COVID-19 and Perceived Changes to Quality of Life, Anxiety, Depression, and Loneliness in Autistic and Other Neurodivergent U.K. Adults

Abstract

Background:

In the United Kingdom, we have experienced many changes to our daily lives as a result of COVID-19. Autistic and other neurodivergent (ND, e.g., those with attention-deficit hyperactivity disorder) adults may be more vulnerable to negative effects of the pandemic due to pre-existing mental health disparities and unmet support needs. Furthermore, there is little research, either pandemic related or otherwise, which considers how the experiences of autistic adults with additional intersecting ND identities might differ from those without.

Methods:

We collected data from an online survey during June 2020 to September 2020 to explore the psychological impact of the coronavirus pandemic on U.K. adults (N = 286, age 18–72 years). Participants included neurotypical (NT) adults (N = 98), autistic adults (N = 73), other ND adults (N = 53), as well as autistic adults with an additional intersecting ND identity (N = 63). We measured and compared levels of quality of life (QoL), depression, anxiety, and loneliness across groups as well as perceived change in these as a result of the pandemic.

Results:

Autistic adults, with and without additional ND identities, had consistently low QoL and high anxiety, depression, and loneliness compared with NT adults. We found no differences in these areas between autistic adults with and without additional intersecting ND identities. In some areas, non-autistic ND participants were also doing poorly compared with their NT peers. Many participants felt that their QoL, mental health, and loneliness had worsened due to the pandemic, and this was largely similar across groups.

Conclusions:

These results highlight that COVID-19 may have led to increased need and demand for mental health services across the U.K. adult population. Both autistic and ND adults may be in particular need of increased (and improved) mental health and well-being support. This is likely because of pre-existing differences in mental health and well-being as well as individuals facing further difficulties as a result of the pandemic.

Community brief

Why is this an important issue?

The coronavirus pandemic has been difficult for many people. Some researchers have found that the pandemic may have been especially difficult for autistic adults and those who are neurodivergent in another way. This might be because autistic and neurodivergent people often experience poor mental health and have a variety of unmet needs.

What was the purpose of this study?

The purpose of this study was to explore quality of life, depression, anxiety, and loneliness in different groups of adults during the pandemic. We also wanted to explore whether people felt that these had become worse during the pandemic. We were interested to explore differences between autistic adults and other groups of neurodivergent adults. This included autistic adults, autistic adults who were neurodivergent in another way, non-autistic neurodivergent adults, and a comparison group of adults who were not autistic or neurodivergent in any way.

What did the researchers do?

We recruited participants to take part in an online survey during June 2020 to September 2020. We advertised the study using social media and research websites. A total of 286 adults from the United Kingdom completed our survey.

What were the results of the study?

Autistic adults had consistently low quality of life and high anxiety, depression, and loneliness compared with the comparison group. This was the same regardless of whether the autistic adults were neurodivergent in another way too.

In some areas, non-autistic neurodivergent participants had lower quality of life than the comparison group. Their depression and loneliness scores were somewhere in between autistic participants' and the comparison group's. Many participants felt that their quality of life, mental health, and loneliness had worsened due to the pandemic. This was similar for participants in all the groups. However, there were also participants who felt much better due to coronavirus restrictions.

What do these findings add to what was already known?

We now know that autistic adults have experienced poor quality of life, mental health, and loneliness during the pandemic. We also found that there were no differences based on whether autistic adults were also neurodivergent in another way (e.g., an autistic adult with attention-deficit hyperactivity disorder). We also know that non-autistic neurodivergent adults have experienced low quality of life during this time. This is important because there has been very little research on other non-autistic neurodivergent adults' experiences during COVID-19 restrictions.

What are potential weaknesses in the study?

Most of our participants were white British and female. This means that our findings may not be relevant to all adults in the United Kingdom. Our study was carried out during the pandemic, which means that we do not know if these differences between the groups will continue to be true in the future.

How will these findings help autistic adults now or in the future?

We hope that these findings will help to argue for more support to be made available to promote good quality of life and reduce mental health difficulties for autistic, and other neurodivergent, adults. Lots of our participants felt that their quality of life and mental health had got worse due to the pandemic. Because of this, it is now even more important that governments make changes to policy and funding to provide better services and support for autistic and neurodivergent adults.

Introduction

Neurodiversity has come to refer to the natural variation in human neurocognitive functioning.¹ From this perspective, those with thoughts and behaviors within a perceived normative range can be considered neurotypical (NT), whereas those outside these perceived norms can be regarded as neurodivergent (ND).² Autistic people and those with attention-deficit hyperactivity disorder (ADHD), along with those with dyslexia, developmental coordination disorder, and others, are among those who can be described, or may identify, as ND.

The coronavirus (COVID-19) pandemic, and measures to limit its spread, have caused changes to the everyday lives of many. Leading experts have called for vital research assessing the psychological impact of the pandemic, especially for those who may be more likely to suffer adverse effects.³ Already, research has demonstrated that the pandemic is associated with self-reported deterioration in mental health for many in the United Kingdom,⁴ and with an increase in the prevalence of anxiety and depression worldwide.⁵

Autistic and other ND adults may be more vulnerable to negative psychological effects of the pandemic. Pre-pandemic research has established that autistic adults commonly experience increased rates of mental health difficulties,^6–8 lower quality of life (QoL),^9,10 and increased loneliness¹¹ compared with their NT peers. Lower baseline well-being for autistic people is concerning, as pandemic-related psychological deterioration could further reduce levels of this.

