Direct Assessment of Health Utilities Using the Standard Gamble among Patients with Primary Intracerebral Hemorrhage

Abstract

Background: Standard gamble (SG) directly measures patients' valuation of their health state. We compare in-hospital and day-90 SG utilities (SGU) among intracerebral hemorrhage patients and report a 3-way association between SGU, EuroQoL-5 dimension, and modified Rankin Scale at day 90. Methods and Results: Patients with intracerebral hemorrhage underwent in-hospital and day-90 assessments for the modified Rankin Scale, EuroQoL-5 dimension, and SG. SG provides patients a choice between their current health state and a hypothetical treatment with varying chances of either perfect health or a painless death. Higher SGU (scale, 0-1) indicates lower risk tolerance and thus higher valuation of the current health state. Logistic regression was used to estimate the likelihood of low SGU (≤0.6), and Wilcoxon paired signed-rank test compared in-hospital and day-90 SGU. In-hospital and day-90 SG was obtained from 381 and 280 patients, respectively, including 236 paired observations. Median (interquartile range) in-hospital and day-90 SGUs were 0.85 (0.40-0.98) and 0.98 (0.75-1.00; P<0.001). In-hospital SGUs were lower with advancing age (P=0.007), higher National Institutes of Health Stroke Scale, and intracerebral hemorrhage scores (P<0.001). Proxy-based assessments resulted in lower SGUs; median difference (95% CI),-0.2 (-0.33 to-0.07). After adjustment, higher National Institutes of Health Stroke Scale and proxy assessments were independently associated with lower SGU, along with an effect modification of age by race. Day-90 SGU and modified Rankin Scale were significantly correlated; however, SGUs were higher than the EuroQoL-5 dimension utilities at higher modified Rankin Scale levels. Conclusions: Divergence between directly (SGU) and indirectly (EuroQoL-5 dimension) assessed utilities at high levels of functional disability warrant careful prognostication of intracerebral hemorrhage outcomes and should be considered in designing early end-of-life care discussions with families and patients.