Alternative diagnostic strategies for coronary artery disease in women: demonstration of the usefulness and efficiency of probability analysis

Alternative strategies using conditional probability analysis for the diagnosis of coronary artery disease (CAD) were examined in 93 infarct-free women presenting with chest pain. Another group of 42 consecutive female patients was prospectively analyzed. For this latter group, the physician had access to the pretest and posttest probability of CAD before coronary angiography. These 135 women all underwent stress electrocardiographic, thallium scintigraphic, and coronary angiographic examination. The pretest and posttest probabilities of coronary disease were derived from a computerized Bayesian algorithm. Probability estimates were calculated by the four following hypothetical strategies: SO, in which history, including risk factors, was considered; S1, in which history and stress electrocardiographic results were considered; S2, in which history and stress electrocardiographic and stress thallium scintigraphic results were considered; and S3, in which history and stress electrocardiographic results were used, but in which stress scintigraphic results were considered only if the poststress probability of CAD was between 10% and 90%, i.e., if a sufficient level of diagnostic certainty could not be obtained with the electrocardiographic results alone. The strategies were compared with respect to accuracy with the coronary angiogram as the standard. For both groups of women, S2 and S3 were found to be the most accurate in predicting the presence or absence of coronary disease (p less than .05). However, it was found with use of S3 that more than one-third of the thallium scintigrams could have been avoided without loss of accuracy. It was also found that diagnostic catheterization performed to exclude CAD as a diagnosis could have been avoided in half of the patients without loss of accuracy.