Diagnostic value of exercise electrocardiography and thallium myocardial scintigraphy in patients without previous myocardial infarction: a Bayesian approach.

The clinical value of combining exertional ECG and postexertional thallium (201T1) scintigraphy was assessed in 160 patients (130 men and 30 women) suspected of having coronary artery disease (CAD) who underwent a coronary arteriography. Based on sex and history, the patients were subdivided in two groups with different prevalences of CAD: Group 1 (high prevalence of CAD = 90%) included 98 men with typical angina pectoris (AP) and group 2 (low prevalence of CAD = 18%) included 32 men and 30 women with atypical AP. Compared with the exertional ECG, myocardial scintigraphy was more sensitive (87% vs 74%) and more specific (89% vs 70%) for the diagnosis of CAD. The combination of the ECG and scintigraphic data was useful if both tests gave concordant results: 100% of true positives (n = 67) and 84% of true negatives (n = 43). In case of discordant results (n = 50), no firm diagnostic conclusion could be made due to the many false-positive (27%) and false-negative (25%) scintigrams. These results are easier to interpret when the prevalence of CAD is taken into account. According to Bayes' theorem, abnormal exercise results confirm CAD when the prevalence is high and normal results rule out CAD when the prevalence is low; also, a normal response to exercise has no predictive value when the prevalence is high. When the prevalence is low, an abnormal ECG or thallium has low predictive value but concordant abnormal responses are highly predictive for CAD (100% of true positives).