Differential Responsiveness and Minimally Important Difference of EQ-5D: Truth or Artifact?
Author(s)
Ling Jie Cheng1, Hwee Weng Dennis Hey, MBBS, MCI2, Gabriel Ka Po Liu, MBBCh, MSc2, Nan Luo, PhD3.
1Postdoctoral Fellow, University of Oxford, Oxford, United Kingdom, 2National University Hospital, Singapore, Singapore, 3National University of Singapore, Singapore, Singapore.
1Postdoctoral Fellow, University of Oxford, Oxford, United Kingdom, 2National University Hospital, Singapore, Singapore, 3National University of Singapore, Singapore, Singapore.
OBJECTIVES: Responsiveness and minimally important difference (MID) are essential for interpreting EQ-5D scores in treatment evaluations. A recent review (Cheng et al., 2024) found consistently larger anchor-based MIDs in surgical versus non-surgical patients. However, it remains unclear whether this is due to baseline scores, treatment type, or patient characteristics. This study compared the responsiveness and MID of EQ-5D across perioperative and postoperative periods using data from a spine surgery registry.
METHODS: We analysed longitudinal EQ-5D-3L, ODI, and SF-36 item 1 data from 595 lumbar spine surgery patients (mean age: 57.1; 54.3% female) across four timepoints: preoperative (TP0), 6 months (TP1), 1 year (TP2), and 2 years postoperative (TP3). We evaluated the responsiveness of EQ index, VAS, and ODI using standardised response means (SRMs), and estimated MIDs for perioperative (TP0-TP2) and postoperative (TP2-TP3) periods using mean change and logistic regression methods. Subgroup analyses focused on patients with low baseline EQ index (≤0.5), VAS (≤50), or high ODI (≥40).
RESULTS: EQ VAS was responsive to both improvement and deterioration in perioperative (SRM: 1.25; 0.76) and postoperative periods (SRM: 0.61; 0.98). EQ index was responsive postoperatively but showed improvements perioperatively even in patients reporting no change or deterioration (SRM: 1.15, 0.88, 0.54). Mean change-derived MIDs were higher perioperatively than postoperatively for EQ index (0.40 vs 0.10) and VAS (17.5 vs 6.9), a trend that remained in subgroups. ODI results and logistic regression-derived MIDs mirrored these findings.
CONCLUSIONS: Findings suggest that both treatment type and baseline scores influence responsiveness and MID estimates when self-rated global health is used as the anchor. The observed treatment effects may reflect response shift, driven by elevated internal standards before or after surgery. These results underscore the need for caution when using global health ratings to anchor responsiveness and MID estimates for function-based HRQoL measures in surgical populations.
METHODS: We analysed longitudinal EQ-5D-3L, ODI, and SF-36 item 1 data from 595 lumbar spine surgery patients (mean age: 57.1; 54.3% female) across four timepoints: preoperative (TP0), 6 months (TP1), 1 year (TP2), and 2 years postoperative (TP3). We evaluated the responsiveness of EQ index, VAS, and ODI using standardised response means (SRMs), and estimated MIDs for perioperative (TP0-TP2) and postoperative (TP2-TP3) periods using mean change and logistic regression methods. Subgroup analyses focused on patients with low baseline EQ index (≤0.5), VAS (≤50), or high ODI (≥40).
RESULTS: EQ VAS was responsive to both improvement and deterioration in perioperative (SRM: 1.25; 0.76) and postoperative periods (SRM: 0.61; 0.98). EQ index was responsive postoperatively but showed improvements perioperatively even in patients reporting no change or deterioration (SRM: 1.15, 0.88, 0.54). Mean change-derived MIDs were higher perioperatively than postoperatively for EQ index (0.40 vs 0.10) and VAS (17.5 vs 6.9), a trend that remained in subgroups. ODI results and logistic regression-derived MIDs mirrored these findings.
CONCLUSIONS: Findings suggest that both treatment type and baseline scores influence responsiveness and MID estimates when self-rated global health is used as the anchor. The observed treatment effects may reflect response shift, driven by elevated internal standards before or after surgery. These results underscore the need for caution when using global health ratings to anchor responsiveness and MID estimates for function-based HRQoL measures in surgical populations.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
MSR77
Topic
Methodological & Statistical Research
Topic Subcategory
PRO & Related Methods
Disease
Musculoskeletal Disorders (Arthritis, Bone Disorders, Osteoporosis, Other Musculoskeletal)