Tree-structured risk stratification of in-hospital mortality after percutaneous coronary intervention for acute myocardial infarction: A report from the New York State percutaneous coronary intervention database

Abstract

Previous risk scores have shown excellent performance. However, the need for real-time risk score computation makes their implementation in an emergent situation challenging. A more simplified approach can provide practitioners with a practical bedside risk stratification tool. We developed an easy-to-use tree-structured risk stratification model for patients undergoing early percutaneous coronary intervention (PCI) for acute myocardial infarction. The model was developed on the New York State PCI database for 1999 to 2000 (consisting of 5385 procedures) and was validated using the subsequent 2001 to 2002 database (consisting of 7414 procedures). Tree-structured modeling identified 3 key presenting features: cardiogenic shock, congestive heart failure, and age. In the validation data set, this risk stratification model identified patient groups with in-hospital mortality ranging from 0.5% to 20.6%, more than a 20-fold increased risk. The performance of this model was similar to the Mayo Clinic Risk Score with a discriminative capacity of 82% (95% confidence interval, 79%-84%) versus 80% (95% confidence interval, 77%-82%), respectively. Patients undergoing PCI for acute myocardial infarction can be readily stratified into risk categories using the tree-structured model. This provides practicing cardiologists with an internally validated and easy-to-use scheme for in-hospital mortality risk stratification.