OBJECTIVES: The increased use of patient-reported outcome measures (PROMs) in clinical practice, policy-making, quality improvement initiatives, and health services programming, have made the interpretation of PROM scores imperative. A minimally important difference (MID), defined as the smallest difference that patients perceive as meaningful, can help stakeholders better understand PROM scores. This systematic review aims to summarize, synthesize, and appraise published MID estimates for the EQ-5D instrument.

METHODS: MEDLINE, EMBASE, PsycINFO, CINAHL, Scopus, and the Cochrane Library were searched (on June 11, 2018) using database subject headings and text words relating to ‘minimally important difference’ and ‘EQ-5D’. Studies with an original MID estimate for the EQ-5D were included. Two teams of reviewers independently screened titles and abstracts, full texts, and extracted data in duplicate. We summarized MID estimates based on methodology, EQ-5D component, and clinical condition.

RESULTS: A total of 2,255 records were identified from our searches. After removing duplicates, 984 studies were screened of which 64 studies met our inclusion criteria for this review. Agreement between reviewers was substantial (Kappa=0.79). For anchor-based approaches, MID estimates were 0.74 to 35 for the VAS (16 studies), 0.01 to 0.64 for the 3L index (32 studies), and 0.01 to 0.14 for 5L index (5 studies). For distribution-based approaches, MID estimates were 2 to 41 for the VAS (20 studies), 0.02 to 0.429 for the 3L index (36 studies), and 0.052 to 0.102 for 5L index (9 studies). Instrument-defined MIDs (3 studies) have also been estimated for the 3L (0.04 to 0.082) and 5L (0.03 to 0.069) index scores. Estimates for different conditions were also summarized.

CONCLUSIONS: Our study identified a wide range of MID estimates for EQ-5D instruments, demonstrating that estimates may vary among different value sets, approaches, anchors, and conditions. These factors should be carefully considered when choosing an appropriate MID to interpret EQ-5D scores.