Towards the measurement of autistic burnout

Introduction

Research on autistic burnout is only just emerging in the literature, despite being discussed by autistic adults online for more than a decade. Terminology, and to a degree the formulation, is derived from occupational burnout, defined in the International Classification of Diseases, 11th Revision (ICD-11) (World Health Organization, 2019) as a condition characterised by exhaustion, perceived reduced professional efficacy and mental distance from one’s job, although this conceptualisation is not uncontested (Hillert et al., 2020). To our knowledge, only four studies have focused specifically on autistic burnout (Higgins et al., 2021; Mantzalas et al., 2021; Phung et al., 2021; Raymaker et al., 2020), although studies on the related phenomena such as fatigue in autistic children (Keville et al., 2021), experiences of autistic students during COVID-19 (Cage & McManemy, 2022) and experiences of masking (or ‘camouflaging’) in autism (Pearson & Rose, 2021) have referenced autistic burnout. All four studies on autistic burnout, despite some differences, have largely described core elements of exhaustion, withdrawal, a heightening of autistic traits and reduced functioning. The emerging literature arguing for the existence of an autistic burnout syndrome, and its devastating impacts, highlights the need for further research and development.

All studies to date have identified characteristics that they argue distinguish autistic burnout from depression and occupational burnout. Raymaker et al. (2020) were the first to publish a definition of autistic burnout. They gathered data from (n = 10) interviews as well secondary analysis of (n = 19) social media accounts and (n = 9) interviews on employment. Their definition included reduced tolerance of stimuli alongside the typical burnout characteristics of exhaustion and reduced function. Higgins et al. (2021), based on the findings from a grounded Delphi method study of autistic adults (N = 23) positioned as experts by lived experience, derived a somewhat different definition. Higgins et al. (2021) argued for interpersonal withdrawal as an essential feature, alongside exhaustion as the core characteristics of autistic burnout, with reduced function, confusion or problems with executive function, and increased intensity of autistic traits, including sensory sensitivities.

Subsequent studies have proposed a variety of potential risk factors. Mantzalas et al. (2021), analysing 1127 posts from Twitter and the Wrong Planet website, endorsed the findings of earlier studies, though also highlighted the potential roles of alexithymia (i.e. problems identifying emotions) and interoception (i.e. sensing of internal body states). Mantzalas et al. (2022) further put forward a theoretical framework intending to provide a holistic perspective outlining potential measurable direct and indirect pathways to autistic burnout. This framework drew from the social-relational model of disability, the neurodiversity paradigm, the job demands-resources model and the conservation of resources theory relating to burnout and stress. Previously, masking and unaccommodating neurotypical environments primarily were the proposed contributing factors, but Mantzalas et al.’s framework attempted to identify multiple potential social, environmental, psychological and individual risk and protective factors, including satisfaction with life, stigma and stimming. In their review and conceptual analysis of masking, Pearson and Rose (2021) discussed social context and stigma driving the need to mask autistic behaviours, such as stimming, and the role alexithymia plays in not being able to identify when stressors are accumulating. The impact of masking on overall mental well-being is a focus of several recent studies (e.g. Cage & Troxell-Whitman, 2019; Cassidy & Rodgers, 2017; Hull et al., 2019). All studies on autistic burnout have highlighted the role of masking.

Where other studies have focused on adult samples, Phung et al. (2021) interviewed eight autistic children and youth (aged 8–18 years, median 14 years) about experiences of autistic burnout, shutdown, meltdown and inertia using definitions based on previous studies (Belek, 2019; Raymaker et al., 2020; Welch et al., 2021). They defined autistic burnout as a state of ‘severe and chronic exhaustion’ (p. 2), inertia as a state of being ‘stuck’ (p. 2) and unable to physically engage in desired activities, meltdowns as a state of being ‘entirely overwhelmed’ (p. 2), expressed externally with a lack of control and cumulative stress, and shutdowns being ‘internal expressions’ of ‘emotional pain’ precipitating withdrawal from surrounding environments (p. 2). Using the language of the autistic children and youth, the phenomena were renamed with burnout as feeling exhausted, inertia as feeling stuck, meltdown as feeling out of control and shutdowns as feeling frozen. Autistic young people used metaphors such as a ‘heavy blanket’, ‘hard piece of dough’ or ‘old computer’ to describe their burnout experiences in particular: ‘a slow old computer that’s trying to run Google Chrome . . . it just uses up a lot of RAM’ (p. 7). Yet, Phung et al. (2021) also found that participants’ descriptions were unclear in distinguishing burnout from shutdown and inertia, though cautioned that young people may not have had the vocabulary needed for clearer delineation and were more commonly exposed to specific terms such as ‘meltdown’.

All four studies to date (Higgins et al., 2021; Mantzalas et al., 2021; Phung et al., 2021; Raymaker et al., 2020) have identified characteristics suggesting a distinction of autistic burnout from depression and occupational burnout. Given the ongoing debate in the literature outside of autism regarding the separation, or lack thereof, of occupational burnout from depression (e.g. (Bianchi et al.,2020, 2021; Parker & Tavella, 2021; Tavella et al., 2020, 2021; Tavella & Parker, 2020; Verkuilen et al., 2021), this is of particular importance. These authors have used various meta-analytic, structural equation and factor analytic approaches, and argued on both sides of a problematic overlap of depression and burnout. In autistic burnout research, one participant in Higgins et al. (2021) reported ‘depression is the side effect with burnout being the cause’ (p. 2362). Studies emerging on autistic burnout, together, propose the existence of a potentially unique autistic burnout syndrome and further suggest that existing measures of occupational burnout may not be suitable in this context, or at least, would not address the full range of symptomatology.

