Time and frequency recurrence analysis of persistent atrial fibrillation after electrical cardioversion

Electrical cardioversion (ECV) has become a mainstay of therapy for the treatment of persistent atrial fibrillation (AF), which is an arrhythmia that affects up to 1% of the general population. The procedure is initially effective, but it is also characterized by a high rate of AF recurrence. As a consequence, it would be clinically useful to predict normal sinus rhythm (NSR) maintenance after ECV before it is attempted. In this respect, several clinical, echocardiographic and demographic parameters have been analyzed by other authors. However, these indices are weak predictors of ECV outcome. In this work, surface electrocardiographic (ECG) recordings were used to extract the atrial activity (AA) signal and parametrize the fibrillatory (f) waves, both in time and frequency, to obtain AF recurrence predictors. Parameters as f waves amplitude (fWA), AA mean power, dominant atrial frequency (DAF), its first harmonic, etc were studied. Obtained results showed that fWA was the most significant predictor of AF recurrence 1 month later. Concretely, 72.73% of the patients resulting in NSR, 83.87% relapsing to AF and 80.0% with unsuccessful ECV, were correctly identified. Therefore, fWA classified satisfactorily 79.37% of the analyzed patients. In addition, a forward stepwise discriminant analysis, with a leave-one-out cross validation approach, proved that fWA and DAF combination provided an improved diagnostic ability of 85.71%. In this case 86.36%, 83.87% and 90% of the patients who resulted in NSR, relapsed to AF and with unsuccessful ECV, were correctly discerned, respectively. In conclusion, fWA could be considered as a promising predictor of ECV outcome during the first month following the procedure. Additionally, time and frequency indices could yield complementary information useful to predict the cardioversion outcome. Finally, further studies are needed to validate the robustness of these parameters and the repeatability of the obtained results on wider databases.