Cardiovascular risk prediction models in patients with stable coronary artery disease

Background: Installment of appropriate measures to prevent adverse events in patients with established, stable coronary artery disease (CAD) may contribute to efficient healthcare. This review gives an overview of existing models for prediction of cardiovascular adverse events in such patients and discusses model performance. Methods: We used a computerized literature search in the EMBASE, PubMed publisher, MEDLINE, Google Scolar, Web of Science and Cochrane databases. Studies were selected if they included patients with stable CAD (stable angina pectoris, myocardial infarction more than 3 months ago or coronary intervention more than 6 months ago) and if they presented a model that included mortality as the endpoint. Results: Sixteen studies met our inclusion criteria. Clinical variables that were included in the models differed highly between the studies. Still, age, smoking status, hypertension, diabetes, cholesterol and heart failure were present in a large part of the models. Several studies examined model discrimination, but the majority paid insufficient attention to calibration and validation. Conclusions: Although multiple prediction models for adverse events have been developed in patients with stable CAD, variables included in these models display large heterogeneity, and model performance is often insufficiently addressed.