Methods

Participants

Volunteer participants were recruited through advertisement distributed by social media, the Australian Longitudinal Study of Autism in Adulthood (ALSAA; Arnold et al., 2019) newsletter, autism organisation websites and newsletters. As an incentive, participants could enter a prize draw for one of four US$100 gift cards. Participants had an independent clinical diagnosis of autism and self-identified an episode of autistic burnout. The consent form made reference to the definition of autistic burnout by Raymaker et al. (2020) and the survey included a screening question ‘Have you had an experience of autistic burnout?’. The survey was anonymous, and N = 141 participants provided useable data, that is, more complete responses beyond the opening demographic questions. Only a small proportion of males participated (16%, n = 22; n = 3 missing), with higher numbers of females (64%, n = 88) and participants reporting another gender (20%, n = 28). The mean age was 40.2 years (SD = 10.8, range: 20.3–71.0; n = 8 missing). Two participants (1%) scored 64 and 65 on the Autism Spectrum Quotient-28 (AQ-28), just below the >65 cut-off which has a sensitivity of 97%, but were nevertheless included in the analysis.

Approximately half of the participants were employed, with the majority having completed high school and gaining a bachelor’s degree or higher (see Table 1). The majority of the sample were diagnosed recently in adulthood: the mean age at diagnosis was 36.9 years (SD = 12.4, range: 3.4–67.1), and the mean years since diagnosis were 3.4 years (SD = 5.3, range: 0–27). Current co-occurring mental health conditions were common, with 113 participants (78%) reporting an internalising mental health condition (i.e. mental health conditions where people internalise their problems, such as depression or anxiety). Responses were collected from October 2020 to April 2021, during which time COVID-19 related lockdowns were occurring. Further demographics and co-occurring mental health conditions are presented in Table 1.

OPEN IN VIEWER

Table 1.

Demographics and mental health conditions (N = 141).

Characteristic	n (%) / M (SD)
Age in years (n = 8 missing)	40.2 (10.8)
Gender (n = 3 missing)
Male	22 (16%)
Female	88 (64%)
Another gender	28 (20%)
Age at diagnosis	36.9 (12.4)
Currently employed (n = 5 missing)	77 (57%)
Education
Completed Year 10 or higher (n = 5 missing)	130 (96%)
Bachelor’s degree or higher (n = 10 missing)	88 (67%)
Marital status (n = 3 missing)
Single / divorced	44 (32%)
Married / de facto	60 (44%)
Separated	14 (10%)
Widowed	6 (4%)
Other / don’t know	14 (10%)
Living situation
Alone	38 (27%)
Couple	61 (43%)
With parents / relatives	24 (17%)
With others	10 (7%)
Other	8 (6%)
Born in Australia (n = 6 missing)	118 (87%)
Ethnicity Caucasian (n = 5 missing)	115 (85%)
Mental health conditions (n = 3 missing)	Current diagnosis	Previous diagnosis	Currently receiving treatment
Anxiety	100 (72%)	24 (18%)	76 (55%)
Depression	66 (48%)	53 (38%)	60 (43%)
Attention deficit/hyperactivity disorder	57 (41%)	5 (4%)	37 (27%)
Social anxiety	56 (41%)	23 (17%)	40 (29%)
Post-traumatic stress disorder	39 (28%)	20 (14%)	22 (16%)
Panic disorder	17 (12%)	25 (18%)	18 (13%)
Eating disorder	13 (9%)	11 (8%)	5 (4%)
Obsessive compulsive disorder	10 (7%)	11 (8%)	10 (7%)
Bipolar disorder	8 (6%)	18 (13%)	7 (5%)
Personality disorder	5 (4%)	13 (9%)	3 (2%)
Pathological grief	5 (4%)	4 (3%)	0 (0)
Substance abuse	1 (1%)	15 (11%)	0 (0)
Tic disorders	1 (1%)	2 (1%)	0 (0)
Schizophrenia	0 (0%)	3 (2%)	0 (0)
Other mental health disorder a	12 (9%)	0 (0)	9 (7%)

SD: standard deviation.

a

Other conditions reported more than once included premenstrual dysphoric disorder and dissociative identity disorder.

Measures

Two unvalidated measures of autistic burnout were employed alongside several standardised measures of conceptually related constructs in the Autistic Burnout Survey. This article reports on a portion of the survey findings focused on the measurement of autistic burnout. We attend to participants’ responses to the open-ended questions in a separate paper (Arnold et al., in press).

