Measurement invariance of the Pediatric Quality of life index in a child and adolescent mental health population.

Abstract

Purpose: To evaluate the measurement invariance of the Pediatric Quality of Life Index (PedsQL) in a population of children and adolescents participating in mental health treatment.

Methods: We analyzed PedsQL data from three child and adolescent mental health service (CAMHS) services (walk-in, day hospital, intensive community treatment). We fit ordered-categorical confirmatory factor analysis (CFA) to the canonical four-domain model and a theory-refined variant with targeted cross-loadings. We then examined multi-group measurement invariance (configural, metric, scalar) across service streams, diagnosis, gender, race/ethnicity (White vs. racialised).) Finally, we assessed item-level bias (i.e., differential item functioning; DIF) with an ordinal logistic approach using an observed-score nested model sequence.

Results: For the CFA, we found that the four-domain structure fit well and that allowing two theory-guided cross-loadings-Physical items 7 and 8 onto the Emotional domain yielded a strong fit. We found full scalar invariance across all groups tested. For DIF testing, no items exhibited meaningful uniform or non-uniform DIF.

Conclusions: PedsQL domain scores demonstrate evidence of structural validity and preliminary evidence of measurement invariance across the CAMHS subgroups evaluated. This provides support for the use of PedsQL scores in CAMHS.