OBJECTIVES

: To examine the measurement properties of the EQ-5D-3L and EQ-5D-5L among total hip and knee replacement (THR/TKR) patients.

METHODS

: Patients (n=24,766) who had THR/TKR in Alberta, Canada between January 2010 and April 2017 and completed the EQ-5D (either 3L or 5L) and WOMAC before surgery were included in this study. In order to compare the two versions in balanced samples, a 1:1 propensity score based on age, gender, and WOMAC score was performed to match patients who completed the EQ-5D-5L and EQ-5D-3L.

Convergent construct validity was examined using Spearman correlation between the EQ-5D and WOMAC. Construct validity was also examined using known-groups defined by gender, number of pre-surgery risk factors, mental health, obesity, and WOMAC score. Discriminatory power were examined using Shannon and Shannon evenness indices.

RESULTS

: After matching, a total of 3,446 pairs of THR (55% female, mean age 66) and 5,428 pairs of TKR (59% female, mean age 67) patients were used in this analysis. The EQ-5D-5L index scores (THR, 0.424; TKR, 0.508) were significantly lower than the matched EQ-5D-3L index scores (THR, 0.526; TKR, 0.590). For each dimension, responses to the EQ-5D-5L were distributed more evenly across levels compared to the EQ-5D-3L, particularly for the mobility dimension (MO). The EQ-5D-5L (index and dimensions) consistently had stronger correlations with the WOMAC than the EQ-5D-3L. The null hypotheses of the known-group analyses were confirmed for both versions reflecting good construct validity. The Shannon and Shannon evenness indices showed that the EQ-5D-5L had stronger discriminatory power than the EQ-5D-3L (the largest improvement: MO).

CONCLUSIONS

: This study demonstrates stronger evidence of validity for the EQ-5D-5L compared to the EQ-5D-3L among THR/TKR patients, especially for MO, which is considered a very important outcome for this patient population. We recommend using the EQ-5D-5L for THR/TKR patients given its superior measurement properties.