Autistic burnout

We employed two measures of autistic burnout, one that we developed and the second created by Raymaker et al. (2020). In our Autistic Burnout Survey, led by autistic peer researcher J.M.H., we developed a series of items based on a combination of the Higgins et al.’s (2021) and Raymaker et al.’s (2020) definitions. Forty-eight of these items questioned participants on ‘During my most recent (or current) autistic burnout experience’ and were related to elements in one or both definitions (e.g. exhaustion, withdrawal, executive functioning, masking), including items relating to meltdowns for the purpose of determining divergent validity (see Table 2). All quantitative survey items were scored on 7-point Likert agreement scale ranging from ‘1’ (strongly agree) to ‘7’ (strongly disagree) with an additional ‘not applicable’ (NA) option. Each question block was followed by open-ended items probing for additional detail or missing concept coverage. J.M.H. determined agreement scale design with open-ended follow-up questions to allow autistic participants a full range of response options and the ability to explain their responses. J.M.H., S.R.C.A. and A.D. developed items, which were then reviewed by all co-authors. For example, given that interpersonal withdrawal was a central characteristic in the Higgins et al.’s (2021) definition, Item 7 asked ‘During my most recent (or current) autistic burnout experience . . . I withdrew from social situations’. Exploratory factor analysis (EFA) and iterative scale reduction applied to these items revealed a grouping of 20 Autistic Burnout Severity Items (ABSI) across four factors (see Table 2). The process to identify the ABSI is described subsequently. These items showed good overall internal consistency (Cronbach’s α = 0.88) with acceptable internal consistency for each factor (range: 0.73–0.86). However, internal consistency at the subscale level was questionable (α = 0.66), which cautions against assuming unidimensionality of the scale. NA responses were set to missing, and those participants who had more than two NA or missing items, were excluded from ABSI total score (n = 2). In subsequent analyses, an average score across answered items was taken to maximise total observations, giving a possible total score ranging from 1 (most severe) to 7 (least severe).

OPEN IN VIEWER

Table 2.

Autistic burnout experience items ordered by highest mean agreement.

Survey item	Obs.	Not applicable (NA) responses	Mean agreement	SD	Minimum	Maximum	Factor loading
1. I felt extremely tired or worn out. a	140	0	1.15	0.45	1	3	0.75
4. I was mentally exhausted. a	141	0	1.20	0.51	1	4	0.73
7. I withdrew from social situations. b	139	2	1.24	0.56	1	3	0.75
8. I had difficulties doing my usual work as well as I typically do. b	135	5	1.30	0.67	1	5	0.37
5. I felt overwhelmed by my environment. b	141	0	1.34	0.83	1	6	0.65
12. I found it more difficult to relate socially to people. b	141	0	1.42	0.81	1	6	0.75
3. I was physically exhausted during the day. a	140	1	1.47	0.76	1	4	0.90
27. I found some of the following more distressing than usual: Sudden or loud noises, bright or flickering lights . . . c	141	0	1.48	1.00	1	7	0.71
25. Communicating with others was more difficult than usual.	141	0	1.48	1.00	1	7
10. I had difficulties with my school or academic tasks.	90	50	1.51	0.78	1	4
6. I withdrew from family members or other people I live with.	139	2	1.58	0.94	1	5
2. I did not have the energy to carry out daily activities. a	141	0	1.60	0.88	1	5	0.71
23. I had a hard time planning my day . . .	140	0	1.65	1.04	1	7
33. My other mental health problems got worse. c	133	7	1.67	1.06	1	7	0.69
20. I found that I took longer to complete tasks.	140	1	1.69	1.14	1	7
22. I was easily distracted. c	139	1	1.76	1.14	1	7	0.54
36. I was overwhelmed with emotions.	141	0	1.85	1.2	1	7
21. During conversations, I took longer to respond to questions. c	139	2	1.88	1.17	1	7	0.58
13. I lost interest in doing activities that I usually enjoy.	140	0	1.89	1.41	1	7
34. I was more self-critical. c	140	0	1.90	1.34	1	7	0.72
32. My physical health conditions got worse.	134	6	1.99	1.39	1	7
9. I was unable to do my usual work.	137	3	1.99	1.33	1	7
14. I was unable to do activities that I usually enjoy.	141	0	2.00	1.36	1	7
28. I found myself avoiding . . . environments that I would typically enjoy (e.g. Christmas, weddings).	133	8	2.02	1.34	1	7
18. I had difficulty remembering instructions given to me. d	140	0	2.04	1.33	1	7	0.52
15. I experienced some time of confusion. d	141	0	2.07	1.29	1	7	0.56
45. I felt very restless and distressed.	140	0	2.13	1.49	1	7
29. My repetitive behaviours / stimming increased . . . c	138	3	2.17	1.34	1	6	0.58
24. I had difficulty maintaining my usual self-care routines . . ..	141	0	2.17	1.69	1	7
40. I felt intense distress.	141	0	2.19	1.59	1	7
30. I had an intense need for ‘sameness’ in my daily activities . . .	141	0	2.27	1.40	1	6
31. I lost my ability to hide that I was autistic . . .. c	138	3	2.38	1.54	1	7	0.50
11. I was unable to do any school or academic tasks.	90	50	2.41	1.61	1	7
16. I experienced gaps in my memory. d	139	1	2.55	1.79	1	7	0.89
17. I had difficulty remembering specific events. d	140	0	2.59	1.72	1	7	0.86
37. I felt emotionally numb.	139	1	2.67	1.82	1	7
35. I would talk to myself.	139	1	3.01	1.79	1	7
26. I was unable to communicate with others.	141	0	3.01	1.63	1	7
43. I would have angry outbursts.	140	1	3.42	1.91	1	7
41. I would pace or walk back and forth a lot.	135	6	3.48	2.09	1	7
39. I felt unbearable internal pain.	138	2	3.64	2.07	1	7
48. I thought a lot about killing myself.	136	5	3.95	2.21	1	7
19. At times, I temporarily forgot things about myself including who I am. d	138	2	4.12	1.95	1	7	0.71
44. I would hurt myself (e.g. bang or hit my head).	135	5	4.30	2.18	1	7
46. My body moved quickly and I could not stop it.	137	4	4.34	2.12	1	7
38. I would yell uncontrollably.	135	6	4.41	2.18	1	7
47. I started to self-harm (e.g. cutting myself).	131	10	5.05	2.06	1	7
42. I became physically aggressive.	136	5	5.30	1.84	1	7

SD: standard deviation; ABSI: Autistic Burnout Severity Items.

a

Item included in the ABSI Exhaustion factor.

b

Item included in the ABSI Overwhelm and Withdrawal factor.

c

Item included in the ABSI Heightened Autistic Self-Awareness factor.

d

Item included in the ABSI Cognitive Disruption factor.

