[Value of the computerized analysis of ST segment depression during exercise without myocardial infarction. Apropos of 807 cases].

The diagnostic value of ST segment changes on exercise were reassessed by computerised analysis in 807 patients without myocardial infarction who underwent coronary angiography. All the stress tests were carried out according to Bruce's protocol with a system of continuous averaging of the ST segment in V5, V2 and VF. An abnormal response was defined by the association of the following three criteria: 1) ST depression less than or equal to 1 mm, 2) the algebraic sum of the depression + ST slope less than or equal to -1, 3) changes occurring during exercise or the first minute of recovery. A significant coronary lesion was defined as at least 50 per cent narrowing of the vessel lumen. In the study population the prevalence of lesions was 55 per cent in men and 18 per cent in women. The sensitivity of exercise stress testing was 69 per cent but the specificity was only 65 per cent. The positive predictive value was 70 per cent in men, 29 per cent in women; the negative predictive value was 90 per cent in women compared with 62 per cent in men. The predictive values depended on the interpretation of the amplitude, morphology and topography of the ST depression. The low sensitivity and specificity were independent of the coronary angiographic criteria and not related to the bias usually encountered in the correlation between stress testing and coronary angiography. These results show that the quantitative analysis of ST changes during computerised stress testing is not sufficiently accurate in itself to detect atherosclerotic coronary artery disease.