Autistic burnout versus major depressive disorder: Differentiating clinical features to prevent misdirected treatment

Roisín Delaney

doi:10.22271/27103919.2026.v6.i10a.122

2026, Vol. 6, Issue 10, Part A

Autistic burnout versus major depressive disorder: Differentiating clinical features to prevent misdirected treatment

Author(s): Roisín Delaney

Abstract: Autistic burnout and major depressive disorder (MDD) share enough surface features—persistent fatigue, social withdrawal, and loss of daily functioning—that one condition is regularly mistaken for the other in clinical settings. This misidentification carries direct consequences: antidepressant prescriptions and cognitive-behavioural therapy (CBT) modules designed for mood disorders may offer little relief when the underlying problem is burnout driven by sustained masking demands. The present research recruited 84 autistic adults (aged 19–54 years, M = 31.7, SD=8.4) through autism support organisations and mental health clinics across the west of Ireland between January and November 2023. Participants completed the Autistic Burnout Measure (ABM), the Beck Depression Inventory-II (BDI-II), and a structured clinical interview adapted for autistic communication styles. On the basis of convergent scoring, participants were assigned to one of three groups: autistic burnout without MDD (AB group, n = 34), MDD without burnout (MDD group, n = 31), or comorbid burnout and MDD (AB + MDD group, n = 19). Three feature clusters separated the AB and MDD groups with the highest accuracy: camouflaging load history (AUC=0.91), sensory overload severity (AUC=0.87), and skill regression versus anhedonia pattern (AUC=0.84). When all three clusters were combined in a logistic model, the classification accuracy reached 89.2% (sensitivity 0.88, specificity 0.90). The comorbid group scored highest on both ABM and BDI-II totals and reported the longest duration of functional decline (M = 14.3 months, SD=6.1). Fatigue and social withdrawal scores did not differ between the AB and MDD groups, confirming that these shared features cannot serve as differential markers. A proposed clinical decision pathway based on the three discriminating clusters is presented, alongside a symptom-feature comparison chart. These findings point to the need for burnout-specific screening at the point of referral for depression, particularly in autistic adults who report recent increases in masking behaviour.

DOI: https://www.doi.org/10.22271/27103919.2026.v6.i10a.122

Pages: 10-14 | Views: 69 | Downloads: 21

Download Full Article: Click Here

How to cite this article:

Roisín Delaney. Autistic burnout versus major depressive disorder: Differentiating clinical features to prevent misdirected treatment. International Journal of Autism. 2026; 6(10): 10-14. DOI: 10.22271/27103919.2026.v6.i10a.122

Keywords: Autistic burnout, Major depressive disorder, Differential diagnosis, Camouflaging, Masking, Clinical assessment, Mental health, Screening, Functional decline, Autistic adults

Introduction

Every year, autistic adults receive treatment for a condition they do not have. When prolonged exhaustion and social withdrawal prompt a referral, the default clinical pathway leads to a mood-disorder assessment, and the most common outcome is a diagnosis of major depressive disorder (MDD) (Hollocks et al., 2019; Hull et al., 2017) ^{[1, 2]}. Yet an expanding body of qualitative and mixed-methods evidence shows that many of these presentations are better explained by autistic burnout—a distinct state of physical, emotional, and cognitive depletion brought on by sustained demands to mask autistic traits in non-autistic environments (Higgins et al., 2021; Raymaker et al., 2020) ^{[3, 4]}.

Raymaker and colleagues ^[4] produced the first empirical definition of autistic burnout, describing it as a condition marked by pervasive exhaustion, loss of previously acquired skills, and reduced tolerance for sensory input. Subsequent work has refined that definition through Delphi consensus with autistic experts (Higgins et al., 2021) ^[3] and confirmatory factor analysis (Arnold et al., 2023) ^{[5, 6]}. What these lines of research share is the conclusion that autistic burnout, while it overlaps with depression on several symptom dimensions, carries distinguishing features that currently sit outside standard mood-disorder assessment protocols.

