Quality-adjusted life expectancies in patients with rheumatoid arthritis - Comparison of index scores from EQ-5D, 15D, and SF-6D

摘要

Background: The aim of this study was to compare quality-adjusted life expectancy (QALE) for patients with rheumatoid arthritis generated from three generic health-related quality-of-life instruments. Methods: Patients from 11 Danish rheumatology outpatient clinics were asked to report current health status using the EuroQol five-dimensional questionnaire (EQ-5D), 15D, and six-dimensional health state short form (derived from SF-12) (SF-6D) instruments. Clinical staff provided data on current disease status (C-reactive protein and Disease Activity Score that involves clinical assessment of 28 joints). National mortality data were retrieved from Statistics Denmark. For each of the three instruments, mean index scores were estimated by gender and 5-year age groups. Partial QALE was estimated for the age interval 30 to 79 years for different subsamples. Results: Although the three quality-of-life index scores were highly correlated, there were statistically significant differences between the average index scores from the three instruments. The 15D provided the highest index score and SF-6D the lowest score. For a 30-year-old patient, the partial QALE ranged from 37.9 quality-adjusted life-years using the SF-6D to 45.6 quality-adjusted life-years using the 15D. The QALE for men and women differed by 6.2%, 4.0%, and 5.3% when the calculation was based on EQ-5D, SF-6D, and 15D index scores, respectively. The largest differences were observed when patients were grouped by functional status (Health Assessment Questionnaire score), where the EQ-5D showed a 50% difference in index score between the best and worst functional group while the SF-6D and 15D showed smaller differences (32% and 14%, respectively). Discussion: This analysis has shown the difference in QALE estimates related to different instruments. The study emphasizes that unless outcome studies use the same instruments and scoring algorithms, the results will not be directly comparable.