Understanding Pediatric Quality of Life in Autism Spectrum Disorder: Variability in Dimensions in Children with Varying Language Levels

Background: Expressive language abilities vary widely in children with autism spectrum disorder (ASD); studies estimate that as many as 30% do not acquire spoken language. Compared to verbal children, minimally verbal (MV) children encounter different challenges that may affect quality of life. There is an urgent need for studies to explore whether existing pediatric quality of life scales are appropriate for use in children with ASD across language levels in order to promote investigations of quality of life that inform development of interventions appropriate for children of different ability levels.

Objectives: To compare the factor structure of the Pediatric Quality of Life Inventory (PedsQL) in MV children with ASD to children with phrase speech and children with fluent speech.

Methods: A sample of 2,380 children was drawn from the Autism-Speaks Autism Treatment Network (ATN) registry using the following criteria: at least 5 years-old at assessment, with ADOS and PedsQL data available. A confirmatory factor analysis did not support the five PedsQL subscales (RMSEA>0.07 or CFI<0.9) for any language group (defined using ADOS module: MV=Module 1, Phrases=Module 2, Verbally Fluent=Module 3 or 4). Thus, an exploratory factor analysis (EFA) was performed in each language group using Quartimin rotation on a polychoric correlation matrix. Velicer’s minimum average partial test was used to determine the number of factors in each solution. Absolute values of standardized regression coefficients ≥0.4 were considered in the factor loadings. In the case of split loadings, the factor with the highest absolute value loading was chosen.

Results: EFA yielded three factors for the MV and Phrases groups and four factors for the verbally fluent children. Internal consistency fell below the acceptable cut-off (alpha=0.7) for one factor in each of the MV and Phrases groups. Across all three language groups, a factor comprised of 5 of the 8 original physical items emerged. Results from MV and Phrases groups also yielded a factor comprised of items positively loading from the Emotion and negatively loading from the Social problems scales, suggesting a factor reflecting negative experiences. In contrast, negatively loading Emotion items and positively loading School items comprised a factor for verbally fluent children that may reflect positive experiences. Verbally fluent children also showed negative loadings of Physical and positive loadings of Social items, possibly reflecting an adaptive function factor.

Conclusions: The original PedsQL factor structure was not supported in the present sample of children with ASD. Furthermore, exploratory factor analyses provided evidence for different empirically-derived scales for children of varying language levels. Taken together, results suggest that the existing PedsQL scales may not be appropriate to assess quality of life in children with ASD and highlight a need for research considering dimensions of quality of life for children with ASD with varying ability levels. Further analyses will be conducted to explore how other factors (e.g., severity of restricted and repetitive behaviors), in addition to language, relate to quality of life on these empirically derived scales.