Prolonged atrial conduction time and inhomogeneous electrical atrial activity have been known to be important electrophysiologic characteristics in patients with paroxysmal atrial fibrillation (PAF). Filtered P wave duration dispersion (Pdisp) is defined as the difference between the maximal and minimal filtered P wave duration of the three X, Y, Z leads of P wave signal averaged ECG (PSAECG). The purpose of this study was to determine and compare conventional PSAECG parameters with this newly developed parameter, Pdisp, for the prediction of PAF risk.

The study population, consisting of 65 patients with PAF (Group 1) and 56 age and sex matched controls (Group 2), was further subgrouped by the presence (Group A) and absence (Group B)of organic heart disease. The PSAECG was recorded using P wave triggered SAECG and the following parameters were measured: total filtered P wave duration (TFPD), vector magnitude, RMS10, RMS20, RMS30 and Pdisp. These parameter values were compared between groups.

TFPD and Pdisp were significantly longer in the PAF group than in the control group (121.2±9.7 msec versus 108.5±12.7 msec, p=0.005:25.2±11.5 msec versus 14.7±7.9 msec, p<0.001, respectively). The presence of organic heart disease and the size of the left atrium did not affect the result. The designation of Pdisp greater than 15 msec as the cut-off value for predicting PAF produced a sensitivity of 85% and a specificity of 75%, and was more accurate than TFPD (sensitivity 72%, specificity 66%).

Pdisp might be a more accurate SAECG parameter for predicting the PAF risk than other conventional parameters.