Psychometric properties comparison between EQ-5D-5L and EQ-5D-3L in the general Thai population.

Abstract

Evidence for the EQ-5D-5L's psychometric properties in the general Thai population is limited. This study aimed to compare ceiling effect, discriminatory power, response redistribution, validity, reliability between the EQ-5D-5L (5L) and the EQ-5D-3L (3L) in the general Thai population. 1200 participants were randomly selected. The Shannon index ([Formula: see text] and Shannon evenness index ([Formula: see text] determining discriminatory power of both EQ-5D versions in each dimension were compared. Test-retest reliability was evaluated using weighted kappa (k) and intraclass correlation coefficients (ICCs). Validity was evaluated by correlations between similar dimensions of the EQ-5D, WHOQOL-BREF, and SF-12v2 and known-groups validity. The ceiling effects for the 3L and for the 5L were compared. The 5L had lower ceiling effects than the 3L (49.08% vs 57.17%, p < 0.01). [Formula: see text] was higher for the 5L than for the 3L, but [Formula: see text] showed otherwise. Moderate correlations were detected between similar dimensions of the EQ-5D and the WHOQOL-BREF and SF-12v2. ICCs and k of the 3L were slightly higher than those of the 5L (ICCs: 0.78 vs 0.71) and (k: 0.42-0.63 vs 0.48-0.61), respectively. Older, female participants and those with comorbidities reported a lower utility index for both versions. Evidence supported use of both EQ-5D versions in the general Thai population. The 5L had better ceiling effects and discriminant activity, while it showed comparable known-groups validity with the 3L. Nevertheless, evidence is limited for the superiority of reliability between these two versions, so more future research is required to investigate it.