Prediction of cardiovascular death in men undergoing noninvasive evaluation for coronary artery disease.

ABSTRACT

OBJECTIVE: To develop prediction rules from clinical and exercise test data identifying patients at high and low risk for cardiovascular events among a group of male veterans. DESIGN: Prognostic study with prospective gathering of data and routine follow-up of consecutive patients referred for exercise testing. Patients only underwent noninvasive evaluation for coronary artery disease. No validation cohort is yet available. SETTING: A 1200-bed Veterans Affairs Medical Center. PATIENTS Of 3609 men referred for exercise testing between 1984 and 1990, 2546 patients remained evaluable after exclusion of those who underwent subsequent cardiac catheterization, those with significant valvular heart disease, and those who had previous coronary artery bypass surgery. MEASUREMENTS Evaluation included recording of clinical data on a standardized form and a standard treadmill test followed by assessment of cardiovascular events. RESULTS: During a mean follow-up period (+/- SD) of 2.75 (+/- 18) years, 119 cardiovascular deaths and 44 nonfatal myocardial infarctions occurred in 2546 patients. The Cox proportional hazards model showed the following characteristics to be statistically independent predictors of time until cardiovascular death history of congestive heart failure or digoxin use, exercise-induced ST depression, change in systolic blood pressure during exercise, and exercise capacity. Using a simple score based on one item of clinical information (history of congestive heart failure or digoxin use) and three exercise test responses (ST depression, exercise capacity, and change in systolic blood pressure), 77% of patients were categorized as low risk (annual cardiac mortality rate, less than 2%), 18% as moderate risk (annual cardiac mortality rate, 7%), and 6% as high risk (annual cardiac mortality rate, 15%; hazard ratio, 10; 95% confidence interval, 6 to 17). This model has not yet been validated. CONCLUSIONS: Variables available from the usual non-invasive work-up of patients with known or suspected coronary artery disease can be used to predict future risk for cardiovascular death.