Identifying Subgroups within PDD-NOS

Background:  The diagnostic criteria for Pervasive Developmental Disorder, Not Otherwise Specified (PDD-NOS) do not define behaviors necessary for the diagnosis. Rather, the diagnosis is assigned to children who exhibit some of the social and communicative impairments common to children with Pervasive Developmental Disorders but fail to meet the detailed criteria of other pervasive developmental disorders, (Walker et al., 2004). The lack of criteria for PDD-NOS suggests there may be a degree of heterogeneity within this population, yet there seems to be little research exploring the similarities and differences between children with PDD-NOS. Identifying subgroups of children within the PDD-NOS population may enhance our capacity for identifying, understanding, and providing services for these children.

Objectives:  To explore potential subgroups within children diagnosed with PDD-NOS at approximately 2 years of age.

Methods:  A hierarchical cluster analysis was conduced to detect clusters within a sample of 100 children diagnosed with PDD-NOS between 18-24 months based upon their subdomain scores from the Autism Diagnostic Observation Schedule (ADOS). Using a series of one-way ANOVAs, external validity of the selected clusters was assessed by comparing the groups’ scores on the Mullen Scales of Early Learning (Mullen) and Vineland Adaptive Behavior Scales (Vineland). 

Results:  Hierarchical cluster analysis suggests a 3 cluster solution best fits the data. Group 1 (n= 45) consists of participants with the highest mean scores (suggesting greatest impairment) on each ADOS subdomain when compared to Groups 2 and 3. Participants in Group 2 (n=24) have the lowest mean scores on each ADOS subdomain compared to Groups 1 and 3. Group 3 (n=31) participants’ mean scores for the combination of Language and Communication and Reciprocal Social Interaction subdomains, as well as the Play subdomain, fall between that of Groups 1 and 2.  Group 3’s mean score for Stereotyped Behaviors and Restricted Interests is most similar to that of Group 1. The one-way ANOVAs used to assess external validity of clusters indicated significant differences between the three clusters on the Visual Reception (F=6.686, p<.05), Receptive (F=8.268, p<.05), and Expressive Language (F=0.630, p<.05) subdomains of the Mullen, as well as the Communication (F=6.467, p<.05) subdomain of the Vineland.  No significant differences were found on the Fine Motor subdomain of the Mullen, nor on the Daily Living and Socialization subdomains of the Vineland.

Conclusions:  Preliminary analyses suggest that three clusters best describe this sample of children diagnosed with PDD-NOS. There appear to be subgroups of children characterized by limited (Group 2), mildly limited (Group 3), or moderately limited (Group 1) play skills, language, and communication abilities. However, the presence of stereotyped behaviors in Group 3 appeared more similar to Group 1 than Group 2. External validation analyses indicate that these clusters differ significantly on measures other than the ADOS. Subsequent analyses will follow children diagnosed with PDD-NOS at age 2 to determine whether cluster assignment remains stable at reevaluation at age 4 and whether cluster assignment at age 2 is significantly related to outcomes at age 4. We will also compare clusters based upon demographic characteristics (e.g. gender).