TCT-343 Does The National Cardiovascular Data Registry Bleeding Risk Score Accurately Predict Bleeding In African American Patients Undergoing Percutaneous Coronary Intervention?

Abstract

Background: The National Cardiovascular Data Registry (NCDR) risk score was validated to predict post-percutaneous coronary intervention (PCI) bleeding. However, this risk prediction model was derived from a predominantly Caucasian population, with an under-represented African American (AA) population of 6%. Differential post-PCI bleeding rates have been reported. Objectives: We aim to analyze the post-PCI bleeding events in a better-represented AA population, and to determine if the NCDR bleeding risk model is improved by the addition of AA race to predict post-PCI bleeding. Methods: From our own PCI registry, bleeding (defined by NCDR CathPCI registry as transfusion, prolonged hospital stay and/or drop in hemoglobin >3g/dL) rates were compared between AA and non-AA patients. A logistic regression model using ROC curves was used to compare the accuracy of the NCDR risk score in predicting post-PCI bleed in AA and non-AA patients, and a net reclassification improvement (NRI) analysis was used to determine...