Coronary risk factors used to predict coronary artery disease by logistic regression analysis.

Risk factor analysis in coronary artery disease was conducted in 303 patients who underwent coronary arteriography to identify associations between personal characteristics and the prevalence of coronary heart disease. Age, sex, obesity, smoking, alcohol intake, hypertension, diabetes mellitus, serum uric acid, total cholesterol, LDL- and HDL-cholesterol, triglyceride, and atherogenic indices were statistically analyzed. All 13 variables were first compared between patients with positive and negative ergonovine tests. Only total cholesterol was significantly different, while significant differences in age, sex, history of diabetes, total cholesterol, LDL- and HDL-cholesterol, triglyceride and atherosclerotic indices were observed between patients with and without organic coronary artery stenosis. A multivariate analysis was performed, and the resulting equation was tested using the remaining patients. Logistic analysis of all 13 variables identified 5 (age, sex, diabetes mellitus, LDL- and HDL-cholesterol) which accounted for the differences between patients with and without significant coronary artery disease and that were validated in the test group. The sensitivity for prediction of coronary artery disease was 75.8%, specificity 68.5%, and predictive accuracy 71.5% in the test group. Thus, risk factor analysis appears to be very valuable in screening subjects with high-risk organic coronary stenosis and in optimizing the preventive and therapeutic modalities, but not in predicting vasospastic subjects.