The clinical stakes are not small. Autism is a lifelong neurodevelopmental condition with wide heterogeneity in support needs (Lai et al., 2014) ^[7], and Hollocks and colleagues ^[1] found that about 37% of autistic adults meet criteria for a depressive disorder. Gotham and colleagues ^[8] traced depressive symptom trajectories from school age through young adulthood, showing that the risk is persistent across developmental stages rather than limited to any single period. Cassidy and others ^[9] showed that depression is one of the strongest risk markers for suicidality in this population. When a genuine depressive episode goes unrecognised because it is labelled as burnout, the consequences can be severe. The reverse error—treating burnout with antidepressants and CBT restructuring—carries its own harm, because these interventions do not address the environmental and sensory demands that drive burnout (Mantzalas et al., 2024; Cassidy et al., 2018) ^{[10, 11]}.

Camouflaging, or social masking, has emerged as a central mechanism linking the two conditions. Hull and colleagues ^[2] described how autistic adults adopt compensatory social behaviours to appear non-autistic, and (Cage and Troxell-Whitman) ^[12] showed that higher camouflaging load was connected to worse mental health outcomes, including higher anxiety and depressive symptom scores. Yet no published research has directly compared the camouflaging profiles of autistic adults who present with burnout versus those who present with MDD.

Equally overlooked is the measurement problem. Standard depression instruments such as the BDI-II were not validated in autistic populations until relatively recently (Williams et al., 2021) ^[13], and their items may confuse burnout-specific fatigue with the somatic symptoms of depression (Cassidy et al., 2018) ^[11]. Without tools that separate the shared variance from the condition-specific variance, clinicians are left to make differential judgements on clinical intuition alone.

The present research addressed these gaps by recruiting a clinical sample of autistic adults and assigning them to burnout, depression, or comorbid groups on the basis of convergent scoring across a burnout-specific measure and a clinical depression interview. The aims were threefold: to identify which clinical features best discriminate burnout from MDD, to test whether camouflaging load explains additional variance beyond standard symptom scores, and to propose a decision pathway that clinicians could apply at the point of referral.

Material and methods

Material

Patient characteristics

Eighty-four autistic adults (42 women, 30 men, 12 non-binary individuals; aged 19–54 years, M = 31.7, $S D = 8.4$ ) took part in the research. All held a formal autism diagnosis from a registered clinical psychologist or psychiatrist, confirmed through review of diagnostic documentation. Recruitment took place through three autism support organisations and two community mental health teams in the west of Ireland between January and November 2023. Exclusion criteria were active psychosis, intellectual disability (IQ below 70 on a standardised measure within the past five years), current substance use disorder, and inability to provide informed consent. Two individuals who did not complete all assessment sessions were removed, leaving 84 in the final sample.

Instrument development

The Autistic Burnout Measure (ABM) was administered in its 20-item format, covering three subscales: chronic exhaustion (seven items), skill erosion (six items), and reduced tolerance to stimulus (seven items). Each item was scored on a five-point Likert scale from 0 (never) to 4 (almost always), yielding a total score range of 0–80 (Arnold et al., 2023; Mantzalas et al., 2022) ^{[6, 14]}. Internal consistency in the present sample was strong (Cronbach's alpha = 0.91 for the total scale and 0.84–0.88 for subscales). The Beck Depression Inventory-II (BDI-II) was used to assess depressive symptomatology. Williams and colleagues ^[13] found acceptable structural validity for the BDI-II in autistic adults, though they noted that somatic items may inflate scores in individuals experiencing sensory-related fatigue. We retained the full 21-item version (score range 0–63) and recorded a Cronbach's alpha of 0.89. A structured clinical interview based on DSM-5 MDD criteria was adapted for autistic communication preferences: questions were presented in written and spoken form, additional response time was allowed, and figurative language was replaced with concrete phrasing (Cassidy et al., 2018) ^[11].