Indeed, COVID-19 era online surveys have reported negative psychological effects of the pandemic on autistic adults. One longitudinal study of 636 U.S. autistic adults found that many experienced high levels of psychological distress,¹² and another study, which included 12 autistic adults, found that a majority reported high levels of stress during the pandemic.¹³ Several online studies have further reported that many autistic adults feel that their mental health has suffered due to the pandemic. This was found by 2 U.K. surveys with autistic adults,^14,15 and a large survey completed by 613 autistic adults and 431 non-autistic adults from Belgium, the Netherlands, and the United Kingdom.¹⁶

COVID-19 research has also highlighted differences in the levels of distress. For example, an Italian survey comparing 45 autistic adults and 45 non-autistic adults found that autistic adults reported higher levels of depression, anxiety, stress, and trauma during the first 2 months of the pandemic compared with non-autistic adults.¹⁷ Furthermore, COVID-19 seems to have reduced community participation among autistic adults,¹⁸ and pandemic-related social isolation is thought to have had a serious negative impact on the mental health and well-being of autistic people.¹⁹

While a handful of studies have explored the psychological impact of the pandemic among autistic adults, there is less research that focuses on adults who may be ND in other ways, and the studies which are available focus on adult ADHD. A small online survey of 24 U.K. adults with ADHD concluded that adults with ADHD were experiencing significant levels of distress during the pandemic.²⁰ Similarly, a large survey of 2055 Israeli adults found that increasing ADHD traits were related to poorer mental health during this period.²¹ However, neither of these studies had information on pre-COVID-19 baseline well-being or asked participants to reflect on how much they felt the COVID-19 pandemic had contributed to their current mental health or emotional distress. However, a small qualitative interview study of four Japanese adults with ADHD found that several reported increases in negative emotions and everyday difficulties related to the pandemic.²²

It is well known that autism and other forms of neurodivergence, for example, ADHD, commonly overlap.²³ Thus, an adult who is ND in one way is also likely to be ND in another way and have intersecting ND identities (i.e., they could be described as “multiply neurodivergent”²⁴). Before the pandemic, a large-scale study using Norwegian registry data identified that adults with diagnoses of both autism and ADHD are more likely to experience several co-occurring mental health difficulties, including anxiety disorders, bipolar disorders, and personality disorders, compared with those who are either autistic or have ADHD alone.²⁵

Adult research into the experiences of those with intersecting autistic and other ND identities is lacking generally, and in relation to the pandemic. We are aware of no published research considering and comparing the psychological impact of the pandemic among autistic and other ND adults, including those who are multiply-ND. Therefore, we do not know whether those with intersecting autistic and other ND identities have experienced an even greater psychological impact from the pandemic.

The current study aimed to provide information about hitherto unresearched groups in the hopes of identifying those who might benefit most from support during and after the current pandemic, determining in what areas support might be most helpful, and informing priorities for future potential health crises. To do this, we examined differences in the mental health, QoL, and loneliness of four groups of U.K. adults during the COVID-19 pandemic: autistic (Aut); other ND, for example, ADHD (ND); autistic plus another neurodivergent identity (Aut+ND); and NT comparison adults. We hypothesized that the NT group would report higher QoL and lower levels of anxiety, depression, and loneliness than all other groups. We expected that in some areas, Aut+ND participants would report the lowest levels of QoL, anxiety, depression, and loneliness.

We also examined differences between the groups in the perceived impact of the pandemic, specifically on QoL, anxiety, depression, and loneliness. In line with findings that autistic adults may have experienced greater changes to their mental health due to the pandemic compared with non-autistic adults,¹⁶ we hypothesized that in some areas, autistic/ND participants would report a greater deterioration in contrast to NT comparison adults.

Methods

Design and procedure

The Quality of Life During COVID-19 (QoLVID) study used a cross-sectional online survey design. We designed and hosted the survey using Qualtrics software,²⁶ which included established questionnaire measures, bespoke items, and free-text response questions. An autistic researcher (D.M.) was part of the research team who conceived and designed the study. The King's College London Psychiatry, Nursing and Midwifery Research Ethics Subcommittee granted ethical approval for this study (Ref: HR-19/20-18279, June 11, 2020).

We advertised the study using multiple online and social media channels (Twitter, Facebook, LinkedIn, Reddit, Autistica Network, MQ participate, Call for participants, and University Recruitment Circulars) to quickly recruit a neurodiverse sample of U.K. adults as the COVID-19 pandemic unfolded in the United Kingdom. Study adverts described that the project aimed “to examine the changes to daily life that people have experienced relating to COVID-19 and how these might have influenced their well-being and mental health.” These adverts also described that we were seeking “autistic, other ND (e.g., ADHD) and neurotypical adults” and relatives of autistic adults to take part in the study. Participants viewing the adverts were able to access study information via an online link, before providing informed consent to participate and for the use of their data for scientific publication. We offered participants the opportunity to enter a prize draw for a 1 in 10 chance of winning a £25 voucher.

The survey was open to collect data between June 11, 2020, and September 30, 2020. The U.K.'s first national lockdown was announced on March 23, 2020, which required school closures, people to stay home, and non-essential businesses to close. By June, schools were beginning to reopen (1st), as were non-essential shops (15th). Further easing of restrictions nationally, paired with regional lockdowns, took place between June and September until the need for further restrictions (including working from home) was announced on September 22, 2020.²⁷

Sample

The final sample included responses from 286 participants who provided demographic information and completed at least one additional measure. Two hundred sixty-nine participants completed the survey by self-report (age median = 29, range 18–72 years), and a further 17 individuals completed the survey on behalf of an autistic adult relative by proxy (age: relative responder, median = 58, range 18–69 years; autistic adult, median = 26, range 18–72 years). We invited proxy-responders to take part in the study to enable us to engage with the perspectives of autistic adults who cannot take part in survey research themselves (e.g., if they are unable to read/write, were too distressed by COVID-19 to engage in research). We found no significant difference in age between the autistic adults contributing data via a proxy-reporter and those completing the study by self-report (Mann–Whitney U, z = 1.85, p = 0.063).

Measures

Autism Spectrum Quotient 10

The Autism Spectrum Quotient 10 (AQ-10) is a short screening questionnaire assessing social and non-social features of autism spectrum conditions in adults.²⁸ For 10 statements, participants are asked to indicate their agreement on a 4-point Likert scale (“Definitely Agree,” “Slightly Agree,” “Slightly Disagree,” “Definitely Disagree”). For four items, a score of 1 is given for “Definitely” or “Slightly Agree” responses, for example, “When I'm reading a story I find it difficult to work out the characters' intentions.” For six items, a score of 1 is given for “Definitely” or “Slightly Disagree” responses, for example, “I usually concentrate more on the whole picture, rather than the small details.” Possible total scores range from 0 to 10, with higher scores indicating increasing autistic traits and a cutoff of 6 or more indicating probable autism. This measure was completed by all participants (N = 286), and internal consistency was α = 0.82.

