Predictors of Treatment Response in the Social ABCs Parent-Mediated Intervention for Toddlers with ASD

Background: The Social ABCs is a parent-mediated intervention for toddlers with, or at-risk for, autism spectrum disorder (ASD). Following pilot evidence of efficacy (Brian et al., 2016), a recent RCT demonstrated that toddlers who received the Social ABCs made significant gains, relative to controls, in functional vocal responsivity to parent prompts, vocal initiations, positive affect, and social orienting; parents learned the techniques (achieved implementation fidelity) and reported increased empowerment (Brian et al., 2017). As with most behavioural/ developmental interventions, outcomes varied across individuals in both studies, raising the question of what factors were associated with degrees of responses. Identification of moderators and mediators of treatment response has been identified as an urgent research priority (Vivanti et al., 2014; IMFAR SIG, 2017), with potential to inform personalized intervention pathways and resource allocation.

Objectives: To explore factors associated with response to Social ABCs treatment.

Methods: Demographics, video-coded indices, and standardized assessment data from 50 parent-child dyads, all of whom had received the Social ABCs intervention, were included (mean child age = 24.0 months; SD = 4.5; range: 12-32 months). Putative predictors were examined using univariate ANOVA and linear regression, with child vocal responsivity (baseline to post-training change; z-transformed) as the primary outcome.

Results: The vast majority (98%) of parents achieved implementation fidelity, and 92% of toddlers demonstrated some improvement, but with individual variability. Regression analyses revealed two significant child-level baseline predictors, with small-to-medium effects (ƒ²): Expressive language (Mullen age equivalent; R² = .12, p = .016; ƒ² = .14), and rate of spontaneous initiations (R² = .09, p = .035; ƒ² = .10), with a negative association for both. Notably, parent fidelity of implementation post-training, but not at baseline, significantly predicted outcome (R² = .14, p = .008) with a medium effect (ƒ² = .16). The following baseline variables were explored but not associated with treatment response: ASD symptoms (ADOS-2 calibrated severity score; p = .40), toddler age (p = .38), parental education (p = .06), sibling status (p = .19), number of words used (p = .56). However, a significant effect was found for child sex (p < .001), wherein boys in our sample demonstrated a greater response. This is likely related to lower baseline expressive language in boys (Mullen t-score; p = .04), and the negative association between baseline language (age equivalent) and response. These relations need further examination.

Conclusions: Preliminary investigation of factors associated with treatment response revealed the importance of baseline language/communication abilities, with greater gains for toddlers who began the program with relatively lower verbal proficiency. This may be explained by a ceiling effect in some, but not all, cases. Parental implementation fidelity following training was the greatest single predictor of children’s progress, supporting the efficacy of the program. Severity of ASD symptoms and age were not associated with treatment response, indicating that this intervention is appropriate for a range of high-risk toddlers. Findings will guide development of a combined moderator/mediator model including interactions between factors, and examination of other key outcomes (e.g., positive affect, social orienting).