IScore

Abstract

Background— A predictive model of stroke mortality may be useful for clinicians to improve communication with and care of hospitalized patients. Our aim was to identify predictors of mortality and to develop and validate a risk score model using information available at hospital presentation. Methods and Results— This retrospective study included 12 262 community-based patients presenting with an acute ischemic stroke at multiple hospitals in Ontario, Canada, between 2003 and 2008 who had been identified from the Registry of the Canadian Stroke Network (8223 patients in the derivation cohort, 4039 in the internal validation cohort) and the Ontario Stroke Audit (3720 for the external validation cohort). The mortality rates for the derivation and internal validation cohorts were 12.2% and 12.6%, respectively, at 30 days and 22.5% and 22.9% at 1 year. Multivariable predictors of 30-day and 1-year mortality included older age, male sex, severe stroke, nonlacunar stroke subtype, glucose ≥7.5 mmol/L (135 mg/dL), history of atrial fibrillation, coronary artery disease, congestive heart failure, cancer, dementia, kidney disease on dialysis, and dependency before the stroke. A risk score index stratified the risk of death and identified low- and high- risk individuals. The c statistic was 0.850 for 30-day mortality and 0.823 for 1-year mortality for the derivation cohort, 0.851 for the 30-day model and 0.840 for the 1-year mortality model in the internal validation set, and 0.790 for the 30-day model and 0.782 for the 1-year model in the external validation set. Conclusion— Among patients with ischemic stroke, factors identifiable within hours of hospital presentation predicted mortality risk at 30 days and 1 year. The predictive score may assist clinicians in estimating stroke mortality risk and policymakers in providing a quantitative tool to compare facilities.