Psychometric performance of proxy-reported EQ-5D youth version 5-level (EQ-5D-Y-5L) in comparison with three-level (EQ-5D-Y-3L) in children and adolescents with scoliosis.

Abstract

To assess the psychometric performance of proxy-reported EQ-5D-Y-5L (Y-5L) in comparisons with EQ-5D-Y-3L (Y-3L) administered by caregivers of patients with juvenile (JIS) or adolescent idiopathic scoliosis (AIS). A consecutive sample of caregivers of JIS or AIS patients were recruited. Redistribution property, ceiling effects, and discriminative power were examined. Known-group validity was determined by examining their ability to detect differences across clinical known groups. Test-retest reliability was assessed using intraclass correlation coefficient (ICC) for EQ-VAS score and Gwet's agreement coefficient (GAC) and percentage agreement (PA) for dimension responses. Furthermore, subgroups were analyzed for comparing test-retest reliability. A total of 130 caregivers were involved in the study. Consistencies between proxy-reported Y-3L and Y-5L were very high for all dimensions (93.8-99.2%). The ceiling effect in the Y-5L was slightly reduced in four dimensions (AR: 0.8-2.3%) whereas increased in "Having pain/discomfort". Greater informativity was found in the Y-5L than the Y-3L. In known-group comparisons of curvature magnitude, curvature type, and treatment modality, Y-5L and Y-3L dimension scales showed hypothesized results. For example, more full-health responses were found in the mild Cobb angle group (Y-5L: 63.1%; Y-3L: 62.2%) than the severe Cobb angle group (Y-5L: 55.6%, Y-3L: 55.6%). EQ-VAS score exhibited low test-retest reliability (ICC: 0.41), whereas dimension scales of both instruments showed satisfactory test-retest reliability (GAC ≥ 0.7 and PA ≥ 70% for all). In most known groups, hard-to-observe dimensions were more reliable for proxy-reported Y-5L than Y-3L. Both the proxy-reported Y-5L and Y-3L are valid and reliable instruments for assessing the HRQoL of JIS or AIS patients.