Validating psychometric properties of generic quality-of-life instruments (WHOQOL-BREF (TW) and EQ-5D) among non-dialysis chronic kidney disease: Rasch and confirmatory factor analyses

Background: Quality of life (QOL) is important for evaluating medical care outcomes. In chronic kidney disease (CKD) population, generic instruments, such as WHOQOL-BREF and EQ-5D, are commonly used for comparing various medical conditions for policy-making purposes. However, their psychometric properties have not yet been validated in non-dialysis CKD population.

Methods: Patients who were in CKD care programs at a medical center in Taiwan were recruited and self-filled WHOQOL-BREF Taiwan version (TW) and EQ-5D questionnaires. Rasch model, confirmatory factor analysis (CFA), and Pearson correlations were used to validate psychometric properties of these two QOL instruments.

Results: 634 participants, aged 66 ± 14 years old, were enrolled. Rasch model revealed almost all items (23 of 26) in WHOQOL-BREF (TW) and all items in EQ-5D were well-fit in their domains. Both instruments demonstrated relatively robust psychometric properties in CFA and Rasch analysis, with WHOQOL-BREF (TW) performing slightly better. Fewer items in WHOQOL-BREF (TW) displayed substantial differential item functioning compared to EQ-5D, particularly across different CKD stages (1 or 2 of 26 versus 4 of 5), diabetes mellitus (DM) and non-DM status (1 of 26 versus 4 of 5), and different age groups. The original domain scores/utility scores were highly correlated with Rasch scores (all absolute correlation coefficients >0.90).

Conclusions: Both WHOQOL-BREF (TW) and EQ-5D have relatively good psychometric properties in non-dialysis CKD population. The direct use of original WHOQOL-BREF (TW) domain scores/EQ-5D utility score for parametric analyses is acceptable non-dialysis in CKD population. However, when using the EQ-5D for cost-effective comparison in CKD patients with different stages or characteristics potential bias should be concerned.