Diagnostic Stability of Autism Before the Concept of Spectrum

The question of how stable an autism diagnosis is over time has long been debated. While many families and clinicians view an autism diagnosis as lifelong, research suggests that diagnostic stability—how consistently a diagnosis persists over time—depends heavily on which diagnostic framework is used. Emerging evidence indicates that before the introduction of the broader ASD category in the DSM-5 (APA 2013), autism diagnoses were more stable and predictable.

Before DSM-5: Distinct Categories, Clearer Boundaries

Prior to DSM-5, the major diagnostic systems—DSM-IV, DSM-IV-TR (APA 1994; 2000), and ICD-9/10 (WHO 1993; 2007)—did not use the term “Autism Spectrum Disorder.” Instead, they grouped several distinct conditions under the umbrella of Pervasive Developmental Disorders (PDDs), including:

• Autistic Disorder (AD)

• Asperger’s Disorder

• Childhood Disintegrative Disorder

• Pervasive Developmental Disorder – Not Otherwise Specified (PDD-NOS)

  
  

Each diagnosis had its own criteria, requiring different combinations of social, communication, and behavioural features. While these categories were sometimes criticised for being overly rigid, they did reflect an important reality — that different developmental patterns could look similar but were not identical. This structure reflected a belief that these were related but distinct conditions.

The Evidence for Diagnostic Stability

A major synthesis of evidence came from Woolfenden et al. (2012), who conducted a systematic review of 23 studies including 1,466 children with PDDs diagnoses. Their results highlighted significant differences in stability between diagnostic subtypes:

• Autistic Disorder (AD) showed high stability—between 53% and 100%, with the best studies reporting 88–89% stability.

• Other PDDs (e.g., such as PDD-NOS) showed far greater variability—between 14% and 61% stability in the best-quality studies.

  
  

These findings suggest that Autistic Disorder, which had stricter diagnostic criteria, was a relatively stable diagnosis. In contrast, the broader, less clearly defined PDD-NOS category was much less stable—especially in children diagnosed before age three. For example, in one meta-analysis of young children with PDD-NOS, stability was only 35% over a three-year period (Rondeau et al. 2011).

This reflected what many clinicians already observed: children with more narrowly defined autism were more likely to retain the diagnosis, while those with milder or atypical presentations often saw their profiles evolve over time.

Clinically, this meant that while most children diagnosed with Autistic Disorder continued to meet criteria over time, many with broader or atypical forms of PDD later moved “off the spectrum” or shifted to other developmental diagnoses. These included language disorders, intellectual disability, or emotional-behavioural conditions.

More recently, Iannuzzi et al. (2025) examined the long-term diagnostic stability of autism in a large Italian cohort using ICD-10 diagnostic categories—just before the transition to DSM-5’s unified ASD model. Their study found:

• Overall diagnostic stability within the F84 (PDDs): 73.6%

• Stability for F84.0 (Childhood Autism): 79.8%

• Stability for F84.9 (PDD-Unspecified): 68.2%

Only 2.8% of children moved completely outside the autism spectrum over the follow-up period, while 3.7% entered it later. These results reinforce earlier findings that ICD-10 autism-related diagnoses were relatively stable—especially for those who met full criteria for childhood autism.

DSM-5: From Autism as One of the PDDs to ASD as the Umbrella Term Encompassing Other Former PDDs

The publication of the DSM-5 in 2013 marked a major conceptual shift. Former PDD diagnoses — Autistic Disorder, Asperger’s, PDD-NOS, and others – were replaced by a single umbrella category: Autism Spectrum Disorder (ASD). The new model introduced three severity levels and various specifiers, reflecting a dimensional view of autism rather than distinct subtypes.

While this change simplified diagnostic practice and improved recognition of the spectrum’s diversity, it also blurred some of the diagnostic boundaries that previously supported stability. The broader spectrum inevitably captures a wider range of presentations—including children with milder or transient features – and, therefore, a lower overall stability rate over time.

Indeed, Woolfenden et al. (2012) cautioned that children with less specific ASD diagnoses, especially those identified in preschool years, were far more likely to lose their diagnosis later. Early identification is essential for timely intervention, but it also risks over-inclusion of children whose symptoms later resolve or change. As Thurm et al. (2019) noted, some early false positives reflect children with other neurodevelopmental issues that mimic autism features but fade as they mature.

*****

The evidence suggests that before autism became a “spectrum,” it was a more stable diagnosis. By including a wider range of developmental profiles under one label, the DSM-5 inadvertently introduced greater heterogeneity — and with it, more diagnostic fluidity.