Abstract
Objective. To evaluate construct validity and external reliability of EQ-5D (Child). Methods and Results. Data were collected within a study conducted in Italy: 457 subjects aged 8-15 years were recruited in primary and secondary schools and 25 Acute Linfoblastic Leukaemia patients were enrolled from two Italian Hospitals. These patients were compared with a sub-sample, similar for age and gender, of children enrolled at schools. Participants self-completed a core-set containing the current Italian version and a number of demographic, socio-economic, relationship (etc.) questions. After 10 days 129 children and adolescents completed the same questionnaire again. To test convergent and discriminant validity, correlations were tested between each EQ-5D item and VAS and each item of other tools included in the core set. To test concurrent validity, responses of children from the general population were compared with those of chronically ill children. Test and retest EQ-5D responses were compared to investigate external reliability. These results refer to 415 valid questionnaires from school children and adolescents (mean=11.7 years old, sd=2.2), 52.0% girls, 129 (31.1%) interviewed twice. There were no missing/not unique responses. There were statistically significant correlations between similar concepts, and no correlation between dissimilar concepts was found. Weighted Cohen K was 0.2–0.5 in every profile item (p<0.01 for Mobility, Self-care, Worry/Happiness/Sadness). VAS Intraclass-Correlation-Coefficient was 0.823. Leukaemia patients more frequently reported problems/trouble in every EQ-5D item and lower mean VAS levels. Different results were found in differently aged children. Conclusion. We verified the psychometric properties of EQ-5D Child version. The current version of the instrument shows relatively good psychometric proprieties but still deserves some improvements.