Objective: Automatic analysis of the frequency content of the fibrillatory baseline on the surface ECG accurately reflects the average rate of atrial fibrillation (AF). This frequency measurement correlates with the behavior of AF and predicts the response to administration of ibutilide, a new antiarrhythmic drug. Neither the temporal pattern of fibrillatory frequency in spontaneous paroxysmal or persistent AF, nor its response to chronic antiarrhythmic medication has been studied so far. Methods and Results: Holter ECG recordings were made in 20 patients during AF. One minute ECG segments were selected for analysis. The frequency content of the fibrillatory baseline was then quantified using digital signal processing. After high-pass filtering, the QRST complexes were subtracted using a template matching algorithm. The resulting fibrillatory baseline signal was subjected to Fourier transformation, displayed as a frequency power spectrum and the peak frequency (f) was determined. In 11 patients (7 male, 4 female, age 62 +/- 10 years) 31 paroxysmal AF episodes were analyzed. Duration ranged from 1 min to 665 min (115 +/- 175 min). Initial mean peak f measured 5.1 +/- 0.7 Hz (range 3.9 to 6.9 Hz). There was a positive correlation between f and AF duration (R = 0.53, p = 0.002). AF of less than 15 min duration (n = 13) showed a lower f (4.8 +/- 0.6 Hz) when compared with longer lasting episodes (n = 18, 5.3 +/- 0.7 Hz, p = 0.03). In short AF episodes f was constant, whereas in longer-lasting episodes f increased to 5.8 +/- 0.5 Hz (p < 0.001) within 5 min. In 9 patients (9 male, age 58 +/- 8 years) with persistent AF oral antiarrhythmic drugs (amiodarone n = 5, sotalol n = 3, flecainide n = 1) were given prior to electrical cardioversion for prophylaxis of AF recurrence. Frequency measurements were obtained at baseline and 3 to 5 days after initiation of drug administration. At baseline mean f measured 6.9 +/- 0.4 Hz. Frequency was reduced by antiarrhythmic drugs to 5.8 +/- 0.4 Hz (p < 0.001). Conclusions: (1) The duration of paroxysmal AF episodes can be predicted using spectral analysis of ECG recordings of AF episodes. (2) An increase in fibrillatory frequency is associated with AF persistence. (3) This technique can be used to monitor the response to antiarrhythmic medication. (C) 1999 Elsevier Science B.V. All rights reserved.