Ritvo Autism and Asperger Diagnostic Scale-14

The Ritvo Autism and Asperger Diagnostic Scale-14 (RAADS-14) is a short 14-item screener designed to identify adults who may have undiagnosed autism.²⁹ For each statement, participants are asked to rate whether and when they believe the statement to have been true to them, using a 4-point Likert scale. For 13 items, agreement indicates greater levels of autistic characteristics, for example, “I get extremely upset when the way I like to do things is suddenly changed.” For these, scores of 0–3 are given for each item (3 = “True now and when I was young,” 2 = “True only now,” 1 = “True only when I was younger than 16,” 0 = “Never true”). A single item, “I can chat and make small talk with people,” is reverse-scored (0 = “True now and when I was young,” 3 = “Never true”). Summing item scores creates a possible total score of 0–42, with higher scores reflecting increasing autistic traits, and a cutoff of 14 or more indicating possible autism. This measure was completed by all participants (N = 286), and internal consistency was α = 0.92.

We chose to use two measures of autistic traits for two main reasons. First, the AQ-10 and RAADS-14 are commonly used in autism research studies. Due to the unique nature of our sample and our participant groups, we wanted to select commonly used measures to allow for comparison with existing research. Despite their common use, there are thought to be several drawbacks to both the AQ-10 and the RAADS-14. We therefore chose to employ both these measures to balance the limitations and criticism of each of these measures separately.

Adult ADHD Self-Report Scale for DSM-5

The Adult ADHD Self-Report Scale for DSM-5 (ASRS-5) is a 6-item self-report screening tool for adult ADHD.³⁰ Each statement asks participants to consider their feelings or behaviors over the last 6 months, for example, “How often do you put things off until the last minute?” Participants rate each item using a 5-point Likert scale, with responses scored 0–4 (0 = “Never,” 1 = “Rarely,” 2 = “Sometimes,” 3 = “Often,” 4 = “Very Often”). Summing all item scores creates possible total scores of 0–24, where higher scores reflect higher levels of ADHD traits. This measure was completed by all participants (N = 286), and internal consistency was α = 0.72.

World Health Organization Quality of Life–Bref

The World Health Organization Quality of Life–Bref (WHOQOL-BREF) is a 26-item self-report measure assessing participant-perceived QoL over the past 2 weeks,^31,32 which has previously been used and validated in autistic adults.³³ The questionnaire consists of 2 global summary questions and an additional 24 items across 4 subdomains (physical, 7 items; psychological, 6 items; social, 3 items; environmental, 8 items). Question phrasing varies, for example, “How often/much/satisfied …,” and a corresponding 5-item response scale is used for each item, for example, “Very dissatisfied,” “Dissatisfied,” “Neither satisfied nor dissatisfied,” “Satisfied,” and “Very satisfied.”

We converted raw sum scores to 0–100 transformed scores for each subdomain in line with standard procedures.³¹ Physical, psychological, and environmental domain items were completed by 270 participants with internal consistency of each domain α = 0.79, 0.84, and 0.79, respectively. Social domain items were completed by 269 participants, and internal consistency of this domain was α = 0.68.

Anxiety Scale for Autism–Adults

The Anxiety Scale for Autism–Adults (ASA-A) is a self-report measure of anxiety specifically designed for use with autistic adults.³⁴ We decided to use this autism informed anxiety measure in recognition that general population anxiety measures may not cover some aspects of anxiety that are particularly relevant to autistic adults.³⁴ It comprises 20 items asking about the experience of anxiety over the past 2 weeks, for example, “I feel anxious in situations where I could make a mistake.” Participants rate each statement using a 4-point Likert scale, scored 0–3 (0 = “Never,” 1 = “Sometimes,” 2 = “Often” 3 = “Always”). Summing all scores creates a general anxiety total score (possible scores 0–60). We used the suggested cutoff of 28 or more to identify those with significant anxiety levels. The measure was completed by 263 participants, and internal consistency was α = 0.95.

Patient Health Questionnaire 9-Item

The Patient Health Questionnaire 9-Item (PHQ-9) is a commonly used self-report screening questionnaire for depression symptoms,^35,36 which has been recently validated in autistic adults.³⁷ Participants respond to 9 items asking about symptoms they may have experienced over the last 2 weeks, for example, “Little interest or pleasure in doing things.” For each statement, participants rate how often each applies to them using a 4-point Likert scale (0 = “Not at all,” 1 = “Several days,” 2 = “More than half the days,” 3 = “Nearly every day”). Item scores are summed to create a continuous total score ranging from 0 to 27, with higher scores indicating greater depression symptom severity. We used a cutoff of 10 or greater to identify those with moderate-to-severe levels of depression (compared with none/mild). An independent meta-analysis indicated that using a single cutoff of between 8 and 11 for this measure is acceptable in identifying major depression.³⁸ The measure was completed by 264 participants, and internal consistency was α = 0.88.

University of California, Los Angeles, 3-Item Loneliness Scale

The University of California, Los Angeles, 3-Item Loneliness Scale (UCLA-3) is a short measure of loneliness consisting of three items.³⁹ Items include “How often do you feel that you lack companionship?” “How often do you feel left out?” “How often do you feel isolated from others?” Participants rate how often they feel each item applies to them using a 3-point Likert scale (1 = “Hardly ever,” 2 = “Some of the time,” 3 = “Often”). Scores from the three items are summed to create a total score ranging from 3 to 9, with higher scores indicating higher levels of loneliness. We used cutoff of 6 or more to identify those likely to be “lonely,” as has been used in previous research.⁴⁰ The measure was completed by 265 participants, and internal consistency was α = 0.81.

