Signal averaged P wave compared with standard electrocardiography or echocardiography for prediction of atrial fibrillation after coronary bypass grafting

Objective-To define the clinical value of the signal averaged P wave (SAPW) and to compare it with the standard electrocardiogram (EGG), echocardiogram, and clinical assessment for the prediction of atrial fibrillation after coronary bypass grafting(CABG). Design-Prospective validation cohort study. Setting-Regional cardiothoracic centre. Patients-201 unselected patients undergoing first elective CABG were recruited over six months. Patients requiring concomitant valve surgery were excluded. Main outcome measures-Age, sex, cardiothoracic ratio, and cardioactive drugs were noted. P wave specific SAPW recordings, EGG, and M mode echocardiograms from which left atrial diameter was measured were performed within 24 hours of surgery. Filtered P wave duration (SAPWD), spatial velocity, and energy were calculated from the SAPW. From the EGG, lead II P wave duration, P terminal force in lead V1, total P wave duration, and isoelectric interval were measured. Patients had Holter monitoring for 48 hours postoperatively and daily ECGs until discharge. Results-Two patients died (1%) and 10 were unsuitable for analysis (5%). Of the remaining 189, 51 (27%) had atrial fibrillation (AF) lasting > 1 hour at a mean of 2 (0.5 to 7) days after CABG. Of the variables examined, only SAPWD (AF group 148 (SD 12), v 142 (14) ms, P = 0.008) and male sex (AF group 96%, v 78%, P < 0.01) were significantly different. A prospectively defined SAPWD of > 141ms predicted atrial fibrillation with positive and negative predictive accuracies of 34% and 83%. Logistic regression analysis identified both male sex and SAPWD as significant independent predictors of postoperative atrial fibrillation. Conclusions-Signal averaged P wave duration was a better predictor of atrial fibrillation after coronary bypass grafting than standard electrocardiographic or echocardiographic criteria. The predictive value of this test is such that it is likely to be useful in the design of prospective trials of prophylactic antiarrhythmic treatment but is of limited use using current techniques in the clinical management of individual patients.