BACKGROUND: The Pediatric Quality of Life InventoryTM Generic Core (PedsQL:GC; Varni et al., 2003) is a widely used assessment of health-related quality of life (HRQoL) in clinical trials. A review of HRQoL assessments in autism spectrum disorder (ASD), concluded that the PedsQL:GC has good-to-excellent internal consistency (.78<ɑ<.91) in children with ASD (Limbers et al., 2009; Viecili & Weiss, 2015) as well as good convergent validity with the Child Behavior Checklist (CBCL) and Repetitive Behavior Scales (RBS-R; Ikeda et al., 2014). However, less has been published on other scales for convergent and divergent validity and longitudinal properties of the PedsQL:GC.
OBJECTIVES: Examine the psychometric measurement properties of the PedsQL:GC in a pediatric ASD population.
METHODS: PedsQL:GC caregiver-reported data from the Autism Treatment Network registry (e.g., Murray et al., 2016) was analyzed at baseline and three consecutive yearly intervals. The measure comprises 23-items in four dimensions. Higher scores indicate better HRQoL. Presented results relate to the total score. Data of 5602 mostly male (83%) children with ASD aged 2-17 years (Mage=5.6y, SDage=3.5y) were analyzed. Available data for different analyses varied between 50-4834 participants. Properties examined include dimension-to-total correlations to measure the internal structure; intraclass correlation coefficients (ICC) to assess test-retest reliability in children with 0- or 1-point change on the clinical global impression severity (CGI-S) scale; associations between the PedsQL:GC with related (convergent validity) and unrelated (divergent validity) constructs; known groups; and exploration of responsiveness.
RESULTS: Dimension-to-total correlations were strong (|r|=0.50-0.62) across all time points. ICCs demonstrated greater test-retest stability of PedsQL:GC total scores between years 1 and 2 (0.73) and years 2 and 3 (0.72), compared to baseline and year 1 (0.61). At baseline, PedsQL:GC total score validity correlations were small with the Vineland-II domains (|r|=0.15-0.29), moderate with the Aberrant Behavior Checklist domains (|r|=0.29-0.45), and moderate-to-strong with the CBCL domains (|r|=0.40-0.59). Divergent validity revealed nonsignificance with written and language scales. Counter to expectations, no correlation was found with the Autism Diagnostic Observation Schedule (|r|=0.004) and IQ (|r|=0.08). Lower PedsQL:GC total scores were observed for the subgroup with parental concern about social interactions (p<0.001), but there were no differences by CGI-S ratings. Responsiveness demonstrated a minimal change on PedsQL:GC total score from baseline to year 3 (M=-1.0, SD=15.23) with no differences in mean change between patients who were categorized as improved, not changed, or worsened based on CGI-S at any time point.
CONCLUSIONS: Results of these analyses demonstrated strong dimension-total correlations, consistent with previous research. Test-retest reliability was generally adequate but was based on a large 1-year follow-up interval. There was a moderate relationship with caregiver ratings of behavior and social problems, and less associations with clinical assessments. However, there was minimal change in the PedsQL:GC, which could potentially be explained by the use of a real world dataset with no specific intervention designed to impact HRQoL. These analyses add to the current literature on the use of PedsQL:GC in ASD, with shorter test-retest intervals and further longitudinal evaluation in an interventional dataset required.