Bespoke items on participant perceived change

Following the ASA-A, PHQ-9, and the UCLA-3 scales, and for each domain of the WHOQOL-BREF, participants were asked questions about their perceived change in anxiety, depression, loneliness, and QoL due to the pandemic. For example, following the PHQ-9 depression scale, participants were asked, “In relation to the feelings and problems we have just asked about, do you think these have changed because of COVID-19?” Responses were recorded using a 5-point Likert scale (“In general these have become much better,” “In general these have become a little better,” “There have been no changes,” “In general these have become a little worse,” “In general these have become much worse”).

Exact wording of the bespoke change question varied for each measure, for example, “In relation to your feelings of loneliness,” “In relation to your experience of worry, anxiety or the physical sensations we just asked about.” For bespoke change, items relating to each domain of the WHOQOL-BREF participants were also given a brief summary and definition of these domains before answering.

Participant groups

We assigned participants to one of four groups based on how they self-identified or the diagnoses they reported. In the case of adults for whom the survey was completed by proxy, this information was provided by the relative/carer completing the survey. We made the decision to include participants who self-identified as autistic or ND based on the assumption that not all individuals who might meet the diagnostic criteria would have received a formal diagnosis (e.g., due to long waiting lists for assessment) or would want one (e.g., due to perceived stigma or adverse experiences with health care providers).

The NT group consisted of individuals who reported that they were not autistic and did not identify as ND in any way (N = 98). The Aut group included diagnosed or self-identified autistic adults who did not feel that they were ND in any additional way (N = 73). The ND group included non-autistic adults who considered themselves ND in some other way (N = 52). The Aut+ND group included diagnosed/self-identified autistic adults who felt that they were also ND in some other way (N = 63).

Participants who identified as ND cited various factors, including dyspraxia, dyslexia, obsessive-compulsive disorder, bipolar disorder, and epilepsy (further information is given in Supplementary Table S1). However, most ND participants reported a diagnosis or self-identification of ADHD; this was true in both the ND group (73.10% of this group cited ADHD) and the Aut+ND group (58.73%).

Scores from autistic and ADHD trait measures (described above) have been used to establish whether self-identified versus formally diagnosed participants, and participants by self-report versus proxy-report, were sufficiently similar in terms of these relevant characteristics to justify combining data. We found no differences in autistic or ADHD traits between formally diagnosed and self-identified participants and no difference between scores for self-report and proxy-report participants (reported in Supplementary Table S2a, S2b). From this, we made the decision to collapse groups to include participants who self-identified or had formal diagnoses and to combine data from participants with self-reports and proxy-reports.

Table 1 presents descriptive and demographic information across groups and in the entire sample. This includes details on participant formal diagnoses, sex, ethnicity, education level, co-occurring conditions, and age, as well as levels of autistic and ADHD traits.

Table 1.

Descriptive and Demographic Information of Participants, Across Groups and in the Full Sample

Characteristic/measure	NT, N = 98	ND, N = 52	Aut, N = 73	Aut+ND, N = 63	Total, N = 286	Statistical comparisons
	N, %
Formal autism Dx	—	—	55	44	—	2 × 2 chi-square test of independence χ2(1) = 0.52, p = 0.472, ϕc = 0.06
	—	—	75.34%	69.84%	—
Other formal ND Dx	—	38	—	47	—	2 × 2 chi-square test of independence χ2(1) = 0.03, p = 0.853, ϕc = 0.02
	—	73.08%	—	74.60%	—
Sex
Female	84	39	50	35	208	3 × 4 chi-square test of independence with Fisher's exact test χ2(6) = 22.49, p = 0.001, ϕc = 0.11 Fisher's exact p = 0.001
	85.71%	75.00%	68.49%	55.56%	72.73%
Male	12	8	13	21	54
	12.24%	15.38%	17.81%	33.33%	18.88%
Other	2	5	10	7	24
	2.04%	9.62%	13.70%	11.11%	8.39%
Ethnicity
Any White background	86	49	67	59	261	2 × 4 chi-square test of independence with Fisher's exact rest χ2(3) = 2.56, p = 0.464, ϕc = 0.05 Fisher's exact p = 0.514
	87.76%	94.23%	91.78%	93.65%	91.26%
Any other ethnic background	12	3	6	4	25
	12.24%	5.77%	8.22%	6.35%	8.74%
Education
Completed higher education qualification	64	33	42	37	176	2 × 4 chi-square test of independence χ2(3) = 1.37, p = 0.712, ϕc = 0.04
	65.31%	63.46%	57.53%	58.73%	61.54%
Co-occurring conditions
Any psychiatric Dx	40	35	47	31	153	2 × 4 chi-square test of independence χ2(3) = 14.27, p = 0.003, ϕc = 0.13
	41.84%	67.31%	64.38%	49.21%	53.50%
Any physical health Dx	16	10	20	25	71	2 × 4 chi-square test of independence χ2(3) = 12.38, p = 0.006, ϕc = 0.12
	16.33%	19.23%	27.40%	39.68%	24.83%
Age
Mean	31.65	30.90	32.71	34.06	32.32	Kruskal–Wallis χ2(3) = 2.13, p = 0.546, ϕc = 0.05
Standard deviation	11.08	9.88	11.59	12.57	11.35
Median	28.00	28.50	31.00	30.00	29.00
Interquartile range	14.00	13.50	15.00	15.00	14.00
AQ-10
Mean	2.90	4.67	7.55	7.67	5.46	Kruskal–Wallis χ2(3) = 145.79, p < 0.001, ϕc = 0.41 The NT group had significantly lower AQ-10 scores than all other groups; the ND group had significantly lower scores than the Aut and Aut+ND groupsa
Standard deviation	2.15	2.29	2.03	1.88	2.98
Median	3.00	4.00	8.00	8.00	5.50
Interquartile range	3.00	4.00	3.00	3.00	5.00
RAADS-14						Kruskal–Wallis χ2(3) = 167.08, p < 0.001, ϕc = 0.44 The NT group had significantly lower RAADS-14 scores than all other groups; the ND group had significantly lower scores than the Aut and Aut+ND groupsa
Mean	9.99	19.38	32.15	33.14	22.45
Standard deviation	8.73	8.87	8.40	7.70	13.24
Median	8.00	20.00	34.00	35.00	25.00
Interquartile range	10.00	14.00	12.00	11.00	24.00
ASRS-5
Mean	8.40	15.17	12.56	14.86	12.12	One-way ANOVA F (3,282) = 44.96, p < 0.001, η2 = 0.32 The NT group had significantly lower ASRS-5 than all other groups; the Aut group had significantly lower scores than the ND and Aut+ND groupsa
Standard deviation	4.22	4.58	3.40	4.46	5.03

Comparison of demographic and descriptive information across participant groups. Bold font denotes overall significant model.

