A Predictive Model for Recurrence of Atrial Fibrillation Based on P-Wave Duration in Patients with Early Persistent Atrial Fibrillation Who Underwent Radiofrequency Catheter Ablation

Abstract

Purpose: To construct a predictive model for the recurrence of atrial fibrillation (AF) based on P-wave duration (PWD) in patients with early persistent AF (PeAF) who underwent radiofrequency catheter ablation (RFCA), with the aim of helping clinicians accurately adjust clinical strategies. Methods: Data from patients with early PeAF, who were admitted to the Department of Cardiology at the authors' hospital were collected. Based on predefined inclusion and exclusion criteria, only those who successfully underwent RFCA for the first time were included in the analysis. Pre- and postoperative clinical, echocardiographic, and electrocardiographic data were collected and recorded. Multivariate logistic regression was used to construct a predictive model for AF recurrence based on PWD. The predictive efficacy of each continuous variable and the predictive model were compared using the area under the receiver operating characteristic (ROC) curve. The corresponding nomogram for the predictive model was constructed. Interaction tests were performed to evaluate the predictive efficacy of the model for AF recurrence. Results: A total of 237 patients were enrolled and divided into two groups: recurrence (n = 59); and sinus rhythm (n = 178). PWD was greater and left atrial appendage emptying velocity (LAAV) was lower in the recurrence group; these differences were statistically significant (p <0.001). The ROC curve for univariate prediction of AF recurrence revealed that the area under the ROC curve (AUC) for PWD and LAAV were 0.7912 and 0.7713, respectively, which were greater than those of other continuous variables. Compared with PWD alone, the multivariate predictive model containing PWD, left ventricular ejection fraction, and LAAV demonstrated no statistically significant difference in AUC (p = 0.0553) but improved the prediction efficiency of correctly reclassifying recurrence rates, net reclassification improvement 14.13% (95% confidence interval: 0.19–28.07%; p = 0.0469). The interaction effect did not significantly alter the effectiveness of the predictive models. Conclusions: The multivariate model based on PWD measured after RFCA demonstrated better predictive efficacy than the univariate model in patients with early PeAF. These results may contribute to evidence supporting the formulation of personalised treatments for patients with AF.