Second, we gathered data using a prepublication version of the AASPIRE Autistic Burnout Measure (AABM) with permission from Raymaker and colleagues. Building on work conducted by Raymaker et al. (2020), this 27-item tool asked participants to rate items on a 5-point Likert scale, ranging from ‘0’ (strongly agree) to ‘4’ (strongly disagree), ¹ yielding a total possible score range of 0–108. Participants respond to items in reference to ‘the past three months’ with item content such as ‘I’ve wanted to isolate myself from others more often than I usually do’ and ‘I’ve felt more mentally exhausted than I usually do’. Lower total sum scores are indicative of greater risk of being in autistic burnout.

Autism severity and screening

The widely used 50-item AQ tool (Baron-Cohen et al., 2001), originally developed to identify autistic traits in adults of at least average intelligence, uses a 4-point Likert scale, ranging from ‘1’ (definitely agree) to ‘4’ (definitely disagree). The AQ-50 has been critiqued on several grounds, including its length, factor structure and whether it is suitable to use as an autism severity measure (Lundqvist & Lindner, 2017). In this study, we gathered subsets of the original 50-item AQ. This included the 28 items selected by Hoekstra et al. (2011) in creating the abridged AQ-28, which we used to determine whether participants met cut-off for autism. The AQ-28 using a cut-off of >65 demonstrates high sensitivity (97%) and specificity (82%) for distinguishing between autistic and non-autistic participants. We also gathered the 12 items (4 not included in the AQ-28) identified by Lundqvist & Lindner, (2017) as having utility to measure autism severity. Good internal consistency was evident in this sample for both the 28-item (α = 0.82) and 12-item (α = 0.81) extracts.

Autistic dispositions

A range of measures were gathered to identify potential unique stressors correlated with autistic burnout, including the Adult Repetitive Behaviour Questionnaire-2 (RBQ-2A). Adapted from the Repetitive Behaviours Questionnaire-2 designed for children, the RBQ-2A (Barrett et al., 2015) is a 20-item tool scored on 4- and 3-point frequency or severity scales, with higher scores indicating greater severity. Following previous studies (Barrett et al., 2015, 2018), the 4-point scaled items were collapsed to a 3-point scale, then a total sum score was used giving a possible score range of 20–60. The RBQ-2A also provides two factor scores of repetitive motor behaviour and insistence of sameness, although only the total score indicating levels of repetitive behaviours was used in this study. It has shown convergent validity with the AQ-50, and good internal consistency in non-autistic (α = 0.73) and autistic samples (α = 0.91), which was replicated in the current sample (α = 0.90).

The Glasgow Sensory Questionnaire (GSQ) (Robertson & Simmons, 2013) is a 42-item tool assessing hyper- and hypo-sensitivities. Items are scored on a 5-point frequency scale, ranging from ‘0’ (never) to ‘4’ (always), giving a maximum total sensory score of 168, with higher scores indicating higher amounts of sensitivities. A single-factor total sensory score has strong internal consistency (α = 0.94), which we replicated in the current sample (α = 0.91).

The Camouflaging Autistic Traits Questionnaire (CAT-Q) (Hull et al., 2019) is the first measure of social camouflaging / masking behaviours, that is, the strategies and actions used by autistic individuals to reduce their autistic characteristics in the context of social interactions. This 25-item scale is scored on a 7-point Likert scale, ranging from ‘1’ (strongly disagree) to ‘7’ (strongly agree) with five reverse-scored items, given a possible score range of 25–175, with higher scores indicating higher amounts of camouflaging. It has shown acceptable test–retest reliability (r = 0.77) and good internal consistency (α = 0.94), replicated in our sample (α = 0.90). The total sum score was used in our study, although subdomains of compensation, masking and assimilation can be calculated. Convergent validity has been tested in autistic (N = 306) samples in comparison with autistic traits (r = 0.34; p < 0.001) using the Broad Autism Phenotype Questionnaire (BAPQ; Hurley et al., 2007), well-being (r = −0.16; p < 0.05) using the Warwick–Edinburgh Mental Well-being Scale (WEMBS; Tennant et al., 2007), anxiety (r = 0.35; p < 0.001) using the Generalised Anxiety Disorder-7 (GAD-7; Spitzer et al., 2006), social anxiety (r = 0.44; p < 0.001) using the Liebowitz Social Anxiety Scale (LSAS; Liebowitz, 1987) and depression (r = 0.28; p < 0.001) using the Patient Health Questionnaire-9 (PHQ-9).

