Systematic review reveals that EQ-5D minimally important differences decrease with increasing baseline score and may vary with treatment type.

Abstract

To provide an updated summary of published anchor-based Minimum Important Difference (MID) estimates for the EQ-5D index (EQ index) and visual analogue scale (EQ VAS) scores and identify factors influencing those estimates. We systematically searched eight electronic databases from January 1990 to March 2023. We examined the association of baseline score, type of score change (improvement/worsening), data source, value set, disease/condition, treatment type (surgical/non-surgical), and type of anchor (clinical vs self-rated) with MID estimates for the EQ-5D-3L and EQ-5D-5L indices, and EQ VAS. Significant variables were used to develop prediction formulas for MID by testing both linear and non-linear regression models. Of 6,786 records reviewed, 47 articles were included for analysis. MID ranges for improved scores were -0.13-0.68 (EQ-5D-3L), 0.01-0.41 (EQ-5D-5L), and 0.42-23.0 (EQ VAS). Surgical intervention and lower baseline scores were associated with higher MIDs for both the EQ indices but not for EQ VAS. The non-linear logarithmic model outperformed the linear model in predicting the MIDs. MIDs based on deteriorated scores were insufficient for quantitative synthesis (mean: -0.02 for EQ-5D-3L; -0.04 for EQ-5D-5L; and -6.5 for EQ VAS). This review revealed that the MID of EQ-5D index scores varies with baseline score and treatment type, indicating that use of a uniform MID may not be appropriate. We recommend using baseline score-adjusted and treatment type-specific EQ-5D MIDs, and call for more MID research, particularly in the context of assessing deterioration in health using this widely used generic health-status instrument.