External validation of ACUITY/HORIZON bleeding risk score among acute coronary syndrome patients in Thai PCI Registry.

Background External validation is essential before implementing a predictive model in clinical practice. This analysis validated the performance of the ACUITY/HORIZON risk score in the most contemporary Thai PCI registry. Methods The ACUITY/HORIZON model was applied and validated externally in 12,268 ACS (acute coronary syndrome) patients. For revision and updating models, the regression coefficient of all predictors were re-estimated and then additional predictors were stepwise selected from multivariate analysis. Results In-hospital bleeding defined by the BARC (Bleeding Academic Research Consortium) criteria was 1.3% (161 patients) and 2.3% (285 patients) by the ACUITY criteria. The calibration of both scales demonstrated overestimation of the original model with C-statistic values of 0.704 for ACUITY major bleeding and 0.793 for BARC 3 or 5 bleeding. For ACUITY major bleeding, the discriminatory power of update model improved substantially when congestive heart failure (CHF), prior vascular disease as well as body mass index (BMI) were considered. The update model demonstrated good calibration and C-statistic of 0.747 and 0.745 with no white blood cell (WBC) count. For BARC 3 or 5 bleeding, good calibration and discriminatory capacity could be observed when CHF and prior vascular disease were added in the update models, with an excellent C-statistic of 0.838 and a lower C-statistic value of 0.835 was obtained in the absence of WBC count. Conclusions The ACUITY/HORIZON score was successfully validated in contemporary predictive and risk-adjustment models for PCI-related bleeding. The update models had good operating characteristics in patients from a real-world ACS population irrespective of bleeding definitions.