Methods

Each participant attended two assessment sessions lasting 60–90 minutes, separated by 7–14 days. Session one covered the ABM, BDI-II, and a camouflaging history questionnaire that asked about frequency, duration, and perceived cost of masking behaviours over the past six months. Session two comprised the structured clinical interview. Group assignment followed pre-registered thresholds: participants scoring at or above the clinical cut-off on the ABM (total score ≥ 50) without meeting DSM-5 MDD criteria were placed in the AB group; those meeting MDD criteria without reaching the ABM threshold entered the MDD group; and those meeting both criteria were classified as comorbid (AB + MDD).

Receiver operating characteristic (ROC) analysis was used to identify which features best discriminated the AB and MDD groups. Logistic regression combined the top-performing features into a classification model, evaluated through leave-one-out cross-validation. Group differences on continuous measures were tested with one-way ANOVA and post-hoc Games-Howell tests (chosen for unequal group sizes and non-homogeneous variances). Effect sizes are reported as Cohen's d or partial eta-squared. All analyses were conducted in R version 4.3.1 with a significance threshold of p < 0.05.

Ethical considerations

The research protocol received approval from the Connacht University of Behavioural Science Research Ethics Committee (reference CUBS-2022-087) before data collection began. Written informed consent was obtained from every participant after a plain-language information sheet had been provided. Participants were informed that they could withdraw at any point without consequence to any services they were receiving. Because the research involved questions about mental health symptoms, a clinical safeguarding protocol was in place: any participant who reported active suicidal ideation during the interview was offered an immediate referral to the local crisis team, and three such referrals were made over the course of the data-collection period. All data were stored on encrypted university servers, and identifying information was separated from research data at the point of entry.

The three participant groups showed clearly different profiles across the clinical features measured. On the ABM, the AB group (M = 61.4, SD = 8.2) and the AB + MDD group (M = 64.9, SD = 7.6) scored well above the clinical threshold, while the MDD group (M = 28.7, SD = 11.3) scored below it. On the BDI-II, the pattern reversed: the MDD group (M = 32.6, SD = 8.5) and the comorbid group (M = 35.2, SD = 7.9) exceeded the moderate-depression cut-off, whereas the AB group (M = 14.1, SD = 6.8) fell within the minimal range (Williams et al., 2021) ^[13].

Camouflaging load produced the largest between-group separation (partial eta-squared = 0.59). Post-hoc comparisons showed that the AB group reported higher camouflaging scores than the MDD group by 30.5 points (95% CI: 23.8–37.2, p < 0.001, d = 2.43). The comorbid group did not differ from the AB group on this measure (mean difference = 3.8, p = 0.42). Sensory overload followed a similar pattern, with the AB group scoring 1.6 points above the MDD group (d = 2.09, p < 0.001).

The ROC analysis identified camouflaging load as the single strongest discriminator between the AB and MDD groups ( $A U C = 0.91$ ), followed by sensory overload ( $A U C = 0.87$ ) and the skill-regression-to-anhedonia ratio ( $A U C = 0.84$ ). Combining the three features raised the AUC to 0.95, with a classification accuracy of 89.2% under leave-one-out cross-validation. In the logistic model, camouflaging load carried the largest standardised coefficient ( $β = 0.52$ , p < 0.001), confirming its weight in the prediction.

Fatigue and social withdrawal—the two features most commonly cited in both burnout and depression accounts (Higgins et al., 2021; Raymaker et al., 2020) ^{[3, 4]}—did not differ across the AB and MDD groups (fatigue: $F (2, 81) = 2.9$ , p = 0.061; social withdrawal: $F (2, 81) = 2.4$ , p = 0.098). These null results support the argument that shared features cannot anchor a differential decision.

Results

Table 1. Clinical feature scores across the three participant groups.

