Measuring the Generalization of Skills in Response to Treatment in Minimally Verbal Children with ASD

Background: Measuring the generalization of skill acquisition and treatment response is crucial for the maintenance of learned behavior. For children with Autism Spectrum Disorder (ASD), generalizing specific skills across contexts can be particularly challenging (Klinger & Dawson, 2001). Given this challenge, researchers and clinicians often rely on a clinical judgment of improvement because there are few standard measures that capture changes seen outside of the intervention setting. Using the Brief Observation of Social Communication Change (BOSCC) may be one way to assess generalization of skills acquired during the clinician-child interactions to a broader context, such as parent-child interactions. 

Objectives: This study aims to explore the relationship between clinicians’ ratings of behavioral improvements observed during treatment of a child with ASD and blind raters’ observations of behavioral improvements during play sessions with a parent. The BOSCC, a novel, objective measure, is used to assess whether children with ASD are generalizing improvements made in intervention to interactions with a parent.

Methods: Data from 150 families participating in adaptive interventions for minimally verbal children with ASD will be presented. Children have a diagnosis of ASD, are between 4.5 and 8.5 years old, and used fewer than 20 spontaneous words during a 20-minute naturalistic language sample. The BOSCC is applied to 10-minute parent-child semi-structured play videos gathered across three time points (baseline, mid-intervention, post-intervention). The BOSCC is then coded by researchers blind to video time point and intervention status. Decreases in BOSCC scores, indicating symptom improvement, from baseline to mid-intervention and from mid- to post-intervention are compared in responders and non-responders defined by Clinical Global Impression ratings of Improvement (CGI-I) from the same time points. Analyses compare children who are able to generalize to other contexts (defined by decreases of >1 SD on the BOSCC) to the remaining children. Analyses also explore whether cognitive level, adaptive skills or ASD symptoms as measured by the Vineland Adaptive Behavior Scales (VABS) and the Autism Diagnostic Observation Schedule (ADOS) predict decreases on the BOSCC.

Results: Analyses of a subset of data show that approximately 60% of children are defined by CGI-I as responders to treatment and 25% decrease ­>1 SD on the BOSCC. These results suggest that although children are able to make improvements within the context of treatment with an individual interventionist, these changes may not be seen in other contexts, such as interactions with parents. No significant differences in cognitive level, VABS domains, or ADOS symptom severity were found between children who were defined as responders on the CGI-I and children who decrease >1 SD on the BOSCC. Additional analyses within a larger dataset will aid in elucidating these results.  

Conclusions: It is important for researchers to recognize the potential limitations of specific skill acquisition seen in singular research contexts. This may be particularly relevant for minimally verbal children with ASD. Consistent with other work, our findings suggest that generalization is a particular challenge for children with ASD and needs to remain a focus of intervention research.