Prediction of Survival in Asian Patients Hospitalized With Heart Failure: Validation of the OPTIMIZE-HF Risk Score

Background: Risk scores predicting in-patient mortality in heart failure patients have not been designed specifically for Asian patients. We aimed to validate and recalibrate the OPTIMIZE-HF risk model for in-hospital mortality in a multiethnic Asian population hospitalized for heart failure. Methods and Results: Data from the Singapore Cardiac Databank Heart Failure on patients admitted for heart failure from January 1, 2008, to December 31, 2013, were included. The primary outcome studied was in-hospital mortality. Two models were compared: the original OPTIMIZE-HF risk model and a modified OPTIMIZE-HF risk model (similar variables but with coefficients derived from our cohort). A total of 15,219 patients were included. The overall in-hospital mortality was 1.88% (n = 286). The original model had a C-statistic of 0.739 (95% CI 0.708-0.770) with a good match between predicted and observed mortality rates (Hosmer-Lemeshow statistic 13.8; P = .086). The modified model had a C-statistic of 0.741 (95% CI 0.709-0.773) but a significant difference between predicted and observed mortality rates (Hosmer-Lemeshow statistic 17.2; P = .029). The modified model tended to underestimate risk at the extremes (lowest and highest ends) of risk. Conclusions: We provide the first independent validation of the OPTIMIZE-HF risk score in an Asian population. This risk model has been shown to perform reliably in our Asian cohort and will potentially provide clinicians with a useful tool to identify high-risk heart failure patients for more intensive management.