Clinician-Derived Social Profiles Predict Play and Friendships with Peers in Children with Autism Spectrum Disorder

Background:

Children with autism exhibit deficits in peer play and social relationships, but there is great variability across individual children’s abilities in these areas. The severity and presentation of restricted/repetitive behaviors also vary across individuals, and there is a lack of information on the direct impact of these behaviors on peer interactions (McConnell, 2002). The heterogeneity in social skills and atypical behaviors makes it difficult to develop targeted treatments that address the challenges children with autism experience in peer play and friendships.

Objectives:

To examine the social profiles of children with autism based on clinician-child interactions in a large, well-characterized sample in order to predict parent-reported quality of peer play and friendship.

Methods:

Utilizing data from the Simons Simplex Collection (SSC), our participants included 1571 children (11.2% female) with ASD between the ages of 4 years and 17 years 11 months (M = 9.75 years, SD = 3.14 years) who completed the Autism Diagnostic Observation Schedule (ADOS) Module 3. We created five social profile composites based on select ADOS items: (1) Social Approach, (2) Social Response, (3) Reciprocity, (4) Nonverbal Behaviors, and (5) Atypical Behaviors. Quality of peer play and friendship was assessed using items from parent-reported responses on the Autism Diagnostic Interview-Revised (ADI-R). We conducted a series of simple linear regressions to determine whether social profile related to peer play and friendship. Second, we added socioeconomic status (SES), Full-Scale IQ, and age to the regressions to account for individual predictors. Finally, we converted each continuous outcome to a binary outcome and ran binary logistic regressions to determine how well our social profiles predicted the quality of peer play and friendship.

Results:

Out of the five social profiles, Social Approach was significantly related to quality of peer play (ß=.08, p=.030), and Atypical Behaviors was significantly related to quality of friendships (ß=.097, p<.001). When considering individual predictors, SES was significantly related to quality of peer play (ß=.08, p=.030), and both FSIQ (ß=-.087, p=.001) and age (ß=.060, p=.020) were significantly related to quality of friendships. A binary logistic regression showed Social Approach was marginally significant in predicting aberrant behaviors in peer play (ß=-.364, p=.064), and Atypical Behaviors significantly predicted aberrant behavior in friendship (ß=-.301, p<.001). Adding individual predictors strengthened the model (χ(3)=23.28, p<001).

Conclusions:

These findings suggest that aspects of social functioning observed in clinician-child encounters may predict quality of peer interactions. Of the five social profile composites, social approach predicted quality of peer play and atypical behaviors predicted quality of friendships. Because initiating interactions is the first step in engaging in peer play, increased difficulties with social approach may impact play negatively. The repetitive behaviors, restricted interests, and sensory needs common in children with autism may inhibit social engagement and, subsequently, interfere with the formation of friendships. Using a large sample and information from a widely-utilized diagnostic assessment, this study provides insight into which specific aspects of social competence should be targeted in ASD interventions.