Examining Reciprocal and Predictive Relations Among Social, Language and Emotion Regulation Skills across Early Childhood in a Longitudinal Sample of Children with ASD.

Background:

Developmental science draws on transactional models to understand stability and change across the lifespan (Sameroff & Mackenzie, 2003; Gottlieb, 2001). Phenotypic presentations of Autism Spectrum Disorder (ASD) are dynamic across development, and variability in individual symptom presentation is well documented across the lifespan. Early social and language skills predict later language, developmental, and social outcomes in ASD (Landa & Kalb, 2012; Smith, Mirenda & Zaidman-Zait, 2007), and emotion regulation (ER) skills have been identified as a key predictor of social and academic development within typically developing (TD) samples (McClelland & Cameron, 2011). Understanding how social, language, and ER skills interact to influence development may provide insight into the mechanisms and processes underlying developmental heterogeneity in ASD.

Objectives:

(1) To examine the stability of and reciprocal relations between social and language skills across early childhood, and (2) To model predictive relations between early ER and later social and language skills during this period.

Methods:

Participants were 277 children with ASD (26% female) drawn from the Autism Phenome Project (APP) at UC Davis MIND Institute. The APP is an ongoing longitudinal investigation of young children with ASD and TD. Participants completed comprehensive behavioral evaluations at study visits 1 (M_age = 35.67 months; SD = 5.68) and 3 (M_age=66.77 months; SD= 10.23). ASD diagnoses were confirmed using the Autism Diagnostic Observation Schedule (ADOS) and Autism Diagnostic Interview-Revised (ADI-R).

Models were developed to represent social, language, and ER latent variables using parent ratings (Vineland Adaptive Behavior Scales; Child Behavior Checklist; Social Responsiveness Scale) and clinician administered measures (ADI-R; ADOS; PPVT-III; Mullen Scales of Early Learning; Differential Ability Scales-II). Measurement models at visits 1 and 3 evidenced good fit (CFI/TLI > 0.93; SRMR <.08; RMSEA < 0.08). Latent variables allow for the estimation of specific constructs using multiple sources of information, while taking the contributions of measurement error into account, providing a more robust estimate of the hypothesized construct (Kline, 2016). Predictive and reciprocal relations between ER, and social and language skills were examined using structural equation modelling, controlling for the influence of early nonverbal IQ.

Results:

Models were estimated using MPlus software 7.4. Results documented stability in language (β = 0.80; p < 0.001) and social skills (β = 0.63; p < 0.001). Early language skills predicted later social skills (β = 0.23, p < 0.001), while early social skills did not account for unique variance in later language (β = 0.06, p = 0.47). Predictive associations were observed between early ER and later social skills (β = 0.19, p < 0.05), with fewer ER difficulties predicting better social skills. ER was not associated with later language.

Conclusions:

These findings suggest that early ER and language skills contribute to variation in social development in ASD. Social and language abilities were relatively stable across early childhood. Further investigation into the mechanisms by which early ER and language skills impact social development is necessary. Illuminating these pathways contributes to increased understanding of the developmental heterogeneity in ASD, and may inform assessment and intervention practices.