than neurotypicals to face loneliness and social isolation, losing the access to their, usually already smaller (e.g., [47,48]), support network appears to be a great burden on adults with autism. At the same time, we also found that the autism group felt more relieved from social stress compared to the nonautism group. This relief from certain social stressors was in line with our qualitative results. However, the qualitative results also brought to light that the pandemic came with new social challenges, such as difficulty reading facial expressions because of face masks.

Routines
On average, the autism group felt more stressed than the non-autism group about the changes to their daily routines that the pandemic is causing. This finding was validated by our qualitative findings as many adults with autism mentioned the loss of routine to be one of the most difficult pandemic-related changes. Moreover, this finding is in accord with the diagnostic criteria (i.e., domain II) of autism (e.g., difficulty with transitions; [49]), and explains the voiced need for help with creating a new routine. Our qualitative findings further reveal that the stress induced by the disruption of routines also translates to prospective anxiety. That is, many adults with autism report experiencing anxiety due to the prospect of the ever-changing pandemic-related rules and the accompanying need to repeatedly adjust their routines.

Implications
Our findings give important insights into how adults with autism can be supported in order to mitigate further mental health problems during the ongoing COVID-19 pandemic and possible future public health crises. Foremost, the majority of adults with autism find it important that they themselves are being consulted for the development of COVID-19 pandemic tips and tools. We therefore recommend collaborations with adults with autism in endeavors aimed at supporting them during the COVID-19 pandemic. Input from adults with autism may also increase satisfaction with the COVID-19 pandemic information and advice, and the tips and tools offered. COVID-19 pandemic information and advice should be clear and coherent, and widely promoted so that it reaches its target audience. Moreover, as our results indicate that most autism-specific information is directed towards parents with children with autism, we encourage efforts into the development of autism-specific information and advice specifically tailored to adults with autism.
In addition, ensuring continued, affordable, and accessible support at this time should be of the utmost priority. This is especially important given the negative impact of the pandemic on the mental health of the majority of adults with autism, and because the number of unmet formal support needs is found to be negatively associated with the quality of life of adults with autism [50]. The impact of the pandemic on for instance the social life and routines of adults with autism is inevitable. However, supporting them in the maintenance of a social network (which can be online as well), in creating alternative routines, and in updating those routines to the ongoing changes could soften the blow. Regarding social support, our results revealed that adults with autism received crucial support from within the autism community, which was previously identified as a coping strategy among adults with autism [45]. Therefore, stimulating adults with autism to engage in social contact within the autism community may be advisable.
When offering remote support, one should preferably provide various communication media as some adults with autism noted to experience anxiety on voice or video calls and therefore prefer text-based messaging. Anxiety experienced over voice or video calls in adults with autism was also previously found (see also for an elaborate discussion of the needs of adults with autism in video calling, [51]). Our results revealed that some adults with autism preferred face-to-face, whereas others commended the accessibility of remote online support. There is still abundant room for further research in determining for whom online healthcare may be beneficial, and how to provide effective remote mental health support to adults with autism during and after the COVID-19 pandemic [52].
Furthermore, to alleviate the anxiety experienced during shopping, one could advertise specific shopping times as autism-friendly, or a buddy system, pairing adults with autism with a buddy who can shop for them, could be introduced.
Lastly, we also found some COVID-19 pandemicrelated changes that were experienced as pleasant, specifically the feeling of social solidarity, and reduced sensory and social overload. For some adults with autism, the reduction in sensory and social overload even led to overall improved living conditions during the COVID-19 pandemic. Although only the small minority of our autism sample preferred the COVID-19 pandemic living conditions, this group should not be overlooked when drawing conclusions about the best ways of taking action. On the contrary, in order to create a more inclusive society, we should learn from these positive experiences that emerged during the COVID-19 pandemic. It would be worthwhile for future research to investigate how we could promote continuity of social solidarity and reduced sensory and social overload post-pandemic.

Strengths and limitations
To our knowledge, this study is the first to investigate the impact of the COVID-19 pandemic and the associated containment measures on adults with autism. Strengths of this study include the inclusion of a control group, the inclusion of multiple countries, the mixed-method approach of qualitative and quantitative data gathering, and the timing of the study, as it was conducted in the very first weeks of lockdown. As the pandemic undeniably has an impact on the whole population, the comparison with adults without autism enables us to gain insight into the specific impact of the pandemic on adults with autism. Data was gathered from three countries (Belgium, the Netherlands, and the UK). We controlled for country-related variation in the analyses and therefore our findings apply to adults with autism across these countries. Including qualitative data allowed us to gather more comprehensive data, to validate our quantitative findings, and to give a voice to the participants to ensure our findings are grounded in their experiences.
Furthermore, the timing of the study gives unique insight into the experience of adults with autism in the very beginning of the pandemic.
The current study also has a number of limitations. First, as with many online survey studies, our sample is likely to reflect some degree of selection bias. Particularly, we expect some underrepresentation of extreme cases, that is, people that were either minimally or very affected by the pandemic. Second, the findings of this study are limited to the immediate experiences in the beginning of the pandemic, and are therefore restricted to the shortterm effects of the pandemic. Future research should also examine the long-term effects of the pandemic, especially now that it is becoming more evident that the pandemic is likely to have a prolonged course. Third, generalizing our findings should be done with care. As mentioned, country of residence only explained a very small part of the variance (± 2.5%) in our study that included residents of Belgium, the Netherlands, and the UK. Still, the findings may not generalize to other countries in-or outside Europe. It is also worth mentioning that participants with Belgian residence were overrepresented in our sample in comparison to the other two countries. Our findings may also not be generalized to other subgroups of individuals with an autism diagnosis, such as children and individuals with intellectual disability. Future research could also focus on more narrow subgroups within the adult population, as we took a rather inclusive and broad approach (e.g., the age ranged between 18 and 81 years). It is important to note that although the groups did not differ on common matched-for variables such as age, gender ratio, and years of education (taken as a rough proxy for SES [53]), there were differences between groups on other variables such as pre-existing mental health difficulties and employment status, with adults reporting more mental health difficulties and lower employment status. However, these differences are representative of what is generally found in the autism population [18,19,46]. Moreover, this makes the greater pandemic-related increase in mental health problems as found for adults with autism even more worrisome. Lastly, we want to emphasize that inherent to group comparisons, the findings may not apply to all individuals in our sample as there is undoubtedly variability in how people with and without autism experience the pandemic.

Conclusion
The results of our study highlight the burden of the pandemic on adults with autism and shed light on how to support them during this COVID-19 pandemic. The greater impact of the pandemic on the mental health of adults with autism emphasizes the need for accessible, affordable (continued) support from health services to manage their mental health. Guidance may focus on the maintenance of a social network, and adjusting routines to the rapid ongoing changes. Further, elevating shopping-related anxiety may help adults with autism cope during this pandemic. Lastly, there is a need for COVID-19 pandemic-related tips, tools and information tailored to adults with autism that are created in collaboration with them. We hope that the findings of this study will increase awareness about the negative impact of the COVID-19 pandemic on adults with autism, and that our recommendations will be translated by the relevant institutions into targeted, helpful strategies. Note: In case not applicable choose 'not more than usual' Scale: significantly less than usual-slightly less than usual-not more than usual-slightly more than usual-significantly more than usual