We aimed to provide a comprehensive summary, synthesis, and appraisal of minimally important difference (MID) estimates for EQ-5D instruments. We conducted a systematic search using relevant terms related to "minimally/clinically, meaningful/ important difference/change" and "EQ-5D" in six major databases, including MEDLINE, EMBASE, PsycINFO, CINAHL, Scopus, and Cochrane Library (up to Jan 2023). We included studies that provided at least one original MID estimate for the EQ-5D. A total of 90 studies reporting 840 MID estimates were included. MID estimates for the EQ-5D-3L index score ranged from 0.075 to 0.8 using distribution-based approaches (239 estimates; 20 studies), from 0.003 to 0.72 using anchor-based approaches (189 estimates; 43 studies), and from 0.038 to 0.082 using instrument-defined approaches (4 estimates; 1 study). For the EQ-5D-5L, MID estimates ranged from 0.023 to 0.115 using distribution-based approaches (17 estimates; 12 studies), from 0.01 to 0.41 using anchor-based approaches (97 estimates; 15 studies), and from 0.037 to 0.101 using instrument-defined approaches (62 estimates; 8 studies). For the EQ visual analogue scale (VAS), MID estimates ranged from 0.96 to 16.6 using distribution-based approaches (87 estimates; 14 studies), and from 0.42 to 51.0 using anchor-based approaches (84 estimates; 24 studies). MID estimates varied by underlying clinical conditions, baseline scores, and direction of change. A wide range of MID estimates for EQ-5D instruments were identified, highlighting the variability of MID across populations, estimation methods, direction of change, baseline scores, and EQ-5D versions. These factors should be carefully considered when selecting an appropriate MID for interpreting EQ-5D scores.
Read full abstract