Objectives: To identify predictors of expressive language (EL) gain in 2-year-olds with ASD enrolled in early intervention.
Methods: Forty-seven toddlers with ASD (23-33 months of age at enrollment) participated in a 6-month nursery school-based comprehensive early intervention for 10 hours per week. A validation sample was included, consisting of the next 28 children enrolled into the program. Pre- and post-treatment assessments included the Mullen Scales of Early Learning (MSEL), Autism Diagnostic Observation Schedule (ADOS), and Communication and Symbolic Behavior Scales Developmental Profile (CSBS). Multiple linear regression models were employed to examine which ADOS items independently predicted pre-post EL gain on the MSEL in both samples while controlling for pre-test EL and nonverbal cognitive level. The identified items were summed to create an ‘EL predictor' algorithm score. A non-parametric Receiver Operator Characteristic (ROC) procedure was then used to determine the most efficacious cutoff score that predicts children who gain in EL at a rate commensurate with increasing chronological age (gain of >6 months in age equivalent score on the MSEL Expressive Language scale between pre- and post-treatment assessments). The strength of the relationship between this novel algorithm and gains in functional EL was examined using an independent measure of EL gain: diversity of words used communicatively (CSBS word inventory score).
Results: Four predictors of EL gain were identified from the ADOS for the first sample: requesting, hand-finger complex mannerisms, functional play, and spontaneous joint attention (SJA) (p<.05). Analyses with the validation sample revealed that Requesting and SJA remained significant predictors (p<.05). Using a cutoff score of two (max=6), in the first sample, our EL algorithm accurately identified 94% of children who met our criteria for age-commensurate EL gain (sensitivity) and 56% of those who did not (specificity); ROC=.74. In the validation sample, this algorithm score accurately identified 84% of those with age-commensurate EL gain and 72% of those who did not meet our criteria for such gain (ROC=.78). Correlations revealed a strong to moderate (r2=.45-.53, p<.05) relationship between gains in requesting and SJA and the CSBS word inventory score.
Conclusions: These data provide strong evidence for the ability to communicate intentionally, albeit nonverbally, as a robust predictor of EL gain in toddlers with ASD enrolled in a classroom-based comprehensive early intervention. While the literature has indicated that social communicative intent is a predictor of EL gain, we offer evidence that imperative communicative ability is also an important predictor of such gain. Thus, a child's awareness of his/her ability to effect and engage with others may be a vital goal for the earliest stages of intervention and, establishing readiness for linguistic acquisition.