Autism Diagnosis Today (4): Late Diagnosis

In recent years, increasing attention has been devoted to the phenomenon of autism diagnosis in adolescence and adulthood. Waiting lists for assessment have grown rapidly. In England, for example, the average waiting time for an autism diagnosis reached 300 days, substantially exceeding the 91-day target recommended by NICE (Fagg & Woodhead 2023). The rising demand for assessment reflects a broader shift in public awareness and clinical practice.

Public discussion often presents this trend as the discovery of a “lost generation” of autistic individuals who were unrecognised in childhood (and excluded from a diagnosis of classic autism) because they supposedly masked their traits. This interpretation has been proposed by Lai and Baron-Cohen (2015), who suggested that increasing awareness and broader diagnostic criteria have allowed clinicians to identify individuals with autism who were previously overlooked.  

This explanation is intuitively appealing, but it deserves critical examination. The growing number of adult diagnoses may not simply represent delayed recognition of the same condition seen in childhood autism. Instead, late diagnosis may reflect diagnostic substitution, broader diagnostic criteria, the interaction between autistic traits and other psychiatric disorders, and changing cultural meanings attached to diagnostic labels.

Expansion of Autism Concept

One important factor behind the rise in late diagnosis is the broadening of the autism concept in recent diagnostic manuals. The publication of the DSM-5 in 2013 merged several previously separate conditions (including Asperger’s syndrome and pervasive developmental disorder-not otherwise specified) into a single category: autism spectrum disorder (ASD) (Zeidan et al. 2022; Huang et al. 2020).

At the same time, broader diagnostic criteria inevitably make the boundaries between autism and other conditions less distinct. Recent research notes that the expansion of diagnostic criteria has contributed to a reduction in observable differences between individuals diagnosed with autism and those without the diagnosis (Dufour et al. 2025). This trend creates additional challenges for clinicians attempting to distinguish autism from complex psychiatric presentations.

Diagnostic expansion is a well-known phenomenon in psychiatry and has also been observed in other conditions where broadening criteria increases prevalence without necessarily indicating a true rise in the underlying disorder.

Late-Diagnosed Autism as a Distinct Version of Autism

Recent genetic research further casts doubt on the assumption that late diagnosis merely reflects “milder” autism. Zhang et al. (2025) challenge this idea. They show that the developmental and genetic profile of autism differs substantially by age at diagnosis. Later diagnosed autism is not a milder version, but a different version. The researchers demonstrated that autism can be associated with two partially distinct polygenic profiles.

One genetic factor was associated with earlier diagnosis and early childhood social communication difficulties. The second factor was linked to later diagnosis, greater socio-emotional and behavioural difficulties during adolescence, and stronger genetic correlations with ADHD and other psychiatric disorders and was particularly evident among women.

These findings suggest that what is currently described as “autism” may in fact encompass multiple partially overlapping conditions ("autisms"), rather than a single homogeneous disorder.

Psychiatric Overlap and Diagnostic Timing

The situation is further complicated by the high prevalence of mental health problems reported in autistic populations.

Numerous studies report elevated rates of anxiety disorders, depression, obsessive–compulsive disorder (OCD), eating disorders, bipolar disorder, personality disorders (PDs) and psychotic disorders among autistic individuals. Anxiety disorders alone affect a substantial proportion of autistic adults, with prevalence estimates well above those observed in non-autistic populations.

Dufour et al. (2025) analysed diagnostic trajectories of 2,799 adults who received their first autism diagnosis between 2012 and 2017 in Quebec. The findings revealed extensive prior psychiatric contact. Over the study period:

·         77.5% had an anxiety disorder

·         58.0% had a depressive disorder

·         49.4% had a schizophrenia spectrum disorder

·         48.3% had bipolar disorder

·         42.9% had a personality disorder

·         33.2% had intellectual or developmental disabilities

These results indicate that adult autism diagnoses often occur in the context of multiple existing psychiatric diagnoses, rather than appearing in isolation.[1]

Another research study also suggests that the presence of multiple psychiatric disorders can significantly influence the age at which autism is diagnosed.

Kavanaugh et al. (2025) analysed more than 50,000 individuals with autism across two large datasets (RI-CART and SPARK). They found that the age of autism diagnosis increased with the number of co-occurring psychiatric disorders:

• no co-occurring conditions: mean diagnosis age 4.3 years

• one or two conditions: 7.1 years

• three or more conditions: around 10 years

Certain conditions were particularly associated with later diagnosis, including depression, ADHD and PDs, whereas conditions such as intellectual disability were associated with earlier identification.

These findings suggest that individuals who receive later diagnoses may represent a distinct subgroup, characterised by more complex psychiatric profiles rather than simply a milder form of the same condition.

Research also suggests elevated rates of suicidality in autistic populations. Studies examining suicidal experiences often highlight factors such as loneliness, hopelessness, trauma, bullying, and difficulties accessing support services. These findings are frequently interpreted as evidence that autistic individuals experience heightened vulnerability to mental illness due to social marginalisation. (Moseley et al. 2025).

