Prediction model for the new onset of atrial fibrillation combining features of 24-hour Holter electrocardiogram with 12-lead electrocardiogram

Background: Several factors that predict new-onset atrial fibrillation (AF) have been investigated using the 24-hour Holter electrocardiogram (ECG) and 12-lead ECG; however, these have been based on each test independently. The aim of this study was to combine findings from the two tests to create a comprehensive, easy-to-use score and to examine its validity. Methods and Results: A total of 502 patients underwent 24-hour Holter ECG and 12-lead ECG were followed up for 6.2 ± 3.5 years, and 66 patients developed new-onset AF. Multivariate Cox regression analyses revealed that total number of supraventricular extrasystoles (SVEs) ≥ 100 beats/day and SVE's longest run ≥ 3 beats on 24-hour Holter ECG and PR interval ≥ 185 ms, amplitude ratio of P wave (aVR/V1) < 1.0 and amplitude of RV5 + SV1 ≥ 2.2 mV on 12-lead ECG were significant independent predictors for developing AF (all p < 0.01). Using these cut-off points, the PAAFS (acronym for risk factors) score was constructed by adding one point for each parameter if the patient met each of the criteria. The area under the curve (AUC) of the PAAFS score was 0.80, compared to the AUCs of 24-hour Holter ECG-only factors (0.73) and 12-lead ECG-only factors (0.72), indicating an improvement in score. The annual incidence of AF for each PAAFS score were 0.0%, 0.2%, 0.7%, 1.9%, 5.6%, and 11.1%/year for scores 0 to 5, respectively. Conclusion: The PAAFS score, which combines findings from 24-hour Holter ECG and 12-lead ECG, was superior to 24-hour Holter ECG and 12-lead ECG alone in predictive accuracy for new-onset AF.