The Interoception Sensory Questionnaire (ISQ) (Fiene et al., 2018) was developed to measure autistic adults’ perception and interpretation of interoceptive sense. The original 20-item tool had a 7-point Likert scale response option, ranging from ‘1’ (not at all true of me) to ‘7’ (very true of me), and showed strong internal consistency (Cronbach’s α = 0.96) for a single factor, strong correlations with a measure of alexithymia (r = 0.76) and significant correlations with a measure of interoceptive awareness (range: −0.28 to −0.15). Subsequently, Suzman et al. (2021) developed a revised 8-item version using a condensed 5-point Likert scale response option and noted multiple redundant item pairs. They also developed an online scoring calculator (https://asdmeasures.shinyapps.io/ISQ_score/) to facilitate the use of latent trait scores. Due to clerical error, we did not gather Item 20 of the original ISQ; however, we substituted the redundant paired Item 19 and calculated ISQ-8 latent trait scores using the online calculator from Suzman et al. (2021). Suzman found strong internal consistency of the ISQ-8 using coefficient omega (ω = 0.90), which was replicated in the eight items from our sample (ω = 0.92)

The recently developed Perth Alexithymia Questionnaire (PAQ) (Preece et al., 2018) is a 24-item tool using a 7-point Likert scale, ranging from ‘1’ (strongly disagree) to ‘7’ (strongly agree) giving a possible score range of 24–168; higher scores indicate higher levels of alexithymia – an inability to identify and describe one’s emotions. Its four subscales and all composite scores show good internal consistency (α = 0.87–0.95), which was replicated in this sample (0.90–0.97). It has shown good convergent validity with expected relationships with emotional regulation measures and measures of depression and anxiety, as well as divergent validity from overall distress. It has also shown good convergent validity (Preece et al., 2020) with the widely used 20-item Toronto Alexithymia Scale (TAS-20).

Depression

The PHQ-9 (Kroenke et al., 2010) measures the severity of depressive symptoms and is used to identify clinical levels of major depressive disorder. Each of its nine items is scored on a 4-point frequency scale: ‘0’ (not at all), ‘1’ (several days), ‘2’ (more than half the days) and ‘3’ (nearly every day). When summed, the scale generates a continuous score ranging from 0 to 27, with higher scores indicating greater depression severity (Kroenke et al. 2010). The PHQ-9 demonstrates excellent internal consistency (α = 0.89) and test–retest reliability (r = 0.84) (Kroenke et al. 2001), and acceptable sensitivity and specificity (>79%) in detecting depressive disorder (Cassidy et al. 2018). It showed good internal consistency in this sample (α = 0.90). It has been previously validated in an autistic sample (Arnold et al., 2019).

Results

Relationships with ABSI

Across the sample, very high rates of depression were evident, with 98% (n = 131) of the complete responses (n = 7 missing) above cut-off for clinical depression on the PHQ-9. Many participants (n = 60, 44%) endorsed the PHQ-9 item relating to suicide ideation or self-harm. Similarly, 53% (n = 69) reported high levels of alexithymia on the PAQ, while a minority (12%; n = 12) had low levels of alexithymia (n = 10 missing). Descriptive statistics for standardised tool and burnout indicators are presented in Table 3.

OPEN IN VIEWER

Table 3.

Descriptive statistics.

Scale (construct)	Obs.	M	SD	Min	Max
ABSI (burnout severity)	140	1.87	0.65	1	4.55
AABM (current burnout)	137	28.47	21.66	0	93
PHQ-9 (depression)	134	23.8	7.1	9	36
CAT-Q (camouflaging)	136	134.77	22	57	175
RBQ-2 (repetitive behaviours)	130	40.56	7.01	21	55
PAQ (alexithymia)	131	106.65	37.06	24	168
AQ-12 (autism severity)	142	40.81	5.69	22	48
AQ-28 (autism screening)	142	89.37	10.53	64	112
GSQ (sensory sensitivity)	132	89	23.72	18	159
ISQ-8 (interoception)	137	54.93	9.63	33	78

ABSI: Autistic Burnout Severity Items; AABM: AASPIRE Autistic Burnout Measure; PHQ-9: Patient Health Questionnaire-9; CAT-Q: Camouflaging Autistic Traits Questionnaire; RBQ-2: Adult Repetitive Behaviour Questionnaire-2; PAQ: Perth Alexithymia Questionnaire; AQ-12: 12-item extract from the Autism Quotient; AQ-28: 28-item extract from the Autism Quotient; GSQ: Glasgow Sensory Questionnaire; ISQ-8: 8-item extract from the Interoception Sensory Questionnaire.

Correlations between measures are presented in Table 4. Both burnout indicators showed significant correlations across all supplementary measures, except, unexpectedly, the AABM did not correlate with the CAT-Q. We note that the ABSI subdomains showed some marginally stronger correlations with discriminant measures over other ABSI subdomains; however, they were all most strongly correlated with the ABSI total score. These marginally stronger correlations with discriminant measures were not significantly different when analysed with a backtransformed average Fisher’s Z procedure, following Hittner et al. (2003). We also report item-level Spearman’s correlations with the PHQ-9, ABSI meltdown items and Spearman’s item-rest correlations for ABSI domain and total scores in Supplementary Table 4. We note ABSI items were more strongly correlated with ABSI domain and total scores than with the PHQ-9, including for additional partial correlations that were conducted between ABSI domain scores and ABSI total score controlling for variation in PHQ-9 scores.

OPEN IN VIEWER

Table 4.

Pairwise correlations (bottom diagonal) and confidence intervals of correlations (top diagonal).