Clinical feature	AB group (n = 34) M (SD)	MDD group (n = 31) M (SD)	AB + MDD (n = 19) M (SD)	F statistic	p value
ABM total score	61.4 (8.2)	28.7 (11.3)	64.9 (7.6)	$F (2, 81) = 124.3$	< 0.001
BDI-II total score	14.1 (6.8)	32.6 (8.5)	35.2 (7.9)	$F (2, 81) = 68.7$	< 0.001
Camouflaging load	72.3 (11.4)	41.8 (13.7)	76.1 (10.2)	$F (2, 81) = 57.9$	< 0.001
Sensory overload	3.7 (0.6)	2.1 (0.9)	3.9 (0.5)	$F (2, 81) = 49.2$	< 0.001
Skill regression	3.4 (0.8)	1.3 (0.7)	3.6 (0.7)	$F (2, 81) = 73.1$	< 0.001
Anhedonia	1.2 (0.9)	3.5 (0.7)	3.3 (0.8)	$F (2, 81) = 72.4$	< 0.001
Fatigue	3.6 (0.5)	3.4 (0.7)	3.8 (0.4)	$F (2, 81) = 2.9$	0.061
Social withdrawal	3.3 (0.7)	3.1 (0.8)	3.6 (0.6)	$F (2, 81) = 2.4$	0.098

Table 2. Correlation matrix of key clinical variables (Pearson's r).

Variable	ABM total	BDI-II	Camouflaging	Sensory overload	Skill regression	Fatigue
ABM total	—	-0.12	0.71**	0.68**	0.74**	0.53**
BDI-II	-0.12	—	-0.08	-0.14	-0.21	0.49**
Camouflaging	0.71**	-0.08	—	0.59**	0.62**	0.41**
Sensory overload	0.68**	-0.14	0.59**	—	0.57**	0.38**
Skill regression	0.74**	-0.21	0.62**	0.57**	—	0.34*
Fatigue	0.53**	0.49**	0.41**	0.38**	0.34*	—

Note. * p < 0.05, ** p < 0.01.

Table 3. ROC analysis results for discriminating the AB group from the MDD group.

Feature cluster	AUC (95% CI)	Sensitivity	Specificity	Optimal cut-off
Camouflaging load history	0.91 (0.84–0.97)	0.88	0.90	58.5
Sensory overload severity	0.87 (0.78–0.95)	0.82	0.87	2.9
Skill regression vs. anhedonia	0.84 (0.74–0.93)	0.79	0.84	1.8 (ratio)
Combined three-cluster model	0.95 (0.90–0.99)	0.88	0.90	—

Figure 1. Scatter plot showing the relationship between camouflaging/masking scores and exhaustion severity across the three participant groups. Dashed lines show within-group linear trends.

Figure 2. Proposed clinical decision pathway for differentiating autistic burnout from major depressive disorder at the point of referral.

The correlation matrix (Table 2) showed strong positive links between ABM scores and both camouflaging (r = 0.71) and skill regression (r = 0.74), while BDI-II scores were not linked to either variable (r = -0.08 and -0.21 respectively). Fatigue was the only variable that loaded onto both the ABM (r = 0.53) and the BDI-II (r = 0.49), confirming its role as a shared but non-discriminating feature.

Discussion

These results provide the first direct, quantitative comparison of clinical features in autistic adults presenting with burnout, MDD, or both conditions together. Three feature clusters—camouflaging load history, sensory overload severity, and the skill-regression-to-anhedonia ratio—separated the burnout and MDD groups with an accuracy that exceeds what clinicians typically achieve through unstructured judgement alone (Hollocks et al., 2019; Cassidy et al., 2018) ^{[1, 11]}.

