Combined time-domain and spectral turbulence analysis of the signal-averaged ECG improves its predictive accuracy in postinfarction patients.

The predictive accuracy of time-domain (TD) late potential analysis of the signal-averaged electrocardiogram in postmyocardial infarction (MI) is limited by the high incidence of false positives in inferior MI. However, frequency-domain spectral turbulence (ST) analysis suffers from a high incidence of false positives, especially in anterior MI. A prospective study was conducted of 262 patients with acute MI to investigate the hypothesis that combined TD and ST analyses of the signal-averaged electrocardiogram could improve its predictive accuracy for serious arrhythmic events in the post-MI period. Abnormal TD criteria were RMS40 less than 25 microV at 25 Hz plus RMS40 less than 16 microV at 40 Hz, and abnormal ST criteria were a turbulence score of 3 or 4. Seventeen patients had arrhythmic events during 10.5 +/- 2.4 months of follow-up evaluation (13 sudden cardiac death judged to be due to arrhythmia and 4 nonfatal sustained ventricular tachycardia). The total predictive accuracy of combined TD and ST (92%) was higher than TD (87%), whereas ST had the lowest total predictive accuracy of 78%. The negative predictive accuracy of all three analyses was high (96-97%). However, the positive predictive accuracy of TD (28%) was higher than ST (14%). Combined TD and ST significantly improved the positive predictive accuracy of the test to 35% in the total group and to 40% in patients with first anterior or inferior MI. The best results were obtained in patients with first anterior MI, where the positive predictive accuracy of combined analysis was 50%.