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)
(1) ABSI (burnout severity)	1.00	[0.43, 0.67]	[0.42, 0.65]	[0.83, 0.91]	[0.78, 0.88]	[0.21, 0.51]	[−0.60, −0.34]	[−0.48, −0.18]	[−0.60, −0.34]	[−0.59, −0.32]	[−0.33, −0.01]	[−0.65, −0.40]	[−0.54, −0.26]
(2) ABSI exhaustion	0.56***	1.00	[0.18, 0.48]	[0.16, 0.46]	[0.18, 0.48]	[−0.02, 0.31]	[−0.40, −0.08]	[−0.34, −0.01]	[−0.31, 0.02]	[−0.33, 0.00]	[−0.30, 0.02]	[−0.46, −0.15]	[−0.39, −0.07]
(3) ABSI withdrawal	0.55***	0.34***	1.00	[0.22, 0.51]	[0.14, 0.44]	[0.01, 0.33]	[−0.36, −0.03]	[−0.32, 0.00]	[−0.28, 0.06]	[−0.35, −0.02]	[−0.36, −0.04]	[−0.34, −0.00]	[−0.25, 0.08]
(4) ABSI cognitive disruption	0.87***	0.32***	0.37***	1.00	[0.44, 0.67]	[0.16, 0.47]	[−0.54, −0.25]	[−0.38, −0.07]	[−0.54, −0.25]	[−0.52, −0.23]	[−0.25, 0.08]	[−0.54, −0.26]	[−0.47, −0.17]
(5) ABSI autistic awareness	0.84***	0.33***	0.30***	0.56***	1.00	[0.18, 0.48]	[−0.57, −0.29]	[−0.48, −0.18]	[−0.64, −0.37]	[−0.59, −0.32]	[−0.32, 0.01]	[−0.64, −0.39]	[−0.55, −0.25]
(6) Raymaker et al. (burnout)	0.37***	0.15*	0.17**	0.33***	0.35***	1.00	[−0.75, −0.56]	[−0.24, 0.10]	[−.43, −0.11]	[−0.46, −0.15]	[−0.37, −0.05]	[−0.44, −0.12]	[−0.42, −0.10]
(7) PHQ-9 (depression)	−0.48***	−0.25***	−0.20**	−0.41***	−0.43***	−0.67***	1.00	[−0.03, 0.31]	[0.25, 0.54]	[0.22, 0.52]	[0.09, 0.41]	[0.25, 0.55]	[0.20, 0.49]
(8) CAT-Q (camouflaging)	−0.34***	−0.18**	−0.16*	−0.23**	−0.35***	−0.07	0.13	1.00	[0.04, 0.38]	[0.06, 0.40]	[0.05, 0.37]	[0.14, 0.46]	[−0.00, 0.33]
(9) RBQ-2 (repetitive behaviours)	−0.48***	−0.15	−0.11	−0.39***	−0.53***	−0.28***	0.39***	0.22**	1.00	[0.37, 0.63]	[−0.05, 0.29]	[0.74, 0.86]	[0.42, 0.67]
(10) PAQ (alexithymia)	−0.47***	−0.17*	−0.18**	−0.38***	−0.47***	−0.31***	0.37***	0.24**	0.52***	1.00	[0.10, 0.42]	[0.39, 0.65]	[0.52, 0.72]
(11) AQ-12 (autism severity)	−0.17**	−0.14*	−0.21**	−0.09	−0.16*	−0.22**	0.26***	0.21**	0.12	0.27***	1.00	[−0.05, 0.29]	[−0.07, 0.26]
(12) GSQ (sensory sensitivity)	−0.53***	−0.31***	−0.17*	−0.41***	−0.53***	−0.29***	0.40***	0.31***	0.81***	0.54***	0.12	1.00	[0.57, 0.76]
(13) ISQ-8 (interoception)	−0.41***	−0.24***	−0.09	−0.33***	−0.40***	−0.27***	0.35***	0.17*	0.55***	0.63***	0.10	0.68	1.00

ABSI: autistic burnout severity items; PHQ-9: Patient Health Questionnaire-9 items; CAT-Q: Camouflaging Autistic Traits Questionnaire; RBQ-2: Adult Repetitive Behaviour Questionnaire-2; PAQ: Perth Alexithymia Questionnaire; GSQ: Glasgow Sensory Questionnaire; ISQ-8: 8-item extract from the Interoception Sensory Questionnaire.

*

p < 0.10. **p < 0.05. ***p < 0.01.

Linear regression with ABSI scores as the outcome with all supplementary measures as predictors, showed significant relationships with scores on depression and camouflaging measures (see Table 5). These significant associations remained when the sample was limited to those who reported having experienced a burnout within the past 3 months. To maximise observations, we also repeated the regression using multiple imputation with 20 imputations and obtained similar results (see Supplementary Materials Table 4). Given concerns as to the unidimensionality of the ABSI, we repeated the linear regression predicting ABSI subdomain totals (see Supplementary Materials Tables 5–9). Patterns of associations changed when predicting subdomain scores, with depression only significantly related in the ABSI Cognitive and ABSI Heightened Autistic Self-Awareness models, and camouflaging and repetitive behaviours significant in the ABSI awareness model.

OPEN IN VIEWER

Table 5.

Linear regression predicting ABSI (n = 106).

