External Validation of the iScore for Predicting Ischemic Stroke Mortality in Patients in China

Background and Purpose-The iScore is a prediction tool developed to estimate the risk of death in patients after hospitalization for an acute ischemic stroke. Our aim was to determine the accuracy of the iScore in patients with ischemic stroke in China. Methods-The iScore was used to predict 30-day mortality rate in 11 656 patients and 1-year mortality rate in 11 051 patients with acute ischemic stroke. These patients were identified from the China National Stroke Registry (CNSR) data set. Model discrimination was quantified by calculating the C statistic. The calibration was assessed using Pearson correlation coefficient. Results-The 30-day and 1-year mortality rates were 5.4% and 14.3%, respectively. The C statistics were 0.825 (95% confidence interval, 0.807-0.843) for 30-day mortality and 0.822 (95% confidence interval, 0.810-0.833) for 1-year mortality. The plots of observed versus predicted mortality rates showed excellent model calibration in the external validation samples from the CNSR (Pearson correlation coefficient, 0.925 for 30-day and 0.998 for 1-year mortality; both P<0.0001). Conclusions-The iScore reliably predicts 30-day and 1-year mortality in Chinese patients with ischemic stroke.