Predicting Growth of Communication Skills of Preschoolers with Autism Spectrum Disorder during Treatment in a Community-Based Pivotal Response Treatment Program

Background:  Understanding the variable responses of children with ASD to current early intervention (EI) programs is a research priority (Vivanti et al., 2014). Attempts to understand this variability have examined global factors (e.g., IQ and age) and their relations to particular outcomes (e.g., IQ changes; Howlin et al., 2009). Increasingly, researchers are attempting to move beyond these immutable factors to examine empirically and theoretically informed predictors (baseline child characteristics). Such research aims to elucidate mechanisms of change, and ultimately to match interventions to individual profiles. A set of variables that together predicted response to Pivotal Response Treatment (PRT; Koegel & Koegel, 2006), in single-case design research (Scherer & Schreibman, 2005; Schreibman et al., 2009) were baseline levels of appropriate toy contact (ATC), social approach, social avoidance, stereotyped and repetitive vocalizations (SRV) and non-verbal behaviours. These variable have not yet been examined as group-level predictors of outcome among children with ASD. An additional variable, baseline positive child affect, is also of interest, given its role in early communication (Kasari et al., 1990) and negative association with emerging ASD (Garon et al., 2009).

Objectives: To determine whether variables identified by Scherer & Schreibman (2005), and child affect, predicted expressive language (EL) outcomes in an unselected sample of preschoolers with ASD in a PRT program.  

Methods: Preschoolers with ASD (n=57) participated; mean baseline chronological age (CA) of 48.81 mo (SD = 8.97), mean baseline ratio IQ scores of 53.38 (SD = 20.35), and mean baseline severity of ASD symptoms (SRS T score) of 77.40 (SD = 9.99). Children were enrolled in a one-year community-based EI program (Bryson et al., 2007). The dependent variable, EL, was assessed by research examiners at baseline and after 12 months of EI, using the Preschool Language Scales, 4^th ed. (Zimmerman, Steiner & Pond, 2002). Independent variables (i.e., ATC, Avoidance, SRV and affect) were coded blindly from baseline videos of children interacting with a therapist or a research examiner.

Relations between putative predictors and EL outcomes were examined using hierarchical linear regression. Baseline EL was entered as a predictor with IQ and chronological age. Based on the patterns of bivariate correlations, ATC, Avoidance, SRV and Affect were entered in a second step.

Results: The first step accounted for 84% (p = .000) of variability in 12-month EL outcomes. Higher baseline EL and IQ scores significantly predicted greater 12-month EL. Step two accounted for an additional 4% (p = .005) of outcome variability. Baseline EL remained a significant predictor. More positive baseline child affect predicted greater 12-month EL outcomes; lower baseline ATC (p = .061) and avoidance (p = .056) were strong trends. 

Conclusions: Positive affect emerged as a predictor of EL outcome in a PRT-based program for preschoolers with ASD, contributing variance in addition to that accounted for by previously established predictors such as baseline language and IQ. The predictor profile described by Sherer & Schreibman (2005) was not observed in this sample.