A critical appraisal of quantitative spectro-temporal analysis of the signal-averaged ECG: predicting arrhythmic events after myocardial infarction.

The objective of this study was to determine if spectro-temporal analysis of the signal-averaged ECG (SAECG) predicts spontaneous sustained ventricular tachyarrhythmias and sudden death in patients prospectively followed after myocardial infarction (MI). A SAECG was recorded in 177 patients 9 +/- 5 days after MI. Spectro-temporal analysis of the SAECG involved incrementing a Hanning window every 3 ms beginning 20 mg before the end of the QRS complex and extending into the ST segment. Quantitative analysis was performed using a cross-correlation function to create a normality factor. A normality factor < 0.3 was deemed abnormal. The SAECG was abnormal in 41% of patients using time-domain analysis and 44% of patients using spectro-temporal analysis. There was no correlation between an abnormal SAECG in the time domain and the frequency domain. Patients with inferior wall MI were more likely to have an abnormal spectro-temporal map (odds ratio 2.26, P < 0.05). Time-domain analysis of the SAECG (relative risk (RR) 2.6) was a statistically significant univariate predictor of arrhythmic events. Spectro-temporal analysis of the SAECG was only weakly (RR 1.8) and not significantly (P = 0.15) associated with the spontaneous occurrence of these arrhythmias. When both time-domain analysis and spectro-temporal analysis of the SAECG were abnormal, the relative risk for an arrhythmic event was increased by 3.3-fold. Quantitative spectro-temporal analysis of high frequency signals within the SAECG cannot by itself predict the occurrence of spontaneous ventricular arrhythmias in patients after MI.