ABSI	Coef.	St. Coef.	SE	t value	p value	95% confidence interval
PHQ-9 (depression)	−0.02	−0.28	0.01	−3.12	>0.001***	−0.04	−0.01
CAT-Q (camouflaging)	−0.01	−0.22	0	−2.56	0.01**	−0.01	0
RBQ-2 (repetitive behaviours)	−0.02	−0.22	0.01	−1.56	0.12	−0.04	0.01
PAQ (alexithymia)	0	−0.16	0	−1.42	0.16	−0.01	0
AQ-12 (autism severity)	0	−0.01	0.01	−0.13	0.9	−0.02	0.02
GSQ (sensory sensitivity)	0	−0.10	0	−0.61	0.54	−0.01	0.01
ISQ-8 (interoception)	0	0.08	0.01	0.65	0.52	−0.01	0.02
Constant	4.29		0.62	6.94	>0.001***	3.07	5.52
R 2	0.40
F-test	9.41
Prob > F	0.00

ABSI: Autistic Burnout Severity Items; PHQ-9: Patient Health Questionnaire-9; CAT-Q: Camouflaging Autistic Traits Questionnaire; RBQ-2: Adult Repetitive Behaviour Questionnaire-2; PAQ: Perth Alexithymia Questionnaire; GSQ: Glasgow Sensory Questionnaire; ISQ-8: 8-item extract from the Interoception Sensory Questionnaire.

***

p < 0.01, **p < 0.05.

Relationships and sensitivity of the AABM

Linear regression with AABM as the outcome variable, limited to those who had experienced burnout in the last 3 months, with all supplementary measures as predictors, showed only depression as a significant predictor (see Table 6). The pattern of findings did not change when all participants were included. To maximise observations, we also repeated the regression using multiple imputation with 20 imputations and obtained similar results (see Supplementary Materials Table 10).

OPEN IN VIEWER

Table 6.

Linear regression predicting AABM for those experiencing burnout within last 3 months (n = 76).

AABM (current burnout)	Coef.	St. Coef.	SE	t value	p value	95% confidence interval
PHQ-9 (depression)	−2.17	−0.74	0.28	−7.69	>0.001***	−2.73	−1.61
CAT-Q (camouflaging)	−0.04	−0.04	0.10	−0.42	0.68	−0.23	0.15
RBQ-2 (repetitive behaviours)	−0.04	−0.01	0.5	−0.07	0.94	−1.03	0.96
PAQ (alexithymia)	−0.05	−0.08	0.07	−0.68	0.50	−0.19	0.09
AQ-12 (autism severity)	0.75	0.16	0.41	1.81	0.08	−0.08	1.57
GSQ (sensory sensitivity)	0.14	0.16	0.16	0.85	0.40	−0.18	0.45
ISQ-8 (interoception)	−0.23	−0.11	0.27	−0.84	0.40	−0.76	0.31
Constant	59.66		24.44	2.44	0.02**	10.88	108.43
R 2	0.552
F-test	11.951
Prob > F	<0.001

ABSI: Autistic Burnout Severity Items; PHQ-9: Patient Health Questionnaire-9; CAT-Q: Camouflaging Autistic Traits Questionnaire; RBQ-2: Adult Repetitive Behaviour Questionnaire-2; PAQ: Perth Alexithymia Questionnaire; GSQ: Glasgow Sensory Questionnaire; ISQ-8: 8-item extract from the Interoception Sensory Questionnaire.

**

p < 0.05. ***p < 0.01.

With regard to our final research question on the predictive validity of the AABM, the AABM asks participants to report on their experiences in the past 3 months. Overlaying kernel density estimations of those who had (n = 103) and had not (n = 33) experienced autistic burnout in the past 3 months suggested poor specificity of the AABM (see Figure 1). Receiver operative curve analysis suggested poor specificity (area under the curve (AUC) = 0.661; n = 136).

OPEN IN VIEWER

Figure 1.

Kernel density scores on AABM versus have you been burnt out in the last 3 months?

Discussion

Emerging research is investigating the existence of an autistic burnout syndrome. From a survey of autistic adults with experience of autistic burnout, we used EFA to construct a preliminary measure of autistic burnout, the ABSI, and, once controlling for other factors, found masking and depression to be associated with more severe autistic burnout. Our findings must be considered with caution given the potential lack of unidimensionality of the ABSI. The ABSI and AABM total scores were only moderatly significantly correlated. The AABM tool may be problematic in not showing associations with masking in multivariate regression analysis and having poor specificity in determining participants currently or recently in burnout, compared to those who perceive they are not currently in burnout. Our finding of masking associated with the ABSI aligns with the testimonies reported in emerging literature of adults who are compelled to mask their autistic features enduring stressors and fatigue leading to autistic burnout. However, masking being a precursor rather than a feature of autistic burnout experiences may explain its lack of association with the AABM. Work is needed to develop and validate assessment tools in this area, with this study providing potential directions for item development. Longitudinal work is needed to determine causative rather than associative factors, including the interrelationship and divergent validity of burnout and depression in autism, especially considering there were some high correlations between ABSI items and subdomains with depression scores. Of importance, all other measures gathered in this study were significantly correlated with the ABSI, and future research with larger samples is needed to determine the significance of these associations.

