Objectives We sought to critically analyze the routine use of conventional coronary angiography (CCA) before noncoronary cardiac surgery and to assess clinical prediction models that might allow more selective use of CCA in this setting. Methods We studied 5463 patients undergoing aortic valve surgery, mitral valve surgery, or septal myectomy with or without coronary artery bypass grafting from 2001 to 2010. Preoperative CCAs were evaluated for the presence of significant coronary artery disease (CAD). Random forests and logistic regression methods were used to determine the predictors of significant (≥50%) coronary stenosis. Results Preoperative CCA was performed in 4711 patients (86%). Two thirds of those with angina, previous myocardial infarction, or percutaneous coronary intervention had significant CAD found on CCA, versus one third of patients free of these risk factors (P <.001). Among 3019 patients without angina, previous myocardial infarction or percutaneous coronary intervention, older age, male gender, diabetes, and peripheral vascular disease independently predicted significant CAD (P <.001 for all; C-index = 0.74). Specifically, a multivariate model with these variables identified 10% (301 of 3019) of patients as having a low (≤10%) probability of coronary stenosis, of whom fewer than 5% had significant CAD and fewer than 1% had left main or triple-vessel coronary disease. Conclusions In the absence of angina, previous myocardial infarction, or percutaneous coronary intervention, preoperative CCA identified significant CAD in only one third of patients. Our clinical prediction models could enhance the identification of patients at low risk of significant CAD for whom CCA might potentially be avoided before cardiac surgery. This strategy may improve the efficiency of cardiac surgical care delivery by diminishing procedure-related morbidity and offering significant cost savings.
ASJC Scopus subject areas
- Pulmonary and Respiratory Medicine
- Cardiology and Cardiovascular Medicine