This Perspective argues that best-estimate clinical diagnosis must be grounded in rigorous developmental anchoring, collateral information, and judicious clinical judgment, and proposes a set of core stratification domains for systematic phenotypic stratification.
Abstract
The reported prevalence of autism spectrum disorder (ASD) has risen dramatically over the past two decades. Although increased awareness, broader diagnostic criteria, and improved access to assessment have corrected historical under-identification, this diagnostic expansion also raises a significant methodological concern: the risk of diagnostic dilution. Increasingly, surveillance systems and clinical cohorts may include individuals whose phenotypic profiles, developmental histories, and functional impairments do not fully align with a developmentally anchored neurodevelopmental presentation of ASD. This challenge is particularly acute in adolescent and adult assessments, where developmental history may be incomplete and standardized instruments or self-report measures may show limited specificity when applied to clinically complex psychiatric populations. Conflating developmentally anchored ASD with partially overlapping clinical phenotypes may reduce the signal-to-noise ratio in genetic, biomarker, neuroimaging, and therapeutic research, contributing to findings that are difficult to replicate or interpret. To preserve diagnostic validity, this Perspective argues that best-estimate clinical diagnosis must be grounded in rigorous developmental anchoring, collateral information, and judicious clinical judgment. It further proposes a set of core stratification domains for systematic phenotypic stratification, including age at first concern and diagnosis, biological sex and sex-related ascertainment factors, language and cognitive trajectories, adaptive functioning, intellectual disability, psychiatric comorbidities, ascertainment source, diagnostic instruments used, collateral developmental documentation, and support needs and functional impairment across contexts and over time. Stratification should not be understood as a restriction on clinical access, but as a scientific requirement for meaningful prevalence estimates and biologically informative autism research.
Background The reported prevalence of Autism Spectrum Disorder (ASD) has risen nearly fourfold over two decades, fuelling debate about whether DSM-5 boundaries still demarcate a coherent clinical category. Between 2020 and 2025, multiple independent research groups have produced a convergent body of critical reflection on this question. Aim This narrative review is an argumentative reappraisal addressing three convergent failures — classificatory, sociocultural, and methodological — and asking what minimum evidentiary standards should govern adult ASD differential diagnosis in complex cases. Four sub-themes traditionally treated separately (sensory profile, female phenotype, personality-disorder differential diagnosis, care-pathway implications) are presented as convergent illustrations of one underlying problem. Methods Narrative integration of peer-reviewed publications (2015–2026) on ASD diagnostic validity, phenotypic and genetic heterogeneity (Type I/Type II partition; Litman et al. SPARK analysis), sensory processing specificity, female phenotype and camouflaging, and differential diagnosis with seven conditions: Borderline, Avoidant, and Schizotypal Personality Disorders; Complex PTSD; ADHD with affective dysregulation; Bipolar Spectrum; OCD-spectrum disorders; and adult disorganised attachment. Literature-identification methods and AI-assisted search with author verification are detailed in Section 1.3. Findings The category aggregates at least two neurobiologically distinct phenotypes — Type I, prototypical, often syndromic, with high genetic load; and Type II, polygenically driven, milder, overlapping with general psychopathology — differentially affected by routine-assessment limitations. The DSM-5 sensory criterion, though neurobiologically grounded, lacks diagnostic specificity. The cross-sectional, single-source, self-report-based model dominating practice is structurally inadequate for Type II presentations and the female phenotype. Recommendations Six minimum standards are specified: structured developmental history with ≥2 informants; multi-context behavioural observation; neuropsychological profiling; granular sensory assessment by modality, direction, and contextual stability; systematic evaluation of alternative diagnoses; and longitudinal formulation with revisability. Specialised pathways should use stepped multidisciplinary triage when differential diagnosis remains unresolved, directing individuals to appropriate parallel or alternative services rather than denying care. A minimum feasible standard for under-resourced settings is articulated alongside the ideal one. Conclusion Restoring diagnostic specificity to ASD is not opposed to the neurodiversity framework. It is the precondition for ensuring that the diagnostic label, when applied, identifies a population for which evidence-based interventions exist and the care pathway is appropriate.
L. Croce, I. Fusaro· Frontiers in Psychiatry· 0 citations
A comprehensive review of recent advancements in ASD research, with particular emphasis on neuroimaging, artificial intelligence (AI), and machine learning (ML)-based diagnostic approaches, highlights the growing potential of AI-driven tools for supporting early ASD diagnosis and emphasizes the need for standardized protocols, external validation, explainable AI, and clinically translatable frameworks.
Kuljeet Singh, Khushi Mogha, S. Moctar· Neurological Sciences· 0 citations
This review systematically summarizes the conceptual definition and clinical phenotypes of profound autism and emphasizes the need to shift the intervention focus toward managing serious comorbidities and establishing augmentative and alternative communication systems.
Yang Xue, Fei-yong Jia· Zhongguo dang dai er ke za z...· 0 citations
There is an urgent need for prospective longitudinal studies to adequately map the aging trajectories of this population of older adults with ASD, which constitute a vulnerable population.
J. P. S. Gontijo, Marcelo H S S Martins, L. M. Alves· Trends in Psychiatry and Psy...· 0 citations
This critical narrative review examines early autism identification as a pathway problem rather than as a single testing event and argues that early autism identification should be evaluated through linked quality measures: response to concern, repeated surveillance following negative or ambiguous screening, referral completion, time to diagnostic assessment, support initiation before diagnostic closure, and equity of access.
R. Kurmashev, Mavile Karaieva· Pediatric Investigation· 0 citations
A narrative review of papers published between 1 January 2015 and 16 February 2026 that were pertinent to comorbidity and autism finds how the variable risk of comorbidity mimics the heterogeneity present in autism, thus inviting further investigations.
E. Dando, Juergen Hahn, D. Geier et al.· Healthcare· 0 citations