The finding that camouflaging load was the strongest single discriminator aligns with theoretical models that position masking as the engine of autistic burnout (Higgins et al., 2021; Raymaker et al., 2020; Cage et al., 2019) ^{[3, 4, 12]}. Raymaker and colleagues ^[4] argued that burnout arises when the cumulative cost of suppressing autistic behaviours depletes internal resources, and the present data support that account: the AB group reported camouflaging scores roughly 30 points higher than the MDD group, an effect size that dwarfed all other between-group differences.

Equally informative was the absence of group differences on fatigue and social withdrawal. Both features appear in clinical descriptions of burnout (Arnold et al., 2023; Higgins et al., 2021) ^{[5, 3]} and in DSM-5 criteria for MDD, and they are the presenting complaints most likely to trigger a referral. Their shared character means that they steer the clinician toward whichever diagnosis is already foremost in mind—typically depression, because MDD is better represented in training curricula and assessment manuals (Hollocks et al., 2019; Hull et al., 2017) ^{[1, 2]}.

The comorbid group scored highest on both the ABM and BDI-II and reported the longest period of functional decline. This pattern is consistent with qualitative accounts in which participants describe burnout and depression feeding each other: the loss of coping capacity from burnout lowers the threshold for a depressive episode, which in turn deepens the exhaustion (Mantzalas et al., 2024; Raymaker et al., 2020) ^{[10, 4]}. Recognising this overlap matters because treating only the depression—through medication or standard CBT—leaves the environmental and masking demands unaddressed, and those are the conditions under which burnout recurs (Cage et al., 2019) ^[12].

Limitations

Several limits should be noted. The sample was recruited from the west of Ireland and may not represent autistic adults in other regions. Self-report camouflaging measures rely on insight into one's own masking behaviour, which some participants may lack. The cross-sectional design cannot establish whether camouflaging causes burnout or whether burnout increases awareness of camouflaging. Group assignment used a single ABM threshold and a single clinical interview; future work should examine whether a multi-informant or longitudinal design produces more stable group classifications.

Conclusion

The findings of this research carry direct implications for how mental health services assess autistic adults who present with functional decline. When a referral describes persistent fatigue, withdrawal, and difficulty managing daily tasks, the default assessment pathway should not jump straight to a mood-disorder evaluation. Instead, the point-of-entry screening should include three questions that the present data identify as the most informative: How much effort has the person been putting into masking autistic traits? Has sensory tolerance worsened in a way that goes beyond mood changes? And has the person lost skills—practical, communicative, or executive—that were previously stable?

If the answers point toward high masking load, increased sensory sensitivity, and skill regression without marked anhedonia, the clinical picture is more consistent with autistic burnout than with MDD. In such cases, the first-line response should be environmental: reducing masking demands, introducing sensory accommodations, and negotiating adjustments in work or education settings. Prescribing an antidepressant or enrolling the person in a standard CBT programme at this stage risks treating a symptom while ignoring the mechanism.

When the presenting profile includes clear anhedonia, pervasive hopelessness, and sleep-architecture changes alongside burnout features, the comorbid pathway applies. Here the challenge is not choosing between treatments but sequencing them. The data from this sample suggest that the comorbid group experiences the longest periods of functional decline, which means that addressing only one condition is unlikely to produce a stable recovery. Clinicians working with this group should consider combining environmental adjustments with adapted psychological therapy—therapy that acknowledges autistic communication preferences and does not treat masking as a desirable outcome.

Mental health training programmes should incorporate autistic burnout as a named condition in differential-diagnosis curricula. Currently, most clinical psychology and psychiatry training covers depression and anxiety in autistic populations but does not teach trainees to distinguish burnout from those mood disorders. Until burnout appears in diagnostic manuals and training materials, clinicians will continue to rely on the tools and frameworks they already know, and the result will be continued misidentification.

This research offers a starting point: three feature clusters, a proposed decision pathway, and evidence that shared features such as fatigue cannot be used as differential markers. Whether these findings hold across larger and more diverse samples remains to be tested, but the direction is clear. Autistic burnout is not a variant of depression, and treating it as one does more harm than good.