Modeling Growth of Internalizing Symptoms from Childhood through Young Adulthood in Autism Spectrum and Developmentally Delayed Samples

Background: Previous reports of longitudinal patterns in internalizing symptoms within autism spectrum disorders (ASD) have focused on narrow childhood age ranges or used dependent variables not specific to affect or anxiety (e.g., Hallett et al., 2010; Taylor & Mailick, 2010). We modeled longitudinal change in individuals with ASD and developmentally delayed (DD) controls using subscales from a widely-used, well-validated measure of internalizing behavior, the Child Behavior Checklist 6-18 (CBCL; Achenbach & Rescorla, 2001).

Objectives: To (1) model growth in anxiety and depressive symptoms from late schoolage through young adulthood in a DD sample; (2) to evaluate differences in developmental patterns of internalizing symptoms based on diagnostic status (ASD versus nonspectrum) and verbal IQ (VIQ); (3) to assess relationships between internalizing growth patterns, baseline predictors (e.g., executive functioning, externalizing behaviors), and distal outcomes (e.g., functional independence, quality of life ratings), controlling for VIQ and autism severity.

Methods: Our sample included 158 participants (n=105 with ASD; n=53 with heterogeneous nonspectrum disorders or early DD) who had between 2 and 7 repeat iterations of the parent-rated CBCL. ASD: age ranged from 6-23 years (M=16 years; SD=4 years); Nonspectrum: 7-24 years (M=17; SD=3). ASD: VIQ ranged from 11-141 (M=62; SD=41); Nonspectrum from 10-139 (M=83; SD=31). As consecutive clinic referrals at age 2, most participants received full diagnostic evaluations around ages 2, 3, 5, 9, and 18-22 years, and families completed parent- and self-report questionnaires at regular intervals of 3–6 months between ages 9 and 24. CBCL Anxious-Depressed, Anxiety, and Affective subscale distributions were positively skewed, so outcomes were treated as count variables; we used mixed-effects Poisson models with a logarithmic function rather than traditional growth models which assume normally distributed residuals. All models contained random intercept and slope terms to capture between-person variability in initial symptom levels and rate of growth over time, with fixed effects of diagnosis, mean-centered age and VIQ, and diagnosis-age interaction.

Results: We observed an interaction between age and diagnosis for the Anxious-Depressed aggregate scale (b=0.005, 95% CI [0.002, 0.008]), with significant increases in symptoms over time in the ASD group alone (b=0.004, 95% CI [0.001, 0.006]). ASD was associated with higher Anxiety (b=0.491, 95% CI [0.068, 0.913]) and Affective (b=0.547, 95% CI [0.138, 0.957]) scores overall. Anxiety subscale scores increased significantly with age in both diagnostic groups, with no difference in rate of change over time (b=-.004, 95% CI [-0.012, 0.003]). Affective subscale scores did not change significantly over time in either group (b=-0.006, 95% CI [-0.015, 0.003]).  Higher VIQ was associated with higher levels of Anxious-Depressed (b=0.012, 95% CI [0.008, 0.016]) and Anxiety symptoms (b=0.006, 95% CI [0.002, 0.009]); VIQ was not associated with Affective scores (b=-0.002, 95% CI [-0.003, 0.006]). 

Conclusions: These preliminary findings support previous claims that individuals with ASD are at particular risk for affect- and anxiety-specific internalizing problems, and provide initial empirical support for the notion of anxiety symptoms increasing into adulthood within ASD. Findings on baseline predictors, including emotion regulation scores, and distal outcomes related to internalizing symptoms also will be presented.