Specifying Pervasive Developmental Disorder – Not Otherwise Specified: The Case for ‘Atypical Autism'

Background: It is an unfortunate paradox that Pervasive Developmental Disorder – Not Otherwise Specified (PDD-NOS) is both the most common and the least satisfactory diagnosis on the autistic spectrum. Its inter-rater reliability is unacceptably low, even amongst experienced clinicians. It has not been formally operationalised in DSM-IV-TR, and is effectively a ‘catch-all’ diagnosis, introduced by necessity in response to symptom constellations that fail to meet more rigorous criteria. PDD-NOS in its current form cannot be considered a useful diagnostic category, as it is not reliable and is unlikely to communicate with precision about an individual’s impairments, the source of their difficulties or their prognosis. As such, PDD-NOS has been described as ‘a work in progress’ and calls have been made for its more specific delineation. Accordingly, attempts have been made to identify homogenous sub-groups within PDD-NOS. It has been reported that around half of people with PDD-NOS have a specific symptom pattern involving significant impairments in reciprocal social interaction and communication, in the absence of repetitive and stereotyped behaviour. Szatmari and colleagues have suggested the ICD-10 term ‘atypical autism’ (AA) be used as a specific descriptor for this group. Whilst a number of commentators have endorsed this suggestion, fewer than 30 people with AA have been phenotypically described in the literature, making it difficult to judge its nosological value.

Objectives: To assess the validity of AA using the framework set out by Kendler for establishing psychiatric diagnoses. Specifically we sought to discover if AA has ‘concurrent validity’ by attempting to differentiate it from autism according to: (1) Core autistic features; (2) associated autistic features; (3) functional impairments commonly associated with autism.

Methods: Participants comprised 567 children consecutively diagnosed with a PDD at two children’s hospitals specialising in the assessment of high-functioning autism. Multi-informant (parent, teacher and child) data were collected using standardised measures (3Di, ADOS, Strengths and Difficulties Questionnaire, WASI). AA (n=228) individuals were compared to those with the full triad of autistic impairments (n=339).

Results: Thirty-nine percent of our PDD sample met criteria for AA. Compared to people with autism, the AA group showed milder social communication difficulties, but the overall pattern of these difficulties was similar in both groups. There were no group IQ differences. The AA group had fewer associated features of autism; showing less sensory sensitivity, fewer eating difficulties, better sleep, superior motor skills and fewer ADHD symptoms. Those with AA showed similar conduct and emotional difficulties to those with autism (home and school), but were less impaired in their peer relations.

Conclusions: ‘Atypical Austism’ as defined by Szatmari and colleagues is the most common group on the autistic spectrum, with a specific pattern of core PDD symptoms. Crucially, AA can be distinguished from full autism according to features independent of diagnostic criteria, and these differences are not attributable to IQ. Despite their lack of repetitive behaviours, children diagnosed with AA will need similar levels of support as those with full autism syndrome.