Frequency Analysis of the P Wave: Comparative Techniques

Abstract

Frequency domain analysis of the signal-averaged P wave (SAPW) may provide additional information over time domain analysis in patients with paroxysmal atrial fibrillation (PAF), but an optimum method has not yet been defined. We compared two different approaches using SAPW and test signals. We analyzed the frequency spectrum of the entire P wave (method A) or of only its terminal 100 msec (method B) using SAPW from 24 patients with idiopathic PAF and 34 normal controls. Absolute powers in frequency bands above 20, 30, 40, 60, and 80Hz (P20, P30, P40, P60, and P80) and power ratios about these frequencies (PR20, PR30, PR40, PR60, and PR80) were calculated. Patients had greater P30, P60, and P80 using method A (P30: 28.6 [+/- 2.2] vs 22.8 [+/- 1.6] microV2.s; P = 0.04; P60: 4.7 [+/- 0.5] vs 3.4 [+/- 0.3] microV2.s; P = 0.03; P80: 1.9 [+/- 0.2] vs 1.3 [+/- 0.09] microV2.s; P = 0.01) but differences were smaller with method B. Moving the P wave endpoint 10 msec into the P wave had no effect on frequency domain parameters using method A but produced important changes using method B. Both methods were also applied to test signals of different durations but equal powers and power ratios. Method A gave predicted values for power and PR40 at all signal durations, but method B produced variable results. Frequency domain analysis of the SAPW shows significant increases in P wave energy in patients with PAF compared to controls. Analysis of the entire P wave is influenced less by signal duration and variations of the P wave endpoint than analysis of the terminal P wave alone and may define patient and control populations more precisely.