Predictors of Expressive Language Outcome over a Two Year Period in Very Young Pre-Verbal Children with ASD

Background:  Extant research suggests that up to two-thirds of children with ASD who are pre-verbal, or have significantly impaired expressive language in toddlerhood, will develop some degree of functional expressive language by age five (Tager-Flusberg & Kasari, 2013; Thurm et al., 2014). Nonverbal cognitive ability, autism symptom severity, adaptive social functioning, and use of specific social skills such as joint attention in toddlerhood have been associated with expressive language outcomes (i.e., remaining pre-verbal or exhibiting improved expressive language; Thurm et al., 2007).

Objectives: We examined associations between nonverbal cognitive ability, adaptive social functioning, autism symptom severity, and use of joint attention skills in toddlerhood and expressive language ability two years later in a sample of very young pre-verbal children with ASD. We sought to identify baseline indices of child functioning that predicted stability in expressive language impairment versus improved expressive language.

Methods: Participants were 56 children (13 girls) with ASD first assessed at 20 to 33 months (T1; M = 28 ± 4 months) and again two years later (T2; M = 52.8 ± 5 months, range = 43-66). Participants were pre-verbal at first assessment, defined by t-scores on the Mullen Scales of Early Learning (Mullen) expressive language scale below 30, which corresponds to the “Very Low” descriptive category. Nonverbal cognitive ability was operationalized with a composite of the Mullen Visual Reception and Fine Motor subscales. Social functioning was assessed by mother-reported Vineland Adaptive Behavior Scales Socialization standard scores. Autism symptom severity was assessed with the Autism Diagnostic Observation Schedule algorithm total score. The Early Social Communication Scales measured joint attention skills, which included response to and initiation of requests, reciprocal social play, and shared attention. Children were assigned to outcome group based on their T2 Expressive Language t-score. Children whose scores remained in the Very Low range (t-scores ≤30) were assigned to the “remained pre-verbal” group (N = 38), and children with Below Average or Average range scores were assigned to the “improved” group (N = 18). T1 age was included in analyses as a covariate.

Results: In bivariate analyses, baseline nonverbal cognitive ability, autism symptom severity, social functioning, and T1 age differed significantly by outcome group (ps < .05). The “improved” group was younger at T1 and demonstrated higher baseline nonverbal cognitive ability and social functioning and less severe autism symptoms than the “remained pre-verbal” group. Subsequent logistic regression revealed that T1 autism symptom severity significantly predicted group membership (β = -.35, p = .002; OR: .70, CI: .56–.88), as did T1 age (β = -.27, p = .007; OR: .77, CI: .63 –.93).

Conclusions:  Our results support the role of autism symptom severity in toddlerhood in predicting expressive language outcomes later in childhood (Weismer & Kover, 2015). Contrary to expectations, nonverbal cognitive ability and parent-reported social functioning were unrelated to expressive language outcome, though our power was limited (Thurm et al., 2014). Identifying early predictors of expressive language outcomes can inform much-needed additional research that focuses specifically on individuals with ASD who have limited expressive language.