Sensitivity and responsiveness of the EQ-5D-3L in patients with uncontrolled focal seizures: an analysis of Phase III trials of adjunctive brivaracetam

Clara Mukuria, Tracey Young, Anju Keetharuth, Simon Borghs, John Brazier

Preference-based measures are required to measure the impact of interventions for cost-effectiveness analysis. This study assessed the psychometric performance of the EQ-5D-3L in adults with uncontrolled focal (partial-onset) seizures. Data from three Phase III studies of an antiepileptic drug (adjunctive brivaracetam; n = 1095) were used. Analysis included correlations between EQ-5D-3L and Quality of Life in Epilepsy Inventory (QOLIE-31P) and seizure frequency. Known group validity was based on ability of the EQ-5D-3L to discriminate between baseline QOLIE-31P total scores, seizure type and number of antiepileptic drugs using effect sizes (ES). Responsiveness assessed proportions reporting highest or lowest scores, overall change using standardized response means (SRM) and change by responder and clinician/patient evaluation groups using ES. Correlations were weak to moderate (ρ = 0.2-0.4) between EQ-5D-3L dimensions and QOLIE-31P subscales, apart from medication effects (ρ < 0.1); seizure frequency was not associated with either measure. Known group analysis had small ES. A quarter (24.9%) of patients had a baseline EQ-5D-3L utility score of 1 (full health) but lower average QOLIE-31P scores. SRMs were small (<0.1) in EQ-5D-3L compared with 0.1-0.4 for QOLIE-31P subscales. Results across the studies were mixed for responder status and clinician/patient evaluation of improvement for EQ-5D-3L. EQ-5D-3L had weak-to-moderate correlations with QOLIE-31P and varied with QOLIE-31P severity groups, but showed less responsiveness than QOLIE-31P. Given this lack of sensitivity, EQ-5D-3L may not be appropriate for measuring the impact of interventions in cost-effectiveness analysis in this population and disease-specific preference-based measures may be more appropriate.