ϕc = Crammer's V effect size calculated by √(χ²/n × df). η² = eta-squared calculated using STATA esizei command.

^a Follow-up pair-wise comparisons significant after Bonferroni correction.

ADHD, attention-deficit hyperactivity disorder; AQ-10, Autism Spectrum Quotient 10; ASRS-5, Adult ADHD Self-Report Scale for DSM-5; Aut, autistic; Aut+ND, autistic and neurodivergent in another way; Dx, diagnosis; ND, neurodivergent, but not autistic; NT, neurotypical; RAADS-14, Ritvo Autism and Asperger Diagnostic Scale-14.

Statistical analysis

We analyzed the study data using STATA Release 16.⁴¹ We visually inspected all noncategorical variables using histograms before inferential analyses. We used a significance threshold for overall models of p = 0.05, and Bonferroni corrected significance levels in follow-up pair-wise tests.

For comparisons between participant groups (NT, ND, Aut, Aut+ND), we used one-way ANOVAs for approximately normally distributed variables. We used corrected t-tests following significant overall ANOVA models. We examined non-normally distributed variables using the Kruskal–Wallis tests followed by corrected Dunn's tests where appropriate.

We collapsed scores from bespoke perceived change items into three categories for analysis (Worse, No Change, Better). To assess whether participant group was related to perceived change in QoL domains, anxiety, depression, or loneliness, we conducted a series of 3 × 4 way chi-square tests. Following an overall significant χ², we used the adjusted standardized residual method.⁴² For 3 × 4 way contingency tables, residuals of ±2.00 indicate results significant at p < 0.05 and residuals of ±2.90 indicate cells significant results after correction (i.e., p = 0.05/12 = 0.00416). We collapsed perceived change scores into three categories for two main reasons: (a) reducing the number of categories meant that all cells had expected frequencies of five or more, so that standard chi-square analyses could be used, and (b) the reduction of categories improved the likelihood of identifying significant cell-wise effects after Bonferroni correction (i.e., as fewer cells are corrected for).

Results

In the results presented below, we have included data from all participants (i.e., self-responders and proxy-responders). We have presented effect sizes for pair-wise comparisons of all scale variables in Supplementary Table S3.

Group differences in QoL

Table 2 shows average QoL scores and group differences. In analyses for all domains, we found that there were significant overall differences in scores between the groups. For physical, social and environmental QoL, participants in the Aut, Aut+ND, and ND groups had significantly lower scores than the NT group, with no other differences between the groups. In the psychological domain, participants in the Aut and Aut+ND groups had lower QoL than NT participants and did not differ from one another. Furthermore, the ND group had intermediate scores with no significant differences from other groups.

Table 2.

Differences in Quality-of-Life Scores Across Groups

WHOQOL-BREF domain	NT	ND	Aut	Aut+ND	ANOVA
Physical
Mean	65.55	52.80	52.89	50.89	F (3, 266) = 9.86, p < 0.001, η2 = 0.10 The ND, Aut, and Aut+ND groups had significantly lower physical QoL than the NT groupa
Standard deviation	19.23	19.26	21.30	16.36
Min	7.14	7.14	3.57	3.57
Max	100.00	85.71	89.29	78.57
Psychological
Mean	50.18	41.09	37.08	39.29	F (3, 266) = 6.32, p = 0.004, η2 = 0.07 The Aut and Aut+ND groups had significantly lower psychological QoL than the NT groupa
Standard deviation	21.22	23.26	20.61	17.69
Min	0.00	0.00	0.00	0.00
Max	100.00	87.50	79.17	79.17
Social
Mean	62.81	51.80	50.48	49.48	F (3, 265) = 6.29, p = 0.004, η2 = 0.07 The ND, Aut, and Aut+ND groups had significantly lower social QoL than the NT groupa
Standard deviation	20.10	24.17	23.57	21.93
Min	16.67	0.00	0.00	0.00
Max	100.00	100.00	100.00	100.00
Environmental
Mean	65.59	54.23	55.49	51.90	F (3, 266) = 9.27, p < 0.001, η2 = 0.09 The ND, Aut, and Aut+ND groups had significantly lower environmental QoL than the NT groupa
Standard deviation	16.89	17.97	18.67	17.03
Min	18.75	15.63	6.25	3.13
Max	100.00	90.63	87.50	81.25

Comparison of QoL domains scores across participant groups. Bold font denotes overall statistically significant difference between the groups.

η² = eta-squared calculated using STATA esizei command.

^a Follow-up pair-wise comparisons significant after Bonferroni correction.

QoL, quality of life; WHOQOL-BREF, World Health Organization Quality of Life–Bref.

Group differences in anxiety scores

Table 3 shows group differences in average anxiety scores and percentages of participants meeting anxiety cutoffs (the table also presents differences in depression and loneliness scores and the percentages of participants meeting cutoffs). Overall, we identified a significant difference in anxiety scores across groups. Specifically, participants in the Aut and Aut+ND groups had significantly higher scores than the NT and ND groups. Our analyses identified that anxiety scores of participants in the Aut and Aut+ND groups did not differ from each other. Similarly, we found no differences between scores of participants from the NT and ND groups.

Table 3.

