Development and validation of a simple model to predict severe coronary artery disease after myocardial infarction: Potential impact on cardiac catheterization use in the United States and Canada

Background: Improved patient selection may optimize the efficiency of cardiac catheterization in both high- and low-rate regions. The purpose of this study was to develop and validate a clinical model for predicting high-risk coronary artery disease (CAD) after myocardial infarction (MI) and to examine the model's potential impact on the use rate of both US and Canadian catheterization practices. Methods and Results: By the use of baseline clinical variables from 1122 patients in the angiographic substudy of the Global Use of Strategies to Open Occluded Arteries in Acute Coronary Syndromes (GUSTO-1) trial, we developed a model that was predictive of severe CAD (left main or triple-vessel disease). The final model, which included prior MI, age, sex, hyperlipidemia, and decreased left ventricular ejection fraction (C-index = 0.70), was externally validated in 781 patients in the GUSTO IIb trial. Although the probability of severe CAD predicted 5-year survival, the frequency of catheterization in both Canada and the United States bore no relationship to severe CAD risk in the GUSTO-1 trial. By use of the model, we estimated that as much as 15% of US catheterizations from both GUSTO-1 and GUSTO IIb might have been avoided, without significantly compromising the number of patients with severe CAD who were identified (sensitivity = 0.94). By applying the model to Canadian practices, an additional 30 cases of severe CAD might have been identified per every 1000 catheterizations performed, without increasing the catheterization rate. Conclusions: The likelihood of severe CAD after ST-elevation MI may be predicted from simple baseline clinical variables. The use of a severe CAD predictive model to guide patient selection might enhance the cost-effectiveness of both aggressive and conservative catheterization practices.