As a generic measure of health-related quality of life among patients awaiting THA or TKA, the three-level version of the EQ-5D (EQ-5D-3L), which has three response levels of severity (no problems, some problems, and extreme problems/unable) to five questions, is widely used. Previous studies indicated that the ceiling effect of the EQ-5D-3L limits its application. The five-level version of the EQ-5D (EQ-5D-5L) was developed to enhance the measurement properties of the tool by adding two levels: slight problems and severe problems. However, only a few small studies have compared the EQ-5D-3L and EQ-5D-5L in patients awaiting THA and TKA.
The purpose of this study was to examine the performance of the EQ-5D-3L and EQ-5D-5L among patients awaiting THA or TKA in terms of (1) response patterns, (2) convergent construct validity, (3) known-group validity, and (4) informativity and discriminatory power.
This is a retrospective analysis of the Alberta Bone and Joint Health Data Repository, which recorded information on all patients receiving hip or knee arthroplasties between April 2010 and March 2017 in Alberta, Canada (n = 37,377). Patients receiving THA or TKA and who completed the EQ-5D and WOMAC at baseline (presurgery) were included in this study (n = 24,766). The EQ-5D-3L was administered to all patients in 2010, and was gradually replaced by the EQ-5D-5L between 2013 and 2016; the EQ-5D-5L reached full application in all clinics by 2017.
A propensity score was used to match patients 1:1 who completed either the EQ-5D-3L or EQ-5D-5L before surgery. Response patterns have been explored using ceiling and floor effects and distribution across severity levels of each dimension. Convergent construct validity was examined using Spearman’s correlation (rho) against the WOMAC. Known-group validity was examined by gender, preoperative risk factors, mental health, obesity, and WOMAC physical function score. Informativity and discriminatory power were examined using the Shannon (H’) and Shannon evenness (J’) indices. A total of 3446 pairs of patients awaiting THA (55% women; mean age, 66 years) and 5428 pairs of patients awaiting TKA (59% women; mean age 67 years) were included in this analysis; the study group included all patients who were kept in the propensity score matching.
Ceiling and floor effects were comparable and small (less than 0.5%) for both versions; the responses across severity levels for each dimension were more evenly distributed for the EQ-5D-5L. Convergent construct validity was stronger for the EQ-5D-5L as it consistently had stronger correlations with the WOMAC overall and domain scores than the EQ-5D-3L (rho(3L-THA), -0.77 to -0.31; rho(3L-TKA), -0.71 to -0.24; rho(5L-THA), -0.71 to -0.17; rho(5L-TKA), -0.64 to -0.17; all p values < 0.001). The hypotheses of known-group analyses were confirmed for both versions. The EQ-5D-5L demonstrated stronger informativity and discriminatory power than the EQ-5D-3L, particularly for the mobility dimension (THA, H’(5L/3L)=1.66/0.37, J’(5L/3L)=0.72/0.23; TKA, H’(5L/3L)=1.66/0.41, J’(5L/3L)=0.71/0.26).
This study demonstrates the superior construct validity, and informativity and discriminatory power of the EQ-5D-5L compared with the EQ-5D-3L among patients awaiting THA or TKA.
Compared with the three-level version, the five-level version of the EQ-5D differentiates between patients awaiting THA and TKA much better based on their mobility, which is a key health aspect or outcome in these patients. Our findings suggest that the EQ-5D-5L is more appropriate for this population.