Power law analysis of the signal-averaged electrocardiogram for identification of patients with ventricular tachycardia: effect of bundle branch block.

Signal-averaged ECGs that use time-domain analysis are useful for the identification of patients at risk for ventricular tachycardia (VT). Bundle branch block (BBB) and other conduction defects reduce the value of this approach, but frequency-domain analysis has shown promise in such patients. The purpose of the present study was to examine a new frequency-domain approach to signal-averaged ECGs in patients with and without BBB: power law scaling (PLS). PLS was performed by plotting the power spectrum of the entire signal-averaged ECG on a plot of log power versus log frequency and determining the slope (beta) by least-squares regression. This method was studied in 346 patients. Results of discriminant analysis revealed better sensitivity, specificity, positive predictive value, negative predictive value, and percentage correctly predicted when this method was compared with time-domain indexes. A large proportion of the variance in PLS (19%) was found to be due to findings in patients with VT; whereas the best time-domain index, duration of the filtered QRS signal, explained only 6% of the variance in the group with VT. Mean levels of PLS (+/- standard deviation) were decreased for the group with VT (-3.55 +/- 0.95) as compared with the group without VT (-4.34 +/- 0.59; p < 0.001), suggesting a decrease in the time correlation of the signal. Thus this method of frequency-domain analysis of the signal-averaged ECG was useful in identifying patients with sustained VT despite the presence of significant conduction defects.(ABSTRACT TRUNCATED AT 250 WORDS)