New criteria for computer interpretation of exercise electrocardiograms in a largely asymptomatic population.

We developed new discriminant functions for analyzing treadmill ECGs from a largely asymptomatic population. Treadmill ECG data were gathered from two patient groups: 70 patients with coronary artery disease with occlusions greater than or equal to 30% by angiography, and 138 patients without coronary artery disease. The group without coronary artery disease consisted of 76 false positive responders to treadmill testing using standard ST segment criteria, 22 supraventricular tachycardia patients (both groups free of coronary artery disease by angiography), and 40 patients at very low risk for having coronary artery disease. ECG leads CC5, CM5, V5, Yh and Z were recorded before, during and after exercise protocol conditions. Computer-averaged ECGs were processed to provide Q, R, S and T-wave amplitudes, ST amplitudes and slope, and QS and RT intervals. Each patient provided over 100 variables per lead for analysis. Stepwise statistical procedures yielded lead-specific linear discriminant functions containing four to six variables/lead. Application of these functions provided sensitivity and specificity in the range 70-84%. When compared with other standard interpretive criteria, these results provided improved diagnostic accuracy for the largely asymptomatic population.