How well do standard stroke outcome measures reflect quality of life?: A retrospective analysis of clinical trial data

Background and Purpose: Quality of life (QoL) is important to stroke survivors yet is often recorded as a secondary measure in acute stroke randomized controlled trials. We examined whether commonly used stroke outcome measures captured aspects of QoL. Methods: We examined primary outcomes by National Institutes of Health Stroke Scale (NIHSS), Barthel Index (BI) and modified Rankin Scale (mRS), and QoL by Stroke Impact Scale (SIS) and European Quality of Life Scale (EQ-5D) from the Virtual International Stroke Trials Archive (VISTA). Using Spearman correlations and logistic regression, we described the relationships between QoL mRS, NIHSS, and BI at 3 months, stratified by respondent (patient or proxy). Using χ2 analyses, we examined the mismatch between good primary outcome (mRS ≤1, NIHSS ≤5, or BI ≥95) but poor QoL, and poor primary outcome (mRS ≥3, NIHSS ≥20, or BI ≤60) but good QoL. Results: Patient-assessed QoL had a stronger association with mRS (EQ-5D weighted score n=2987, <i>P</i><0.0001, <i>r</i>=−0.7, <i>r</i>²=0.53; SIS recovery n=2970, <i>P</i><0.0001, <i>r</i>=−0.71, <i>r</i>²=0.52). Proxy responses had a stronger association with BI (EQ-5D weighted score n=837, <i>P</i><0.0001, <i>r</i>=0.78, <i>r</i>²=0.63; SIS recovery n=867, <i>P</i><0.0001, <i>r</i>=0.68, <i>r</i>²=0.48). mRS explained more of the variation in QoL (EQ-5D weighted score=53%, recovery by SIS v3.0=52%) than NIHSS or BI and resulted in fewer mismatches between good primary outcome and poor QoL (<i>P</i>