Predictors of Initial Language Level and Rates of Language Growth in Young Children with ASD

Background:

There has been considerable interest in identifying predictors of language outcomes in young children with ASD (e.g., Bopp & Mirenda, 2011; Charman et al., 2003, 2005; Paul et al., 2008; Thurm et al., 2007). Prior investigations have typically involved relatively small samples or not included consideration of growth trajectories across time. Examination of a large, well-defined sample of young children at successive developmental levels is needed to provide a clearer understanding of predictors of language development over time in children with ASD. In addition to addressing this need, the current study is the first to investigate the role of calibrated ADOS severity scores as a potential predictor of language outcomes.

Objectives:

The goal of this study was to examine predictors of early language comprehension and expressive language abilities in toddlers with ASD and predictors of rates of language growth over time.

Methods:  

One hundred twenty-one children with ASD participated in a longitudinal investigation of language development. Mean age at each assessment was approximately 2 ½ (Visit 1), 3 ½ (Visit 2), 4 ½ (Visit 3), and 5 ½ years (Visit 4).  Autism spectrum diagnoses were determined using comprehensive evaluations including the ADI-R and ADOS/ADOS-T. Language abilities were measured by the Auditory Comprehension (AC) and Expressive Communication (EC) subscales of the Preschool Language Scale (PLS-4). Predictors included nonverbal cognition (Bayley-III), socialization (Vineland Adaptive Behavior Scales, VABS), maternal education as an index of SES, and calibrated ADOS severity scores (Gotham et al., 2009).

Results:  

A random slope and intercept model using Hierarchical Linear Modeling was fit to PLS-AC and PLS-EC standard scores across 2 ½ -5 ½ years. When considered individually, cognition, socialization, and maternal education were each significant positive predictors and ADOS severity was a significant negative predictor of both intercept and slope. Entering all four significant predictors into the model at once revealed that the 2 ½ year PLS-AC score (intercept) was primarily predicted by nonverbal cognition (p<.001), whereas the 2 ½ year PLS-EC score was predicted by nonverbal cognition and socialization (ps<.001). ADOS severity at 2 ½ years was a significant negative predictor of change across the three-year time span (slope) for PLS-AC (p=.004) and PLS-EC (p<.001). Additionally, nonverbal cognition at 2 ½ years was a positive predictor of PLS-EC growth over time (p<.001).

Conclusions:  

Nonverbal cognition was a robust predictor of language comprehension and expressive language level at 2 ½ years for toddlers with ASD, such that toddlers with higher nonverbal cognitive abilities performed better on language measures. Initial expressive language level was also positively predicted by socialization skills. Higher degrees of autism severity at 2 ½ years were predictive of slower rates of language growth across three years for both comprehension and production  and higher nonverbal cognition was predictive of faster rates of expressive language development.