While this may sometimes be the case, the relationship between autism and psychiatric conditions is likely to be complex and multidirectional.

Many psychiatric disorders produce behavioural patterns that resemble simplified descriptions of autistic traits. Social withdrawal, reduced eye contact, rumination, cognitive rigidity, and restricted interests can emerge in anxiety disorders, depression, trauma-related conditions, and several PDs (Fombonne 2020). E.g., social anxiety may involve avoidance of eye contact and interaction, while mood or PDs may produce rumination, social isolation, or narrow behavioural patterns.

In adulthood these patterns may converge into a behavioural profile characterised by social difficulties, repetitive thinking, and restricted activities. Developmental psychologists refer to this phenomenon as equifinality: different developmental pathways leading to similar outcomes.

The challenge becomes particularly acute because commonly used adult diagnostic tools are not fully specific. Research has shown that scores on autism assessment instruments can be significantly confounded by co-occurring psychiatric symptoms (Havdahl et al. 2016).

Reduced diagnostic specificity has also been documented when autism instruments are used in adults with schizophrenia, bipolar disorder, or other mood disorders (Matsuo et al. 2015).

In such contexts, distinguishing autism from other psychiatric conditions requires careful differential diagnosis, not merely the application of autism-specific assessment tools.

Consequently, adult behavioural patterns alone cannot reliably distinguish autism from other psychiatric conditions. Without clear evidence of early developmental abnormalities, retrospective interpretation becomes highly uncertain.

Diagnostic Substitution: From Personality Disorders to Autism

The overlap between autism and PDs has received particular attention. Several PDs share behavioural features with autism, including interpersonal difficulties, social withdrawal, rigid thinking, and emotional dysregulation. Schizotypal and avoidant PDs, for example, may involve limited social relationships and unusual communication styles, while borderline personality disorder (BPD) can involve identity disturbance and emotional instability.

These similarities create genuine challenges for differential diagnosis. Some qualitative studies have examined individuals who were first diagnosed with PDs and later received autism diagnoses. In one study of ten adults previously diagnosed with BPD, participants reported that the PD label felt highly stigmatising, whereas the autism diagnosis was experienced as validating and less blame-focused (Tamilson, Eccles, Shaw 2025).[2]

Such findings are often interpreted as evidence that autism had previously been misdiagnosed. However, the interpretation is not straightforward. Small qualitative studies cannot determine whether earlier diagnoses were incorrect, whether autism was present but unrecognised, or whether later autism diagnoses reflect reinterpretations of complex psychiatric histories.

More broadly, the shift from PD diagnoses toward autism may partly reflect changing cultural attitudes. PDs are often perceived as highly stigmatising and associated with personal blame, whereas autism is increasingly framed within the language of neurodiversity and difference rather than pathology. In such a context, autism may become a more acceptable explanatory label for longstanding psychological difficulties.

Misdiagnosis or Diagnostic Substitution?

A common explanation for late autism diagnosis is that individuals were previously misdiagnosed with other psychiatric conditions. Several studies indeed report long histories of psychiatric diagnoses preceding an autism diagnosis in adulthood.

For example, in a clinical study of adults receiving their first autism diagnosis, Fusar-Poli et al. (2022) found that the median age of diagnosis was 23 years, despite the first contact with mental health services occurring around age 13. Over this period, many participants had received alternative diagnoses, most commonly psychotic disorders, PDs, and depression.

These findings are often interpreted as evidence of widespread misdiagnosis.

However, the extensive symptom overlap suggests a more complex picture.

The overlap between autistic traits and other psychiatric syndromes is substantial, particularly in areas such as social cognition, emotional regulation, and interpersonal functioning.

A narrative review by Figueiredo and Caixeta (2025) highlights the diagnostic difficulty created by these shared features. For example:

·         both autism and schizotypal PD can involve social withdrawal and unusual interpersonal styles

·         autism and BPD may involve emotional dysregulation or identity disturbance

·         autism and avoidant personality disorder may present with similar patterns of social avoidance.

Because of these overlaps, distinguishing between autism and PDs often requires longitudinal developmental information, neuropsychological assessment, and multidisciplinary evaluation. In adults, such information is frequently incomplete or unavailable.

Thus, what is often described as “misdiagnosis” may in some cases reflect genuine diagnostic ambiguity.

Camouflage and the Myth of Hidden Autism

One of the most widely cited explanations for late autism diagnosis is the concept of camouflaging (or masking). According to this view, some autistic individuals — particularly females — learn to imitate typical social behaviour so effectively that their autism remains undetected until adolescence or adulthood. However, the explanatory power of this concept remains uncertain and the empirical evidence supporting this explanation remains limited.

First, social masking is not unique to autism. All individuals adjust their behaviour in social contexts in order to meet expectations, manage impressions, or avoid negative evaluation. Distinguishing autism-specific camouflage from ordinary social adaptation is therefore inherently difficult.