Differences in Anxiety, Depression, and Loneliness Scores Across Groups

Measure	NT	ND	Aut	Aut+ND	Statistical comparisons
Anxiety (ASA-A)
Mean	20.94	26.16	35.16	33.69	One-way ANOVA F (3, 259) = 21.89, p < 0.001, η2 = 0.20 The NT and ND groups had significantly lower anxiety scores than the Aut and Aut+ND groupsa
Standard deviation	12.45	11.65	13.58	10.92
Min	0.00	1.00	1.00	5.00
Max	47.00	57.00	60.00	57.00
N above cutoff	26	20	51	36
% of group	28.57	40.82	73.91	66.67
Depression (PHQ-9)
Mean	10.99	13.60	15.30	15.51
Standard deviation	6.97	7.05	6.61	5.46
Median	10.00	13.00	15.00	16.00	Kruskal–Wallis χ2(3) = 21.18, p < 0.001, ϕc = 0.16 The NT group had significantly lower depression scores than the Aut and Aut+ND groupsa
Interquartile range	11.00	11.50	11.00	9.00
Min	0.00	1.00	0.00	5.00
Max	24.00	27.00	27.00	26.00
N above cutoff	46	33	53	49
% of group	50.55	68.75	75.71	89.09
Loneliness (UCLA-3)
Mean	5.54	6.10	6.74	6.75
Standard deviation	1.94	2.05	1.77	1.69
Median	6.00	6.00	7.00	7.00	Kruskal–Wallis χ2(3) = 19.64, p < 0.001, ϕc = 0.16 The NT group had significantly lower loneliness scores than the Aut and Aut+ND groupsa
Interquartile range	3.00	4.00	2.00	2.00
Min	3.00	3.00	3.00	3.00
Max	9.00	9.00	9.00	9.00
N above cutoff	47	32	55	43
% of group	51.65	65.31	78.57	78.18

Comparison of anxiety, depression, and loneliness across participant groups. Bold font denotes overall statistically significant difference between the groups.

ϕc = Crammer's V effect size calculated by √(χ²/n × df). η² = eta-squared calculated using STATA esizei command.

^a Follow-up pair-wise comparisons significant after Bonferroni correction.

ASA-A, Anxiety Scale for Autism–Adults; PHQ-9, Patient Health Questionnaire 9-item; UCLA-3, University of California, Los Angeles, 3-Item Loneliness Scale.

Group differences in depression scores

Overall, we found that there was a significant difference in depression scores across groups. Participants in the Aut and Aut+ND groups did not differ from each other but had significantly higher depression scores than the NT group. In addition, participants in the ND group had intermediate scores, which did not significantly differ from any other group.

Group differences in loneliness scores

Overall, we found that there was a significant difference in loneliness across groups. Similar to the pattern found with depression scores, the Aut and Aut+ND groups did not differ from each other but had significantly higher loneliness scores than the NT group. Furthermore, participants in the ND group had intermediate scores, which did not significantly differ from any other group.

Experience of multiple above cutoff difficulties across groups

Venn diagrams in Figure 1 show the overlap of participants meeting cutoffs for anxiety, depression, and loneliness by group. We used the Kruskal–Wallis test to compare the number of above-threshold problems (0–3 from anxiety, depression, and loneliness) across groups. We found an overall significant difference in the number of above cutoff problems experienced by participants in each group: χ²(3) = 35.22, p < 0.001, ϕc = 5.93. Specifically, participants in the Aut and the Aut+ND groups did not differ from each other but had significantly more problems (medians = 3) than those in the NT group (median = 1). Furthermore, those in the ND group had an intermediate number of problems (median = 2), which was not significantly different from any other group.

FIG. 1.

Venn diagrams showing the proportions (with N) of participants meeting anxiety, depression, and loneliness cutoffs across groups. Darker shading represents greater proportions of participants from that group occupying that section of the Venn. Aut, autistic; Aut+ND, autistic and neurodivergent in another way; ND, neurodivergent, but not autistic; NT, neurotypical.

Perceived change in QoL, anxiety, depression, and loneliness

Perceived change in the sample overall

Figure 2 presents the percentages of participants reporting positive, negative, or no change in QoL, anxiety, depression, and loneliness across the entire sample. In general, we found that more participants perceived that their QoL, mental health, and loneliness had worsened—rather than improved—due to the COVID-19 pandemic.

FIG. 2.

Participant perceived change in QoL, anxiety, depression, and loneliness due to COVID-19 for the whole sample (N = 286).

QoL, quality of life.

Perceived change across groups

We used several 3 × 4 way chi-square tests of independence to determine whether the proportions of participants who perceived no changes, worsening, or improvement due to COVID-19 was related to participant group. Our analyses identified that perceived change due to the pandemic was unrelated to participant group for most variables. This was the case for perceived change ratings for: psychological QoL, χ²(6) = 6.89, p = 0.331, ϕc = 0.07; social QoL, χ²(6) = 2.47, p = 0.0.87, ϕc = 0.04; environmental QoL, χ²(6) = 6.38, p = 0.380, ϕc = 0.06; depression χ²(6) = 4.62, p = 0.593, ϕc = 0.05; and loneliness χ²(6) = 10.39, p = 0.109, ϕc = 0.08.

For the remaining two variables, we found that participant perceived change ratings were related to participant group: physical QoL χ²(6) = 16.63, p = 0.011, ϕc = 0.10; anxiety χ²(6) = 13.36, p = 0.038, ϕc = 0.09. Supplementary Figure S4 depicts perceived change ratings for physical QoL and anxiety across participant groups. For physical QoL, no individual cells met corrected significance thresholds. At the p < 0.05 level, we identified that a lower proportion of participants in the ND group felt that their physical QoL had not changed (14%, residual −2.05) compared with the other participant groups (29%–37%).

For anxiety, no individual cells met corrected significance thresholds. At the p < 0.05 level, we identified that a higher proportion of participants in the Aut+ND group felt that their anxiety had improved (24%, residual 2.49) compared with the other participant groups (7%–12%).

