Representing Behavioral Phenotypic Heterogeneity within Autism Spectrum Disorders (ASD): Questions and Answers from the Simons Simplex Collection

Background: In several studies, clinical diagnoses and/or certainty ratings made by clinicians within a research group have added to the value of standardized diagnostic and cognitive measures in longitudinal predictions for young children with ASD.

Objectives: The objectives of this study were to 1) assess how such clinical judgments, made across a number of tertiary clinical centers for autism, would compare to standardized assessments and 2) whether the clinical judgments would independently contribute to predictions of current impairment.

Methods: The sample consisted of 414 4 to 17 year-old probands with an autism spectrum disorder (clinical diagnoses: 65.7% autism, 21.7% PDD-NOS, 12.6% Asperger’s) from 12 North American clinics. All probands met CPEA-STAART criteria for ASD based on the ADI-R and ADOS (Risi et al., 2006).  With appropriate IRB approved consents, families were recruited through the Simons Simplex Collection, a genetic and phenotypic repository for families with one child with ASD and, in most cases, at least one typical sibling. Children with severe to profound nonverbal intellectual disability, cerebral palsy, extensive birth complications, Fragile X, or Down syndrome or first/ second degree relatives with ASD were not included.  Probands were assessed with the ADI-R, ADOS, a standard hierarchy of cognitive and language tests, and other behavioral and adaptive measures. Extensive efforts were made to ensure cross-site reliability of administration and scoring of the standardized measures, as well as procedures for making DSM IV diagnoses.

Results: Use of different clinical diagnoses within ASD was highly variable across sites, with some sites diagnosing most or all children with autism and other sites reporting much higher frequencies of PDD-NOS or Asperger, though the means and distributions of standardized scores on the ADI-R, ADOS and cognitive measures were very similar across sites, with a few exceptions.  Use of different ASD diagnoses were related to verbal IQ, age of proband and severity of repetitive behaviors, but also reflected comparative composition of the sample within a site (e.g., how an individual proband compared to other children seen at that site).  Adaptive scores and IQs for children with diagnoses of Asperger Syndrome were consistently higher than those of children with autism diagnoses; the relationship of PDD-NOS to either of the other diagnoses, adaptive functioning and all other measures was highly variable, as were proportions of children receiving different diagnoses within very similar distributions of standardized scores.

Conclusions: In the past, within-site clinical judgments of subtypes of ASD (autism, PDD-NOS, Asperger) have offered a useful source of information about functioning for young children with spectrum disorders.  However, in a multi-site study of older children, such judgments were far less interpretable.  Predictable factors could be identified within most sites which related to different clinical diagnoses, however, the particular factor of greatest influence varied considerably across sites, even when a standard battery of instruments was used and procedures were highly similar.  Implications for the validity of clinical diagnoses within ASD will be discussed.