Second, autism is defined as a neurodevelopmental condition with early onset. Establishing the validity of an adult diagnosis therefore requires evidence that core features were present during early development.

Psychiatrist Eric Fombonne (2020) has argued that adult diagnostic evaluations must therefore include careful reconstruction of developmental history using multiple sources: family interviews, educational records, work history, and other documentation. Autism-specific assessment tools alone are insufficient, particularly because their scores can be influenced by co-occurring psychiatric symptoms such as anxiety or mood disorders. The claim that large numbers of autistic individuals remain completely undetected throughout childhood due to successful camouflage has also been questioned.

Moreover, the conceptualisation of therapeutic “masking” within this literature raises important questions. Participants reported that certain treatments encouraged them to modify or suppress behaviours, which they interpreted as harmful masking. Yet many psychotherapeutic approaches for PDs aim precisely to develop greater behavioural regulation, emotional control, and adaptive social strategies. From a clinical perspective, encouraging individuals to manage impulsivity or interpersonal conflict cannot automatically be regarded as harmful.

This does not imply that autism cannot occasionally be overlooked in childhood, particularly in individuals with subtle presentations, but rather that the hypothesis of widespread undetected autism requires stronger developmental evidence than is currently available.

“Female Autism Phenotype”

Closely related to the camouflage hypothesis is the idea of a distinct “female autism phenotype.” This concept proposes that autism manifests differently in females, often with subtler symptoms that evade traditional diagnostic criteria.

However, evidence supporting a clearly distinct female phenotype remains inconsistent. Comparative studies show only modest differences between males and females with autism. Females may report somewhat lower levels of restricted or repetitive behaviours, but differences in social communication and early cognitive abilities tend to be small (Kaat et al. 2020).

Furthermore, some observed differences may simply reflect general sex differences in behaviour that are also present in typically developing populations rather than autism-specific mechanisms. As Fombonne (2020) argues, it may be more appropriate to view these patterns as sex differences within autism, rather than evidence for a separate diagnostic entity.

Cultural Concept of Autism

The contemporary rise in adult autism diagnosis is also linked to broader cultural changes in how autism is conceptualised.

Fombonne (2020) notes that popular discourse often reduces autism to a simplified set of personality characteristics (e.g., “quirky,” introverted, or highly focused on specific interests). In such simplified forms, autism becomes almost indistinguishable from ordinary personality variation.

This conceptual shift risks conflating personality style, social anxiety, or life difficulties with neurodevelopmental disorder. When complex adult psychological outcomes are retrospectively mapped onto simplified autism narratives, the result may be an overinterpretation of behaviours that have multiple possible explanations.

Paradox of Subclinical Traits

A striking paradox emerges when comparing the goals of childhood autism intervention with the motivations of some adults who seek diagnosis.

Early intervention programmes aim to improve communication, social understanding, and adaptive functioning. When interventions are effective, some children eventually fall below diagnostic thresholds – a phenomenon described as loss of autism diagnosis (LAD). In other words, the clinical objective is often to reduce autistic traits to subclinical levels and improve adaptive functioning.

Yet in some cases adults seeking autism assessments present with traits that fall within this subclinical range. These may include social awkwardness, preference for solitude, intense interests, or sensitivity to sensory stimuli – traits that also occur within ordinary personality variation.

This contrast highlights a fundamental difference between childhood and adult perspectives on diagnosis. In childhood, clinicians aim to reduce problems and barriers to independent adaptive functioning until they no longer meet diagnostic criteria. In adulthood, similar levels of difference may motivate individuals to seek diagnostic recognition.

The reasons for this shift are likely varied. For some people, an autism diagnosis provides a coherent explanation for lifelong feelings of difference. For others, it offers access to supportive communities or identity frameworks. Increasing awareness through online communities and social media may also contribute to rising diagnostic demand.

Nevertheless, the paradox remains: behavioural traits that clinicians attempt to minimise in childhood may later become the basis for diagnosis in adulthood.

What Late Diagnosis Actually Tells Us

The rise in adult autism diagnosis is often framed as the discovery of a previously unrecognised “lost generation.” Yet the available evidence suggests a more complex picture. Rather than simply uncovering hidden autism, many late diagnoses appear to emerge at the intersection of broader diagnostic criteria, psychiatric comorbidity, and changing cultural interpretations of autistic traits.

In this sense, the phenomenon of late autism diagnosis may reveal as much about evolving psychiatric categories and contemporary cultural narratives as it does about autism itself.

_____________________

[1] In fact, for almost half of the cohort the first autism diagnosis occurred during hospitalisation, frequently within psychiatric services. Other studies reach similar conclusions. Adults who receive their first autism diagnosis often present to healthcare systems primarily because of mental health difficulties, including anxiety, depression, OCD, ADHD, or PDs.

[2] However, such studies must be interpreted cautiously. The research involved only ten participants, recruited through interviews, and therefore primarily reflects subjective narratives rather than objective diagnostic verification.