Background: Although warfarin is extremely effective in reducing the risk of ischemic stroke from atrial fibrillation (AF), it also increases the risk for significant hemorrhage. We assessed rates, predictors, and short-term mortality after hospitalization for major hemorrhage in a cohort of AF patients on and off warfarin.
Methods: We prospectively studied 13,559 adults with AF. Patients hospitalized for hemorrhage were identified from clinical databases and confirmed using chart review. Major hemorrhage was defined as fatal, transfusion of ≥ 2 units packed blood cells, or into a critical anatomic site; separate analyses were performed for intracranial hemorrhage (ICH) and extracranial hemorrhage. We used multivariable log-linear regression to assess for predictors of hemorrhage, adjusting for demographic and clinical characteristics.
Results: Out of 30,994 person-years of follow-up, we identified 78 ICHs in patients on warfarin and 51 off warfarin. Out of 30,830 person-years, there were 98 major extracranial hemorrhages on warfarin and 116 off warfarin. Although rates of hemorrhage increased with age, absolute rates of hemorrhage in anticoagulated patients were similar to those in non-anticoagulated patients. In patients aged 80 and older, the annual rate of ICH was 0.8% on warfarin and 0.7% off warfarin; for extracranial hemorrhage, the rate was 0.7% on warfarin and 1.0% off warfarin. In patients who developed ICH while taking warfarin, independent risk factors included age ≥ 80 (OR 2.8 [1.1-7.2] compared to age <60) and hypertension (OR 1.7 [1.0-2.8]).
Predictors of ICH off warfarin were age ≥ 80 (OR 6.2 [1.7-23]) and prior stroke (OR 2.4 [1.1-5.0]). Independent predictors of extracranial hemorrhage were age ≥ 80 (OR 3.8 [1.1-13] on warfarin and 3.8 [1.7-8.6] off warfarin) and prior gastrointestinal hemorrhage (OR 5.3 [2.9-9.8] on warfarin and 2.0 [1.1-3.6] off warfarin). INR ≥ 4.0 (compared to INR<4.0) was the strongest predictor of hemorrhage in patients taking warfarin: OR 15 [8.1-29] for ICH and OR 18 [10-30] for extracranial hemorrhage. Of the 48 deaths, 81% were from ICH, with a 30-day mortality rate of 50% after ICH on warfarin and 28% off warfarin. In contrast, 30-day mortality after major extracranial hemorrhage was relatively low (5% for patients on warfarin and 10% off warfarin).
Conclusions: In this largest prospective study to date of individuals with AF, absolute rates of major hemorrhage were low and similar in patients on and off warfarin, even in the most elderly. Hemorrhage contributed to few short-term deaths and the majority of deaths resulted from ICH, not extracranial hemorrhage. The benefits of warfarin in reducing ischemic stroke continue to outweigh the risk of hemorrhage, especially in older patients.