Results of comparisons using different subsets of study participants

We compared results from analyses with the total sample (N = 286) and self-responders only (N = 269) and found that patterns of significant findings were the same for 14 of 15 analyses performed (the 2 discrepancies are described in Supplementary Table S5a, S5b). From this, we concluded that the pattern of findings was largely unaffected by the inclusion of data from proxy-reporters.

As described previously, the majority of the ND and Aut+ND group participants reported a formal diagnosis or self-identification of ADHD. We compared results from analyses using the original sample groups (NT = 98, ND = 52, Aut = 73, Aut+ND = 63) with more restricted groupings focusing on autism and ADHD (i.e., excluding those with other ND identities, resulting in: NT = 98, ADHD = 38, Aut = 99, Aut+ADHD = 37). Patterns of significant findings were the same for 10 of 15 analyses performed (the 5 discrepancies are described in Supplementary Table S5). For the majority of analyses, we therefore assume that patterns of results have not been affected by the inclusion of participants with a broader range of ND identities.

Discussion

This is the first study to examine and compare the psychological impact of the COVID-19 pandemic on autistic adults, other ND adults, autistic adults with additional intersecting ND identities, and a NT comparison group. Our findings provide important insight into the QoL, mental health, and loneliness of ND adults at this time. Furthermore, this wider range of ND identities (and the intersectionality between autistic and other ND identities) is less well studied, even outside the pandemic context.

Differences in QoL, anxiety, depression, and loneliness between the groups

Overall, participants from both autistic groups (Aut and Aut+ND) did not differ and had the lower QoL and higher levels of anxiety, depression, and loneliness compared with the NT group. For physical, social, and environmental QoL, other ND adults also had lower QoL than the NT group, which was similar to levels among autistic adults. Those in the ND group had anxiety scores similar to the NT participants and lower scores than both autistic participant groups. The ND group had psychological QoL, depression, and loneliness scores intermediate between the NT and autistic groups. We also found that the comparison group also reported fewer above cutoff problems compared with both autistic adult groups.

Before the pandemic, research has shown that autistic adults have low QoL,^9,10 elevated rates of mental health difficulties,^6–8 and increased loneliness compared with non-autistic adults.¹¹ COVID-19 era studies have also identified that autistic adults have experienced greater levels of stress, anxiety, and depression than non-autistic adults.¹⁷ Given these findings, it is unsurprising that autistic adults reported the poorest well-being in every area measured in our study. However, we found no differences in anxiety, depression, loneliness, or QoL scores between autistic adults with and without additional neurodivergence (most often ADHD).

From our findings, this would suggest that autistic adults with additional intersecting ND identities experience no additional problems in relation to QoL, mental health, and loneliness compared with autistic adults without additional ND identities. This is somewhat surprising, as a study using Norwegian registry data found that adults with autism and ADHD had higher rates of anxiety disorders, bipolar, and personality disorders than those with autism or ADHD only.²⁵ Furthermore, emerging evidence from a preprint study of 628 autistic adults found that ADHD traits were significantly related to lower QoL in each domain, over and above the impact of autistic traits.⁴³

Discrepancies between our results and previous findings could have been due to a number of factors; however, without further investigation into this underresearched area, this is likely to remain unclear. First, we grouped our participants based on a range of self-identified ND identities and reported diagnoses. It is possible that the specific intersection of autism and ADHD, as studied previously, is associated with more pronounced differences in mental health and QoL compared with the intersection of autism and other ND identities more broadly. A second consideration is that our data were collected in the context of a global pandemic; thus, we cannot rule out that our findings do not fully represent difference between these groups before or following the pandemic period in the future.

Our participants from the ND group also had lower physical, social, and environmental QoL scores than NT participants, which was similar to both autistic groups. Given that the majority of this group comprised adults with ADHD, this is in line with the finding that increasing ADHD traits were associated with lower general life satisfaction in 2055 Israeli adults during the pandemic.²¹ This study also reported that ADHD traits were positively related to general psychological distress. In contrast, our findings indicated that ND participants were doing as well as the NT comparison group in terms of anxiety and were intermediate in their depression levels.

This discrepancy could be due to several reasons, for example, differences in measures, locale, traits versus groups' approaches, and statistical power. It is also possible that a significant degree of the relationship found between ADHD traits and psychological distress by Pollak et al.²¹ may have been driven by unmeasured autistic traits. Thus, we argue that it is imperative to consider autism, ADHD, or other forms of neurodivergence together to identify their independent effects on mental health and QoL. Research taking this approach could, in turn, help identify subgroups of ND adults who may benefit most from support in particular areas.

Perceived changes in QoL, anxiety, depression, and loneliness due to COVID-19

Overall, high percentages of participants reported a perceived negative change to their QoL (37%–68%), anxiety (57%), depression (61%), and loneliness (57%) due to COVID-19; in most cases, this appeared to be no different across the NT, ND, Aut, and Aut+ND groups. This finding was counter to our expectations, as other pandemic research has found that autistic adults were more likely to report worsening of anxiety, depression, and specific worries than non-autistic adults.¹⁶

This discrepancy may be because Oomen et al.¹⁶ employed a more detailed measurement of perceived change (e.g., adapting PHQ-9 items to assess perceived change and generating a sum score), whereas the current study used a single item to capture perceived change for each area assessed. Alternatively, this difference in findings may be due to high rates of co-occurring psychiatric conditions in our NT comparison group (41.84%) compared with rates in the non-autistic group (18%) in Oomen et al.'s study. This is particularly important, as a survey of 16,338 young people and adults in the United Kingdom found that those with existing mental health conditions before the pandemic were more likely to report that their mental health had become worse due to COVID-19 restrictions.⁴

However, there were two areas where perceived change due to the pandemic differed across our participant groups: physical QoL and anxiety. Comparing proportions visually (Supplementary Figure S4), fewer ND participants felt that their physical QoL had not changed (14%), and more felt that this had improved due to COVID-19 (35%) compared with other participant groups. For anxiety, however, a greater proportion of the Aut+ND group felt that this had improved (24%) due to the pandemic compared with other groups. However, as these differences in observed versus expected proportions did not meet corrected significance levels, limited conclusions can be drawn.