The EFA in this study, alongside emerging literature (Higgins et al., 2021; Mantzalas et al., 2021; Phung et al., 2021; Raymaker et al., 2020), highlights the need to further investigate the phenomenon of autistic burnout. Our study, similar to other published literature, suggests that burnout is distinct from meltdowns, although its disambiguation from depression remains currently unclear. Some of the ABSI domains showed marginally stronger correlation with depression than other subdomains, although overall were more strongly correlated with the ABSI total score. Unidimensionality of the ABSI is questionable, and regression models predicting subdomain scores showed varying patterns of relationships. Also not explored in our design is the separation or interrelatedness of autistic burnout, shutdowns, catatonia (i.e. a disorder characterised in the Diagnostic and Statistical Manual of Mental Disorders (5th ed., DSM-5) as the presence of three of the following symptoms: catalepsy, waxy flexibility, stupor, mutism, negativism, agitation, posturing, stereotypes, mannerism, grimacing, echolalia and echopraxia; (Ghaziuddin et al., 2021; Vaquerizo-Serrano et al., 2022) and inertia (Buckle et al., 2021), which appears less clear. Particularly some of our participants’ experiences included changes in motivation, mutism or stupor. The children and youth in Phung et al. (2021) did not report a clear distinction, with burnout, shutdowns and inertia all being thematically grouped as ‘exhausted and/or frozen’ (p. 6). Of interest, there is content alignment between the shutdowns and burnout described by Phung et al.’s (2021) participants with the content of ABSI items. Together with the unclear duration and both chronic and acute experiences of autistic burnout (Arnold et al., in press; Higgins et al., 2021), emerging literature may point towards an overarching autistic exhaustion syndrome which encompasses burnout, shutdown and inertia, occurring across the lifespan, and is possibly related to autistic catatonia, though more research in this area is needed.

Conceptual issues in the criteria for autistic burnout may require more work before the development of autistic burnout measurement tools. The prepublication AABM tool appeared somewhat problematic in two areas. First, it shows no relation to masking, which is indicated in the literature and in our findings with the ABSI. We note that the literature discusses masking as a stressor leading to autistic burnout onset, and it is plausible that masking is a component of a complex causal pathway where it is no longer significant once in the experience of burnout. Second, having poor specificity in determining those who report currently being in autistic burnout, although our sample ideally would have included autistic adults who had never experienced autistic burnout. Although, the identification of the ABSI did not follow all processes ideal in the development of assessment tools (Boateng et al., 2018), it may provide useful material for future assessment tool development in this area. Better conceptualisation and screening tools will help identify autistic burnout sooner, and could, in some instances, prevent misdiagnoses (see Arnold et al., in press). We deliberately refer to autistic burnout as a syndrome, while acknowledging limitations of preliminary findings from emerging studies, and that this terminology is potentially stigmatising and the problematic history of the medical model within autism (Pellicano & den Houting, 2022). We recommend that autistic burnout is considered a diagnosable condition. Doing so may assist in reducing misdiagnoses and in developing validated prevention, treatment and support pathways. This would be particularly important to clinicians without extensive autism knowledge. There are potential risks though situating autistic burnout within the psychiatric nosology, with similarity to the pathologisation of depression (Ratnayake, 2022). Further research is needed with sensitivity to the most beneficial approaches to supporting autistic adults.

Masking and depression being the most predictive factors in regression models give some direction towards supports and prevention of autistic burnout. The significance of masking highlights the locus of stressors driving autistic burnout. This challenges us to promote social change, build more neurodiverse spaces and accommodation within the neurotypical world, reducing the need to mask. Relatedly, encouragement from autistic pride movements and active inclusion (Weaver et al., 2021) from neurotypical people might lead some autistic adults to lessen their masking or unmask.

The interrelationship of autistic burnout with depression appears to mirror the relationship of occupational burnout and depression, and the ongoing, years-long conceptual uncertainty between these disorders. Of interest, the Sydney Burnout Measure, which measures occupational burnout, developed by Parker et al. (2021), shows good sensitivity but poor specificity and requires clinical reasoning to exclude depression in diagnosing occupational burnout. Similar to arguments presented by Parker and Tavella (2021), such as differences in precipitant, attribution of causation, severity of depression, primary symptom and anhedonia, we would contend there are important differences between depression and autistic burnout, that they could occur independently, though likely that an autistic person who is experiencing burnout is more at risk of developing depression. For example, many participants highlight an onset associated with stressors, such as masking, with a primary symptom of exhaustion, not depression. This will be another area for future disambiguation. Furthermore, Parker and Tavella (2022) suggest that ‘widespread acceptance by the lay community of burnout as a distinct and relatable syndrome suggest it is worthy of independent designation’ (p. 1065), and we believe this is mirrored in the autistic community for autistic burnout. Future work may benefit from the inclusion of clinical reasoning, as suggested by Parker et al. (2021), to disambiguate autistic burnout from depression and other conditions that could be misdiagnosed. Future research should also consider the various approaches from the occupational burnout literature that have examined the depression and burnout overlap (e.g. Bianchi et al.,2020, 2021; Tavella et al., 2021; Verkuilen et al., 2021). We note that Tavella et al. (2021) argue that the presence of depression symptoms in burnout does not suggest that burnout and depression are synonymous. Tavella et al. (2021) applied bifactor analyses to a set of 137 items generated from existing burnout and depression measures, arriving at a 34-item general burnout factor that included only four depression-related items and may present a model for future autistic burnout work. Also missing from our study and needed in future research is an exploration of the unique stressors arising from an unaccommodating neurotypical systems and structures reportedly leading to autistic burnout, and the relationships of autistic burnout with anxiety and social anxiety.

Conclusion

Autistic people report that autistic burnout is a debilitating condition. This condition is only recently being investigated in the scientific literature. More work is needed to develop valid screening and measurement tools, including further definitional work around duration and relationships with shutdown and inertia. In common with the occupational burnout construct, further work is needed to disambiguate autistic burnout from depression and other conditions that are potential misdiagnoses. The significant relationship of masking with autistic burnout highlights core phenomena of exhaustion, withdrawal and cognitive overload in autistic people driven by stressors amplified in autistic people. Efforts to reduce the stress and frequency of masking are needed, including acceptance and valuing of autistic people in society, increase in accommodations for autistic people and promotion of autistic pride.

Supplemental Material

Supplemental Material