21721
Validity Testing of a Social Communication Classification System of Functioning for Preschool Children with Autism Spectrum Disorder (ASD)
Objectives: To examine the construct validity of the ACSF:SC across settings (i.e., home and clinic) and the spectrum of abilities of preschool children with ASD.
Methods: Support for ACSF:SC construct validation was proposed with a priori hypotheses by the research team based on Pearson correlation coefficients between ACSF:SC ratings (parents or professionals) and sub-domains of other measures. Construct validity was examined using concurrent ratings of the ACSF:SC with prominent ASD measures that evaluate function and/or social communication to assess: (1) convergent validity: how well the ACSF:SC correlates with measures possessing domains related to social communication; and (2) discriminant validity: weak correlations with unrelated functional domains. Professionals and parents completed the ACSF:SC ratings, and a concurrent self-report measure (Social Responsiveness Scale 2nd edition [SRS-2]). Parents also completed two interview-based measures: the Vineland Adaptive Behavior Scale-2 (VABS-2) and the Autism Diagnostic Interview-Revised (ADI-R) algorithm. For the ADI-R algorithm, only Social Interactions (section A) and Communication (section B) were completed.
Results: The sample included 36 children whose ACSF:SC and SRS-2 were completed by their parents. The mean age (SD) was 4.3 (0.9) years (85% male). 27 (75%) parents completed an interview to collect data for the VABS-2 and ADI-R algorithm. 33 professionals completed ACSF:SC and SRS-2 ratings. Descriptive data showed that children categorized as either verbal or non-verbal based on the ADI-R algorithm had an ACSF:SC across most of the 5 ability levels (Table 1). Mean composite scores for parent scores on the SRS-2 and VABS2 were generally increasing across improved functioning on the 5 ACSF:SC levels (i.e., ACSF:SC rating of 1 = highest ability) (Table 2). Convergent validity testing between the SRS-2 subdomain (Social Communication and Interaction [SCI]) and ACSF:SC showed statistically significant correlations for capacity (95% confidence interval[CI]), for parents: 0.52 (0.23,0.73) and for professionals: 0.63 (0.36,0.80), respectively. Discriminant validity comparing the ACFS:SC tool with an unrelated sub-domain of function on the VABS-2 (i.e., Motor Skills) demonstrated no significant correlations for capacity, -0.34 (-.64,0.04).
Conclusions: Without a gold standard, testing validity of a novel tool is challenging, requiring other measures that may be only partly related to its core construct (i.e., social communication functioning). These data demonstrated that ACSF:SC ratings by both parents and professionals were related to SRS-2-defined social communication. Parent ratings of children identified as having both verbal and non-verbal abilities also demonstrated abilities across most of the five ACSF:SC levels, indicating that social communication ability is not contingent on verbal ability. These data are promising starting point for an ongoing validation process of the ACSF:SC.