A Validated Model to Identify Patients With Low Likelihood of High-Risk Coronary Artery Disease Anatomy

Abstract

In stable coronary artery disease (CAD), revascularization improves outcomes only for patients with high-risk coronary anatomy (HRCA). We sought to derive and validate a prediction model, incorporating clinical and exercise stress test characteristics, to identify patients with HRCA. We conducted a retrospective analysis of patients undergoing exercise stress testing at Cleveland Clinic (2005 to 2014), followed by invasive coronary angiography within 3 months. We excluded patients with acute coronary syndrome, known CAD or ejection fraction <50%. HRCA was defined as left main, 3-vessel, or 2-vessel disease involving the proximal left anterior descending artery. Clinical and stress test predictors of HRCA were identified in a multivariable logistic regression model, internally validated with 1,000-fold bootstrapping. The model was then externally validated at the University of Pittsburgh Medical Center (2017 to 2019). The model was derived from 2,758 patients with complete data. HRCA was identified in 418 patients (15.2%) in the derivation cohort. The model consisted of 10 variables: age, male gender, hypertension, hypercholesterolemia, diabetes mellitus, family history of premature CAD, high-density lipoprotein, chest pain, exercise time, and Duke Treadmill Score. Bias-corrected c-statistic was 0.79 (95% confidence interval 0.77 to 0.81) with excellent calibration. In all, 762 patients (27.6%) had a predicted probability and observed prevalence of HRCA <5%. In the validation cohort, the model had a c-statistic of 0.79 (95% confidence interval 0.74 to 0.85) and 210 patients had an observed prevalence of HRCA <5% (40%). In conclusion, an externally validated prediction model, based on clinical characteristics and exercise stress test variables, can identify stable patients with CAD who have HRCA.