Optimal measurements for detection of coronary artery disease by exercise electrocardiography

Abstract

Exercise electrocardiograms were analysed by digital computer in order to select criteria for optimal separation of normal subjects and patients with coronary artery disease. The results of a training group of 86 ostensibly healthy men and 52 patients were verified in a test series of 43 patients and 43 controls. All subjects had a normal QRS complex and ST-T segment at rest. ECG measurements included ST amplitudes and slopes at fixed intervals after the end of QRS, time-normalized ST amplitudes, the negative ST areas, and Chebyshev waveform vectors. Visual reading of the ECG according to the classical criterion of a 0.1 mV horizontal ST depression yielded a sensitivity of 50 and 51% and a specificity of 94 and 95% in the training group and the test group, respectively. Best results with a single ST measurement were obtained by the ST60 amplitude in lead X which was adjusted for heart rate according to the gradual ECG changes during exercise in the normal subjects. The sensitivity was 81% and the specificity 93%. In the test group these figures were 70 and 93%. Combination of measurements in linear discriminant functions resulted in some further improvement of these figures.