Incidence and Predictors of Extracranial Bleeding on Oral Anticoagulants for Stroke Prevention in Patients With Atrial Fibrillation: A COMBINE-AF Analysis
Deborah M. Siegal, Marc Carrier, M. Cecilia Bahit, Jackie Bosch, Alexander P. Benz, Brian Bergmark, Robert P. Giugliano, Christopher B. Granger, Victorien Monguillon, Martin O’Donnell, Jonas Oldgren, Manesh R. Patel, Rohin K. Reddy, Lars Wallentin, Qilong Yi, Andre Zimerman, John EikelboomBACKGROUND:
Extracranial bleeding is the most common complication of oral anticoagulant (OAC) therapy for atrial fibrillation (AF), but its clinical importance for patients may be underrecognized. We sought to characterize extracranial bleeding events according to standardized severity definitions, identify baseline risk factors for bleeding, and quantify their population attributable fraction in patients with AF receiving OACs.
METHODS:
We analyzed patients receiving OACs from 5 pivotal randomized trials testing a direct OAC or warfarin in patients with AF (COMBINE-AF [A Collaboration Between Multiple Institutions to Better Investigate Non-Vitamin K Antagonist Oral Anticoagulant Use in Atrial Fibrillation]). The primary outcome was extracranial clinically relevant bleeding, defined as a first episode of extracranial major or clinically relevant nonmajor bleeding according to International Society on Thrombosis and Haemostasis criteria. The Kaplan-Meier method was used to calculate the cumulative incidence of bleeding by category. Multivariable Cox regression models were used to estimate adjusted hazard ratios (HRs) with 95% CI. Logistic regression models were used to calculate average population attributable fraction with 95% CI.
RESULTS:
Of 73 737 patients treated with OACs, 10 634 experienced clinically relevant extracranial bleeding over a mean follow-up of 705 days (cumulative incidence, 26% [95% CI, 18%–35%]; 7.6 per 100 person-years). This included 3188 major bleeds (cumulative incidence, 7% [95% CI, 6%–7%]; 2.1 per 100 person-years) and 7446 clinically relevant nonmajor bleeds (cumulative incidence, 19% [95% CI, 12%–28%]; 5.2 per 100 person-years). The distribution of bleeding sites differed by severity, with gastrointestinal bleeds comprising 26% of clinically relevant bleeds, 49% of major bleeds, and 15% of clinically relevant nonmajor bleeds. Risk factors for extracranial bleeding were consistent across severity bleeding categories, and baseline covariates in our multivariable models accounted for 66% to 69% of the population attributable bleeding risk.
CONCLUSIONS:
Extracranial clinically relevant bleeding is common among patients with AF treated with OACs and may more accurately reflect the overall burden of bleeding than major bleeding alone. Our models explained about two-thirds of the average population attributable risk, suggesting that additional unmeasured or unknown factors contribute to bleeding risk.