A sequential approach to the diagnosis of coronary artery disease using multivariate analysis

There has been considerable interest in recent years in enhancing the accuracy of noninvasive tests in diagnosing coronary artery disease. The recognition that no currently available test is a perfect predictor has led to the use of probability analysis as a means of assessing the presence or absence of coronary disease. In this article we present a multivariate approach to the diagnosis of coronary disease. One hundred forty-seven patients undergoing coronary angiography, thallium-201 imaging, and exercise ECG were studied. Patients were classified according to age, sex, and typical vs atypical chest pain. Sequential stepwise logistic regression analysis was performed to develop probability statements prior to testing, after exercise ECG, and after exercise ECG and thallium-201. The results indicate that this sequential approach can be used to develop strategies for the diagnosis of coronary disease in the same way as Bayes' theorem, while permitting integration of multiple characteristics into one model.