Note: Most Internet Explorer 8 users encounter issues playing the presentation videos. Please update your browser or use a different one if available.

Linear Versus Multiphase Models Testing Outcomes for Infants At-Risk for ASD in a Randomized Controlled Trial of a Parent-Mediated Intervention

Friday, 3 May 2013: 14:00-18:00
Banquet Hall (Kursaal Centre)
G. T. Baranek1, L. R. Watson2, L. Turner-Brown3, S. Field4, E. Crais2, L. Wakeford2 and L. M. Little5, (1)Allied Health Sciences, University of North Carolina at Chapel Hill, Chapel Hill, NC, (2)University of North Carolina at Chapel Hill, Chapel Hill, NC, (3)Psychiatry, University of North Carolina at Chapel Hill, Chapel Hill, NC, (4)Frank Porter Graham Child Development Institute, University of North Carolina at Chapel Hill, Chapel Hill, NC, (5)University of North Carolilna at Chapel Hill, Chapel Hill, NC
Background:  Studies of early behavioral interventions for toddlers with ASD are emerging (Carter et al., 2011; Dawson et al., 2010), but little research examines the efficacy of interventions with infants at-risk for a later diagnosis of ASD in community samples. Intervening with infants at-risk prior to full emergence of diagnostic symptoms may be more efficacious and has implications for prevention.

Objectives:  (1) Evaluate preliminary efficacy of a parent-mediated intervention (Adapted Responsive Teaching [ART]) for one-year-olds at-risk for ASD in improving child developmental outcomes and parent responsiveness.  (2) Compare the ability of linear versus multiphase growth models to summarize within-subject variation in outcome variables across two study phases (short-term: pre- to post-treatment; long-term: post-treatment to follow-up).

Methods: This RCT tested effects of ART versus a control condition (Community Services [CS]), using an intent-to-treat analysis. Infants identified through birth records were screened for ASD risk with the First Year Inventory at 12 months of age; scores >95th% were assessed and subsequently enrolled. Using a 2:1 randomization ratio, 16 families enrolled (ART=11; CS=5). ART families received a 6-month home-based intervention designed to enhance parent responsiveness and child social-communication and sensory-regulatory functions. We compared linear and multiphase models as models for growth across the two phases of interest.  Initial estimates for the unknown parameters from both models were obtained using maximum likelihood estimation (MLE).  Small sample sizes motivated us to adopt a Bayesian simulation approach  (Yuan & MacKinnon, 2009).We simulated sampling from the posterior distributions of model parameters using the Monte Carlo Markov Chain (MCMC) method implemented by the software WinBugs (Lunn, Best & Spiegelhalter, 2000).

Results: Despite the small sample, interval estimates from MCMC and MLE were largely similar; thus we report the more commonly used MLE method.  The more parsimonious linear model provided a better fit for only a few variables, 3 of which demonstrated statistically significant associations between treatment assignment and post treatment growth (i.e., ART improved CSBS-CQ total score & Mullen Receptive Language, and worsened on SEQ Hyperresponsiveness).  With multi-phase models, statistically significant associations between treatment status and growth were concentrated in the first (active treatment) phase of the study.  After the first phase, 5 outcomes had significant treatment impacts – specifically, ART parents were less directive (MBRS) and their children showed less sensory hyporesponsiveness (SEQ) and higher levels of expressive and receptive communication and socialization (Vineland).  In contrast, statistically significant differential growth by treatment status in the second, follow-up phase was largely absent.  AIC statistics indicated that the multi-phase model provided a better fit than the linear model for all 5 of these variables. 

Conclusions:  Although some variables showed linear trends, the multi-phase model was preferred in most cases.  Findings supported greater gains for ART versus CS, although these gains were concentrated in the first (pre- to post-treatment) phase, while growth in the second phase (post-tx to follow-up) was substantially attenuated. Gains in socialization, communication, and sensory functions, as well as in parent behaviors, provide preliminary support for this parent-mediated intervention.  Issues of maintenance of treatment effects require further research.

| More