An Externally Validated Model for Predicting Long-Term Survival after Exercise Treadmill Testing in Patients with Suspected Coronary Artery Disease and a Normal Electrocardiogram

This study identified 33,268 patients with normal electrocardiograms undergoing treadmill stress testing, and attempted to develop a post-treadmill test prediction rule for adults with suspected coronary artery disease. Demographic, clinical, and stress test data were collected at a single institution on patients aged 30 years or older undergoing stress testing between September of 1990 and December of 2002. Analysis was focused on age, gender, smoking history, diabetes mellitus, hypertension, and history of typical angina. Clinical data obtained included proportion of predicted metabolic equivalents achieved, occurrence of ST-segment depression, test-induced angina, abnormal heart rate recovery, and frequent ventricular ectopy during recovery. The primary outcome measured was all-cause mortality. A predictive nomogram was developed and then externally validated at a separate institution. Sixty-four percent of patients that were identified as intermediate or high risk by the Duke Treadmill Score were reclassified by the nomogram as low risk. When compared to the Duke Score, 21% of all patients were reclassified using the nomogram. Subsequent external validation of the nomogram was performed at a separate institution in 5821 patients, where it was found to successfully identify the low-risk subset. The authors of this study concluded that their nomogram worked well for predicting mortality in patients with normal electrocardiograms and was more discriminatory than the Duke Treadmill score.