Comparative accuracy of clinical tests for diagnosis and prognosis of coronary artery disease.

The discriminant accuracy of 14 variables derived from clinical evaluation, cardiac fluoroscopy, exercise electrocardiography, thallium scintigraphy and radionuclide angiography was assessed with respect to the diagnosis of angiographic coronary artery disease (CAD) among 607 patients undergoing coronary angiography, and with respect to the prognosis of subsequent cardiac death or nonfatal infarction among 4,104 patients followed for 1 year after testing. Discriminant accuracy (quantified in terms of the area under a receiver-operating characteristic curve for each variable) ranged from 50 to 73% for diagnosis, and from 54 to 77% for prognosis. Although there was a great deal of variability from test to test, variables representing direct or indirect manifestations of myocardial ischemia tended to correlate better with prognosis. Thus, variables derived from radionuclide angiography and thallium scintigraphy tended to have more prognostic accuracy than diagnostic accuracy, while fluoroscopy tended to have more diagnostic accuracy than prognostic accuracy. The pattern associated with clinical evaluation and exercise electrocardiography was less discernable. Accordingly, the accuracy of noninvasive tests with respect to diagnosis and prognosis of CAD should be separately determined based on individual empiric observation.