Within- and between-patient variation of the signal-averaged P wave in coronary artery disease

Objectives: To estimate interobserver, within-patient and between-patient variation of the signal-averaged P wave. To determine whether demographic, clinical, conventional ECG information, and coronary angiographic data are associated with the signal-averaged P wave duration in patients with documented coronary artery disease. Background: A prolonged signal-averaged P wave may indicate the presence of a substrate for atrial tachyarrhythmias and may predict subsequent development of atrial fibrillation. However, information on variation, reproducibility, and determinants of the signal-averaged P wave are sparse. Methods: One hundred ninety-three patients with angiographically documented coronary artery disease underwent two consecutive procedures of signal-averaging of P waves (SAECG1 and SAECG2). Interobserver, within-patient, and between-patient variation of the signal-averaged P wave was estimated (coefficient of variation: SD/mean). Multiple linear regression analysis was applied to identify parameters independently associated with signal-averaged P wave duration ISA-P). Atrial late potentials were considered if SA-P > 140 ms, and logistic regression analysis was applied to identify parameters associated with the presence of atrial late potentials. Results: The interobserver, within-patient, and between-patient coefficients of variation for the signal-averaged P wave duration were 7.5%, 6.0%, and 8.4%, respectively. The:signal-averaged P wave duration correlated significantly with standard ECG P wave duration and height of the patient (r = 0.59). Forty-nine percent of the patients had atrial late potentials. P wave duration in the standard ECG correctly classified 73% (140/188) of the patients with respect to atrial late potential positivity or negativity. The sensitivity was 67% and the specificity was 78%. Agreement on the presence or absence of atrial late potentials between two observers was present in 71% (136/193) of the patients, and in 78% (151/193) between SAECG1 and SAECG2. Conclusions: The signal-averaged P wave has limited reproducibility in patients with coronary artery disease, and a normal or abnormal signal-averaged P wave can be predicted from the conventional ECG with high accuracy. It is recommended that the signal-averaged P wave be compared with the standard ECG P wave duration in follow-up studies with the aim of predicting atrial fibrillation.