Nonetheless, this supports the idea that some aspects of COVID-19 restrictions and lockdown may have been beneficial for autistic and other ND adults. Similar findings have emerged from studies in which autistic adults have reported: being less tired at the end of the day,¹⁷ reduced sensory and social overload, and benefits of having more time for themselves, with their families or working from home,¹⁶ or overall decreases in psychological problems⁴⁴ in relation to restrictions and “lockdown living.” In the aftermath of the current pandemic, it may be beneficial to identify these positive changes and support individuals to adopt these into their lifestyles going forward where possible. In this respect, we agree that “This may be a really good opportunity to make the world a more autism-friendly place,”⁴⁵ but also hope that everyone (ND or otherwise) might be supported to carry forward positive changes they may have experienced during the pandemic.

While perceived change appears essentially the same across groups, these changes may have a more prominent impact on autistic groups (and in some cases, those who are ND in other ways) as their QoL was already lower and their anxiety, depression, and loneliness already higher than NT adults. Thus, treatments for co-occurring mental health difficulties, and support to improve QoL and combat loneliness, are now even more vital to address the inequalities faced by autistic and other ND adults.

Strengths and limitations

Our study has some important strengths and limitations. Many studies of autistic adults, or those with other conditions, commonly adopt a two-group comparison design (e.g., autistic vs. non-autistic). A strength of our study is the unique design, which has allowed us to examine and compare the experiences of autistic adults (with and without additional intersecting ND identities), other ND adults (e.g., those with ADHD), and non-ND adults living in the United Kingdom during the COVID-19 pandemic.

Our selection of measures is also a strength of this study. The PHQ-9 has been recently validated for use with autistic adults,³⁷ and our anxiety measure (ASA-A) was even specifically designed for use with autistic adults.³⁴ Similarly, the WHOQOL-BREF is validated for use in autistic populations³³ and is one of the most widely used QoL measures for this population.⁴⁶

Volunteer sample and gender

We advertised our study through various social media and research recruitment platforms to maximize the number and diversity of respondents. The recruitment of this type of convenience sample means that there are likely to be a number of biases relating to our findings. For example, our survey sample might differ from the broader U.K. population in rates of mental health difficulties. Our NT comparison group reported particularly high levels of co-occurring psychiatric conditions (41.84%), and around half scored above cutoffs for depression (50.55%) and loneliness (51.65%).

It is unclear whether this reflects the impact of a selection effect of who chose to take part in our study (e.g., due to the fact that we posted study adverts using the Participate platform by the Charity MQ: Transforming Mental Health). Therefore, our comparison group may not represent the general U.K. population now or at other time points. Given the high rates of mental health difficulties reported by our comparison group, it is even more striking that our autistic groups still showed significantly worse mental health and QoL.

Across all sample groups, most participants were female (61%–86%)—which is commonly the case in volunteer samples, including those reported in recent COVID-19-related research.^4,16 This limits the generalizability of our findings and also means that we could not explore gender differences across the groups. This is regrettable given that other studies have identified that females have experienced greater adverse effects as a result of the pandemic, both in the general population⁴ and among autistic adults.¹² Although sample biases such as these are common with online research, further work may be needed to identify how researchers can improve sample representativeness even when online convenience sampling is the only viable option.

Inclusion of proxy-reports by relatives

To include and consider the experiences of autistic and ND individuals who would be unable to complete an online survey, we aimed to recruit relatives of autistic and ND adults to act as proxy-reporters. This approach has been adopted previously for autistic adults (e.g., for QoL^47–49), although its validity has been questioned.⁵⁰ However, our final sample only included 17 relatives providing proxy-reports. Thus, we have very little data from this group to either analyze separately or compare with the data provided by self-responders.

Relevance of findings to different groups of ND adults

Our novel grouping approach used in this study can be seen as a strength of this work. However, including participants with a broad range of ND identities does pose issues to the generalizability of our findings. Specifically, most participants from our ND and Aut+ND groups had ADHD, meaning that the findings related to these groups may be more relevant to those with ADHD compared with those with other types of ND identities.

Study time frame and cross-sectional data collection

Further considerations related to our study findings concern the time period of data collection and the cross-sectional nature of the study. The time frame for data collection (June 11, 2020 to September 30, 2020) was relatively short and can now be considered to represent an early period in the pandemic. As such, it remains to be seen how representative our findings will be compared to later stages of the pandemic and into the future. Additionally, the survey was cross-sectional in nature; participants answered questions at a single time point, although they reflected on how they felt their mental health and QoL had changed as a result of the pandemic. It is important to note that biases in participant recall may mean that these perceived changes do not accurately represent actual changes in mental health and QoL. Furthermore, the lack of longitudinal data means that we cannot make any causal inferences based on the associations that have been identified.

Conclusions

Our study found that autistic adults in the United Kingdom reported consistently poorer QoL, anxiety, depression, and loneliness compared with NT participants during the COVID-19 pandemic. In some domains of QoL, other ND adults also reported lower scores than NT adults. Many ND and NT people felt that the pandemic had harmed their well-being in these areas, but a small minority felt that these had improved.

Our findings highlight that there may now be an increased need for effective mental health services and support for adults across the United Kingdom. Autistic and other ND adults, in particular, should be a high priority group for additional support, both due to continued disparities in mental health and well-being and further difficulties they may face as a result of the pandemic.

Authorship Confirmation Statement

S.J.C., D.M., E.C., and F.H. all contributed to the study conceptualization, selection of measures, and survey questions. S.J.C. designed online questionnaires, led study advertising efforts, extracted data, conducted data analysis, and drafted the article. D.M., E.C., J.A.-B., and F.H. provided critical revisions on the article draft. All coauthors have reviewed and approved this article before submission. This article is not published, in press, or submitted elsewhere.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

MQ: Transforming Mental Health have supported this work through studentship funding to Simone J. Capp. Francesca Happé is part-funded by the NIHR Maudsley Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King's College London.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

Supplementary Figure S4

Supplementary Table S5