Systemic Embolic Events in Atrial Fibrillation: An Individual Patient Data Meta-analysis of 71 683 Participants Randomized to NOAC Versus Warfarin

BACKGROUND: Systemic embolic events (SEEs) are a serious but underrecognized complication of atrial fibrillation. Although non–vitamin K antagonist oral anticoagulants prevent ischemic stroke (IS), their efficacy in SEE and the clinical characteristics of patients who experience SEE remain poorly understood. METHODS: We analyzed individual patient data from 4 pivotal randomized trials enrolling patients between 2005 and 2010 comparing non–vitamin K antagonist oral anticoagulants versus warfarin in atrial fibrillation. We characterized the incidence, clinical features, management, and outcomes of clinically overt SEE and compared results in these patients with patients who had an IS. RESULTS: Among 71 683 patients, 188 experienced SEE (26 with concurrent IS), yielding an annualized event rate of 0.13% per patient-year, compared with 1.25% per patient-year for IS (n=1797). Among 171 patients with SEE as their first event, median age was 75 years (interquartile range, 68–80), 49.7% were female, and mean±SD CHA 2 DS 2 -VASc score was 4. 7 ±1.5. Compared with IS, patients with SEE had higher rates of peripheral arterial disease (PAD, 16.5% versus 5.4%; P <0.001), previous myocardial infarction (24% versus 17%; P =0.02), previous vitamin K antagonist exposure (57% versus 46%; P =0.007), worse renal function (median creatinine clearance 58 versus 62 mL/min; P =0.02), and higher incidence of nonparoxysmal atrial fibrillation (86% versus 80%; P =0.047). Interventions (surgical or percutaneous) were performed in 62 patients (31%) with SEE. Standard-dose non–vitamin K antagonist oral anticoagulants reduced the risk of SEE by 29% compared with warfarin over a median follow-up of 25.2 months (interquartile range, 17.5–32.0; hazard ratio, 0.71 [95% CI, 0.51–0.99]; P =0.04). Thirty-day mortality after SEE was similar to IS (18% versus 17%), and SEE was associated with a nearly 3-fold increased risk of long-term mortality compared with patients without SEE or IS (hazard ratio, 2.85 [95% CI, 2.11–3.85]). Independent predictors of SEE included peripheral artery disease, smoking, nonparoxysmal atrial fibrillation, female sex, previous myocardial infarction, previous stroke or transient ischemic attack, vitamin K antagonist experience, and renal dysfunction. CONCLUSIONS: In this large individual patient data meta-analysis, non–vitamin K antagonist oral anticoagulants significantly reduced the risk of SEE compared with warfarin. Although SEEs were approximately one-tenth as frequent as IS, they were associated with comparable mortality and substantial morbidity.