The EQ-5D-5L and minimal important change in Long COVID

Introduction: The EQ-5D-5L is the most commonly used patient-reported outcome measure in Long COVID (LC). Despite its frequent use, there have been few studies reporting LC-specific metrics to identify and interpret meaningful change. The aim of the study was therefore to determine the Minimal Clinically Important Difference (MCID) and Minimal Important Difference (MID) measures for the EQ-5D-5L in LC.

Methods: Data were collected from a national study (LOCOMOTION) evaluating LC services in the UK, involving participants completing the EQ-5D-5L on at least 2 occasions. The EQ-5D domains were categorised using Paretian classification of health states, and the probability of superiority was used to determine changes in health states over time. EQ-5D-5L profile scores were converted into health utilities using the UK-specific algorithm. The MCID was derived using 0.5 standard deviation and the MID by a 0.2 effect size.

Results: A total of 423 people (283 females, 67%) with LC completed the EQ-5D at 2 time points (median time interval: 196 days). Most participants reported problems in at least 1 EQ-5D domain. Only around 25% of participants noted some improvement. The MCID estimates were 0.11 for the EQ-5D-5L and 10.6 for the EQ-5D-5L VAS. The MID for the EQ-5D-5L was 0.03. Some differences in the change metrics were observed depending on baseline health states and timing of the follow-up assessment.

Conclusion: Long COVID specific estimates of the MCIDs and MIDs were derived for the EQ-5D-5L and EQ-5D VAS. The MCIDs will facilitate the evaluation and interpretation of meaningful change in patient health states in LC, both at the individual level and more broadly in health economic assessments of LC management, intervention and rehabilitation programmes.