Effects of a Parent-Mediated Intervention on Expressive Language Among Toddlers with Autism: A Randomized Controlled Trial

Background: The purpose of this presentation is to report results of a randomized controlled trial pilot study of a parent-mediated intervention, focused primarily on enhancing parent support of toddler joint attention (Schertz et al., 2011). The Joint Attention Mediated Learning (JAML) program is a relationship based, developmentally oriented intervention that does not directly target expressive language promotion, but has consistently demonstrated large effects on joint attention (Schertz et al., 2013; 2017). Prior research has shown a relationship between establishment of joint attention and language development in young children with ASD (Markus et al., 2000; Bottema-Beutel, 2017). However, for preverbal toddlers with autism, it is not clear whether receipt of brief parent-mediated interventions, which do not intentionally target expressive language, are sufficient for improving expressive language outcomes.

Objectives: The current study examines effects of the Joint Attention Mediated Learning intervention on expressive language of toddlers with autism.

Methods: Participants included 30 toddlers diagnosed with autism. Participants were randomly selected from a larger sample (Schertz et al., 2017) to include a subset of 15 families who received the JAML program and 15 control group families. Demographics of the subset of 30 families are presented in Table 1. Pre- and post-intervention assessment videos of parent-child interactions at home were coded using the Early Communication Indicator (ECI; Greenwood, Walker, & Buzhardt, 2010). Based on direct observation, gestures, vocalizations, single words, and multiple words are coded for occurrence based on six minutes of direct observation of parent-child interactions in family homes. Coders were naïve to treatment condition and time point. Interobserver agreement (IOA) was examined for 20 percent of the videos. Mean IOA was 88%, ranging from 80% to 100% (SD=5.14). Analysis consisted of General Linear Mixture Models (GLMM), using lme function of the R nlme package, for parametric data (gestures). GLMM (Poisson, quasi-Poisson, and negative binomial) analyses were conducted with the glmmPQL function from the MASS package for non-parametrically distributed data, evidencing zero-inflation (single-words and multiple word). GLMM logistic regression, using lme4’s glmer function, analyses were conducted on whether or not any word was uttered.

Results: Analyses of gestures, single words, and multiple words did not show significant treatment by time differences within this low powered sample. However, there was a significant treatment X time interaction wherein any word utterance (Z=3.31, p < .001) increased significantly for the JAML group (Pre= 20%, Post=67%; OR=8.0[95%CI=1.52-42.04] p<.001), but not the control group (Pre=33%, Post=40%; OR=1.33[95%CI=0.30-5.91], p=.71). See Figure 1.

Conclusions: In contrast to families in the control condition, families receiving the JAML intervention showed significant pre- to post- assessment gains in any word utterance during interactions with their parents at home. Although the JAML program does not directly target language acquisition, it had large effects on expressive language of toddlers with autism. These results are encouraging and warrant sufficiently powered analyses with the full study sample and other studies. Limitations of this study as well as considerations for future study will be presented.