Utility of the Autism Diagnostic Observation Schedule and the Brief Observation of Social and Communication Change for Measuring Outcomes for a Parent-Mediated Early Autism Intervention

Abstract

Measuring outcomes for autistic children following social communication interventions is an ongoing challenge given the heterogeneous changes, which can be subtle. We tested and compared the overall and item-level intervention effects of the Brief Observation of Social Communication Change (BOSCC), Autism Diagnostic Observation Schedule (ADOS-2) algorithm, and ADOS-2 Calibrated Severity Scores (CSS) with autistic children aged 2-5 years from the Preschool Autism Communication Trial (PACT). The BOSCC was applied to Module 1 ADOS assessments (ADOS-BOSCC). Among the 117 children using single or no words (Module 1), the ADOS-BOSCC, ADOS algorithm, and ADOS CSS each detected small non-significant intervention effects. However, on the ADOS algorithm, there was a medium significant intervention effect for children with "few to no words" at baseline, while children with "some words" showed little intervention effect. For the full PACT sample (including ADOS Module 2, total n=152), ADOS metrics evidenced significant small (CSS) and medium (algorithm) overall intervention effects. None of the Module 1 item-level intervention effects reached significance, with largest changes observed for Gesture (ADOS-BOSCC and ADOS), Facial Expressions (ADOS), and Intonation (ADOS). Significant ADOS Module 2 item-level effects were observed for Mannerisms and Repetitive Interests and Stereotyped Behaviors. Despite strong psychometric properties, the ADOS-BOSCC was not more sensitive to behavioral changes than the ADOS among Module 1 children. Our results suggest the ADOS can be a sensitive outcome measure. Item-level intervention effect plots have the potential to indicate intervention "signatures of change," a concept that may be useful in future trials and systematic reviews. LAY SUMMARY: This study compares two outcome measures in a parent-mediated therapy. Neither was clearly better or worse than the other; however, the Autism Diagnostic Observation Schedule produced somewhat clearer evidence than the Brief Observation of Social Communication Change of improvement among children who had use of "few to no" words at the start. We explore which particular behaviors are associated with greater improvement. These findings can inform researchers when they consider how best to explore the impact of their intervention.

Keywords: Autism Diagnostic Observation Schedule; Brief Observation of Social Communication Change; autism spectrum disorder; intervention; outcome measures; trials.

Figure 1

Box plots with scatter of ADOS‐BOSCC, ADOS algorithm, and ADOS CSS totals at baseline and endpoint across intervention groups for Module 1. ADOS: Autism Diagnostic Observation Schedule; ADOS‐BOSCC: Brief Observation of Social and Communication Change‐version for ADOS; PACT: Preschool Autism Communication Trial; TAU: treatment‐as‐usual.

Figure 2

Forest plot of intervention effect size estimates for the Module 1 total and subdomain scores. Negative effect sizes are in favor of PACT. ADOS: Autism Diagnostic Observation Schedule; ADOS‐BOSCC: Brief Observation of Social and Communication Change‐version for ADOS; CSS: Calibrated Severity Scores; PACT: Preschool Autism Communication Trial; RRB: restricted and repetitive behaviors; TAU: treatment‐as‐usual.

Figure 3

Forest plot of intervention “signature of change”: Model effect estimates for the items of the ADOS‐BOSCC and ADOS (Module 1) with 95% confidence intervals corrected for multiple comparisons within each measure. This graph plots model estimates, while effect sizes are listed in the right‐hand column. Confidence intervals of the model estimates were corrected using the Dubey/Armitage‐Parmar adjustment, which accounts for there being multiple correlated outcomes. Fourteen items make up the ADOS‐2 Module 1 score, but four items differ depending on the language level of the child (indicated with dotted confidence intervals). The intervention models for “Response to Joint Attention” and “Intonation” are therefore only conducted with the 47 Module 1 children who remained in the “Few to No Words” category at both timepoints. The intervention models for “Pointing” and “Stereotyped Language” are only conducted with the 35 children who remained in the “Some Words” category at both time points. *Items marked with an asterisk had poor inter‐rater reliability. ADOS: Autism Diagnostic Observation Schedule; ADOS‐BOSCC: Brief Observation of Social and Communication Change‐version for ADOS; JA: joint attention; M1: Module 1.