Variation in the estimation of quality-adjusted life-years by different preference-based instruments

OBJECTIVES. To assess the interchangeability of preference-based health-related quality of life tools and compare the potential gains in quality-adjusted life years (QALYs) in patients with musculoskeletal disease. METHODS. Consecutive patients visiting a rheumatology clinic completed health-related quality of life assessments at baseline and 3, 6, and 12 months with the EuroQo1 (EQ-5D), Health Utilities Index (HUI3),and Short-Form 6D (SF-6D). Patients rated their health changes retrospectively and responses were categorized into three groups: better, same, and worse. Correlations and repeated measures analysis of variance with post hoc contrasts and a Bonferroni correction were used to assess interchangeability of tools. RESULTS. Results were based on 161 cases with complete baseline data and 98 cases with data at baseline and 12 months. Correlations ranged from 0.66 to 0.79. An interaction effect showed that for the better group, the EQ-5D showed a significantly greater mean improvement (0.15) than the HUI3 (0.07) or the SF-6D (0.05). For the Worse group, the EQ-5D showed a significantly greater mean decrease (0.19) than either the HUI3 (0.05) or the SF-6D (0.03). QALYs differences between the better and worse groups were significantly greater (0.23) with the EQ-5D than with the HUI3 (0.11) or the SF-6D (0.09). CONCLUSIONS. Although results moderately support the idea that the three tools are measuring a similar underlying construct, the tools are not interchangeable because they are scaled differently and produce varying results. These findings have potential implications for the interpretation and comparability of health outcome studies and economic analyses. Possible approaches are sensitivity analysis or standardization of scores before calculation of QALYs.