Predictive model for late atrial arrhythmia after closure of an atrial septal defect

Objectives: To develop a quantitative event-free prediction model of late atrial arrhythmia after atrial septal defect (ASD) repair. Background: The clinical management of ASD is driven by risk factors that determine the occurrence of late atrial arrhythmia. Methods: Data from ASD type secundum patients, included in the Belgian Congenital Heart Disease Registry, were analyzed. Based on review of the literature, age at repair, gender, pulmonary hypertension, atrial arrhythmia before and within one month after repair were included in the model. Using Cox-regression analysis, a weighted risk score was derived, which was validated using the Brier score. Results: A total of 155 patients (117 women; median age at follow-up 53.9 years, range 18.0-78.8) having 349 follow-up years was included. Thirty-nine patients (25.2%) presented with late atrial arrhythmia. Multivariate analysis showed that a mPAP ≥ 25 mmHg (HR 4.39; 95%CI 2.17-9.09; P < 0.0001), the presence of atrial arrhythmia before (HR 3.52; 95%CI 1.75-7.14; P = 0.002) and ≤ 1 month after repair (HR 6.62; 95%CI 2.38-20.00; P < 0.0001) and gender (HR 2.18 95%CI 1.11-4.35) were associated with late atrial arrhythmia. A risk score (0 to 28 points) to predict atrial arrhythmia free survival was derived for follow-up times ranging from one to 5 years. Mean Brier score for the model was 0.10. Conclusions: We formulated a well validated risk model to predict arrhythmia-free survival in ASD patients undergoing ASD repair. Further research is needed whether this model can be used for individual clinical risk stratification and whether the model can be adapted for application in other congenital heart defects.