Background: Abnormal conduction, structure, and function of the atrial myocardium predispose to atrial fibrillation (AF) and stroke. The usefulness of electrocardiographic indices in predicting stroke or systemic embolism (SSE) in patients undergoing cardioversion (CV) for AF remains unknown, especially in those at low estimated risk.

Objective: We systematically evaluated the performance of various P-wave abnormalities (PWAs) in predicting SSE 30 days after CV (derivation cohort) and in the long term (validation cohort).

Methods: Electrocardiograms (n = 1773) of AF patients undergoing an acute CV were manually reviewed. The 30-day post-CV data were used to derive a composite PWA variable. The electrocardiographic findings were validated by the long-term follow-up of patients with no anticoagulation. Electrocardiograms of 27 CAREBANK study patients with right atrial appendage biopsies were further analyzed for histopathologic validation.

Results: During data derivation, the best performance was found with a combination of prolonged P-wave (≥180 ms), deflected P-wave morphology in lead II, biphasic P-waves in inferior leads, or increased P-terminal force (≥80 mm·ms) as markers for extensive PWA. In the validation cohort, 219 of 874 (25.1%) had extensive PWA. During a median follow-up of 4.9 years, there were 51 patients (5.8%) with SSE in total. In a competing risk model, PWA predicted SSE (adjusted hazard ratio, 2.1 per category; 95% CI, 1.4-3.1; P < .001). Areas under the curve for SSE at 3 years were 0.77, 0.79, and 0.86 for PWA, CHA₂DS₂-VASc, score, and their combination, respectively. On histologic evaluation, extensive PWA was associated with interstitial fibrosis (P = .033).

Conclusion: Novel electrocardiographic PWA classification provided additional prognostic insight in AF patients.