Wavelet Analysis of Signal-Averaged Electrocardiograms.

Abstract

Listen

Translate article

Background:Spectral techniques for the analysis of signal-averaged electrocardiogram (ECG) are superior to the standard time-domain analysis in special circumstances but none of these techniques is better than the time-domain analysis in the prospective risk stratification of survivors of myocardial infarction (Ml). Aim:This study applied wavelet decomposition of signal-averaged ECGs to prospective risk stratification of Ml survivors and compared its performance with that of conventional time-domain analysis. Methods:Eligible patients were aged 75 years without atrial fibrillation, important noncardiac disease, bundle branch block, ventricular preexcitation, permanent pacemaker, or a history of cardiac surgery. Of 754 eligible patients, signal-averaged ECG recordings were available in 551. During a 2-year follow-up, 32 patients suffered from cardiac death (CM), 21 potentially arrhythmic death, 19 sudden cardiac death (SCO), 19 developed nonfatal sustained ventricular tachycardia (VT), and 5 were resuscitated from an ECG-documented ventricular fibrillation. Signal-averaged ECG indices from both analyses were compared in patients with and without different follow-up events. The positive predictive characteristics (dependency of positive predictive accuracy on sensitivity) for predicting different follow-up events based on both modes of analysis were compared at selected levels of sensitivity. Results: Time-domain signal-averaged ECGs were strongly correlated with incidence of VT (P = 0.01). Positive wavelet analysis was more strongly correlated with this event (P < 0.005) and with cardiac mortality (P < 0.05). For all events, wavelet analysis gave higher positive predictive accuracy (PPA) than the time-domain analysis, e.g., the prediction of SCD at sensitivity of 25%, wavelet and time-domain analyses gave PPA of 20.0% and 6.5%, respectively (P < 10-10). Conclusion:Wavelet decomposition analysis of signal-averaged ECGs is superior to the standard time-domain analysis in predicting post-MI events. In particular, this analysis identifies not only those post-MI patients who are at high risk of nonfatal sustained VT but also those who are at risk of SCD. A.N.E. 2000;5(1):20–29