Autism Diagnosis Today (3): Loss of Autism Diagnosis

Autism has long been described as a lifelong neurodevelopmental disorder/condition. Core behavioural features, typically emerging in infancy or early childhood, have been assumed to persist throughout the lifespan. The research, however, suggests that within the vast heterogeneity of autism, not all individuals follow a single, fixed trajectory. Various reports have documented the presentation of "acquired autism" (regression) following a period of typical development. Less widely acknowledged, but increasingly supported by evidence, is the reverse phenomenon: some individuals who clearly met diagnostic criteria for autism in early childhood later function outside those thresholds. Autism, therefore, may not be innate or lifelong for everyone (Whiteley, Carr, & Shattock, 2019).

Isolated reports of children “growing out of autism” can be traced back to early psychiatric literature (Rutter 1970; von Knorring & Hägglöf 1993). Through the 1980s and 1990s, such reports remained largely anecdotal and lacked systematic follow-up. (At the time, such cases were usually dismissed as rare anomalies, misdiagnoses, or the effects of unmeasured interventions.) Further research, however, has confirmed that a subgroup of individuals who met criteria for ASD in childhood no longer meet diagnostic criteria later in development.

Terminology and definitions

Several terms have been used to describe this group, which capture the absence of current symptoms, co-occurring conditions, and functional impairment: “optimal outcome”, (Fein et al. 2013; Mukaddes et al. 2014); “optimal progress” (Moulton et al. 2016); “bloomers” (Fountain, Winter, Bearman 2012); “best possible outcome” (Costello & Maughan 2015); “very positive outcome” (Anderson, Liang, Lord 2014); “recovery” (O’Hara & Szakacs 2008; Mukaddes et al. 2014).[1] Eigsti, Fein, and Larson (2023) use the term loss of autism diagnosis (LAD) to describe individuals who no longer meet ASD criteria, assessing long-term outcomes not only through standardised measures and caregiver reports, but also from the individuals’ own perspectives.

Ozonoff (2013) explicitly challenged the dogma that autism must be lifelong. Citing Fein et al. (2013), who carefully documented individuals with incontrovertible early ASD diagnoses who later fell below diagnostic thresholds, Ozonoff argued that the possibility of “recovery” should not be treated as taboo. This marked a turning point in which LAD began to be taken seriously as a legitimate clinical and research phenomenon rather than dismissed as misdiagnosis or anomaly. This label emphasises the change in diagnostic status and captures a real and measurable shift in symptomatology.

Debates over language mirror deeper disagreements about what autism represents: a stable identity, a developmental disability, or a heterogeneous cluster of conditions with multiple developmental pathways. Importantly, when autism is discussed in terms of identity, it should be recognised that a medical diagnosis is not identity. Even when an autism diagnosis is no longer met, individuals may retain autistic traits or features associated with the broader autism phenotype, often at a subclinical level.

Evidence for LAD

A growing body of systematic research supports the existence of LAD, although it remains relatively uncommon. Prospective longitudinal studies suggest that approximately 9% of children with well-established ASD diagnoses later move off the spectrum (Fountain et al. 2012; Anderson et al.,2014; Moulton et al. 2016). Retrospective studies report comparable proportions (Mukaddes et al. 2014).

Key predictors of LAD include less severe early autistic behaviours, stronger adaptive and motor skills, and higher cognitive abilities (Sutera et al. 2007; Moulton et al. 2016). Early intensive intervention appears to play a significant role: Orinstein et al. (2014) found that children who lost their diagnosis had received significantly more therapeutic input between ages of two and three. Mukaddes et al. (2014) similarly linked LAD to early language development and higher intellectual functioning.

Outcomes following LAD are heterogeneous. Some LAD individuals demonstrate cognitive, social and communicative abilities indistinguishable from their typically developing peers (Fein et al. 2013). Others retain subtle pragmatic language differences (Suh et al. 2016), ADHD-like traits (Eigsti & Fein 2022) or increased vulnerability to anxiety and mood disorders (Horwitz et al. 2020). LAD is therefore best understood as one developmental trajectory among many, rather than a definitive or uniform endpoint.

Comorbidities and Broader Outcomes

An important question is whether losing an autism diagnosis is associated with broader improvements in mental health and daily functioning. Some findings are encouraging. In a long-term (for nearly 20 years) follow-up of males diagnosed with Asperger syndrome, Gillberg et al. (2016) found that those who no longer met ASD criteria were also more likely to be free of psychiatric comorbidities. Given the high prevalence of anxiety and depression among autistic individuals (Hollocks et al. 2019), this is noteworthy.

Yet results across studies are mixed. Shulman et al. (2019) reported that LAD does not consistently reduce psychiatric risk. More recent work by Sevin, Dogan and Ozbaran (2025) found that although LAD individuals showed higher cognitive capacity and lower symptom severity than their ASD peers, 80% still carried at least one DSM-5 diagnosis. Thus, LAD cannot be equated with the absence of clinical challenges. In short, LAD should not be equated with the absence of mental health difficulties.

Conceptual and Sociological Implications

The existence of LAD has significant implications for how autism is conceptualised. Scientifically, it undermines the idea of autism as a single, fixed entity. The move in DSM-5 toward a unified umbrella diagnosis (Harker & Stone 2014) has increased recognised heterogeneity, making it increasingly difficult to identify universal biological or behavioural markers. Acknowledging multiple developmental trajectories, including LAD, supports the view that “autism” may be better understood as a set of related disorders/conditions – “the autisms” – rather than a singular phenomenon (Whiteley 2017).

Sociologically, LAD intersects with debates over “autistic identity”. Autistic self-advocacy groups have expressed concern that framing LAD in terms of “recovery” stigmatises those However, to suggest that studying LAD itself is harmful risks conflating scientific inquiry with moral judgment.

The empirical fact remains that some individuals no longer meet diagnostic criteria for ASD. Disregarding this reality out of concern for stigma would undermine scientific integrity and reality of those who moved off the spectrum. For clinicians and families, recognising LAD can provide a more nuanced understanding of prognosis and encourage further investigation into the biological, environmental, and therapeutic factors that shape developmental trajectories.

Several questions remain: First, longitudinal research is needed to track LAD individuals into adolescence and adulthood, examining whether subtle social or cognitive difficulties re-emerge under new life demands (Eigsti, Fein, Larson 2023). Second, biological correlates of LAD may shed light on mechanisms of symptom remission, potentially clarifying whether autism is best understood as a single biological entity or a convergence of heterogeneous pathways (Eigsti & Fein 2022).

To treat LAD as taboo out of fear of stigmatisation is to deny families, clinicians, and autistic individuals themselves the knowledge that trajectories can differ. Autism is not a single story — and for some, that story includes losing the diagnosis.

__________________  

[1] Each term carries connotations — scientific, clinical, and sociocultural. “Recovery” suggests illness and cure, terms that resonate poorly with many ‘autistic self-advocates’. “Optimal” risks implying that remaining autistic is “suboptimal.” However, it is important to note that ASD is a medical diagnosis. Individuals who prefer non-medicalised framings do not need a formal diagnosis. Instead, their characteristics may be more appropriately described in terms of “subclinical autistic traits” or the “broad autism phenotype (BAP)”. These constructs allow for the recognition of autistic traits outside formal diagnostic boundaries and can be examined within psychological rather than strictly psychiatric paradigms (T.D. 2025).

Nonetheless, debates over terminology reflect deeper disagreements about what autism “is”: an immutable identity, a developmental disability, or a heterogeneous set of conditions with multiple trajectories.