Effect of conduction defects on the signal-averaged electrocardiographic determination of late potentials

To determine the effect of cardiac conduction defects on the signal-averaged electrocardiogram (ECG) and on its ability to noninvasively identify patients predisposed to ventricular tachycardia (VT), standard 12-lead ECGs and signal-averaged ECGs were obtained in 213 patients with normal conduction and 186 patients with various conduction defects. Sustained VT was induced by programmed stimulation or occurred spontaneously in 122 patients. Two-way analysis of variance showed that conduction defects and VT were associated with changes in 3 signal-averaged ECG parameters: duration of the filtered QRS, duration of the terminal QRS under 40 microV and the mean amplitude of the terminal 40 ms of the QRS. Stepwise multiple logistic regression identified 3 variables that distinguished the patient with VT with a sensitivity of 62%, a specificity of 63% and a positive predictive accuracy of 63%. These 3 variables, listed in order of importance, were conduction defect score, duration of the filtered QRS and mean amplitude of the terminal 40 ms of the QRS. These data indicate that conduction defects have systematic effects on signal-averaged ECG parameters independent of those seen in patients predisposed to VT. These effects mandate the adjustment of the definitions of late potentials in the presence of conduction defects.