Convergent, Discriminant, Concurrent validity and Reliability of the EQ-5D(CHILD): Results
MetadataShow full item record
OBJECTIVES: Generic Quality-of-life (QoL) instruments allow to compare QoL in populations different for culture, languages, health. If utilities are derivable, QoL data can be used in economic research. Unfortunately, children often remain excluded from being part of these populations, hence information for optimal decision-making in pediatric context can be limited. The Child-Task-Force operating within the EuroQoL group is translating and validating an EQ-5D child version. Within the ChildTask-Force, the Italian team conducted a validation study involving 8–15 years old children in Italy. METHODS: Children from the general population were enrolled at schools, those with chronic diseases enrolled at hospitals. Participants self-completed a core-set containing: EQ-5D(child), open-questions on reasons why they reported “no”, “some”, “a lot of” problems with each EQ-5D-domain, standard measures for comparisons like PEDsQL, a scale to self-rate global health, the World-HealthOrganization health scale, questions on demographic, socioeconomic, health status. After 10 days some kids underwent to a re-test interview. Convergent and Discriminant validities were tested by computing correlation coefficients between every couple of items/scores, as appropriate; concurrent validity was tested by comparing the EQ-5D(child) responses of chronically sick versus general population children; reliability was tested by calculating agreement levels of test with re-test responses. RESULTS: In total, 415 valid children from schools (31.1% were re-tested), 11 Acute Linfoblastic Leukaemia patients form hospitals were enrolled. Relevant Spearman’s correlation coefficients (>0.2, p < 0.0001) were found between every conceptually similar items/scores; not significant coefficients were found between most of not conceptually similar items. Weighted Cohen’s Kappa was 0.2–0.5 in the profile items. IntraclassCorrelation-Coefficient for VAS = 0.715. Leukemia patients more frequently reported problems/trouble in every EQ-5D item, lower median (92.5 vs 98.0) and mean (83.0 vs 93.9) VAS. CONCLUSION: EQ-5D(child) shows acceptable levels of validity and reliability. Further research will be conducted to improve its properties and allow appropriate use in both research and clinical practice.