We recruited 973 patients (mean age, 81 years) with atrial fibrillation from the primary care setting and randomly assigned them to receive anticoagulation with warfarin or aspirin.
Atrial fibrillation is generally found in patients older than 75 years, with about 1 in 10 people in this age group having the disorder.1,2 These individuals, however, have been notably underrepresented in previous randomized controlled trials evaluating the efficacy of anticoagulation therapy for stroke prevention among patients with atrial fibrillation.3,4 This has led to concern about the applicability of the trial evidence to elderly patients, especially because it is well recognized that the risk of severe bleeding while taking anticoagulants increases with age.5
The principal alternatives to warfarin (Coumadin) are antiplatelet agents such as aspirin, which are convenient, if less effective, options.6 A meta-analysis of individual patient data from trials in which patients were randomly assigned to receive antiplatelet agents or anticoagulants did not show clear evidence of an overall benefit of anticoagulation therapy in patients older than 75 years; a significant reduction in risk of ischemic stroke was potentially offset by a significant increase in the risk of serious hemorrhage.4 Therefore, we conducted the Birmingham Atrial Fibrillation Treatment of the Aged (BAFTA) study, which compared the effects of warfarin therapy with aspirin therapy in a primary care population of elderly patients with atrial fibrillation.
Subjects and methods
We recruited 973 patients from general practices in England and Wales between 2001 and 2004. These subjects were a mean age of 81 years and had atrial fibrillation confirmed on electrocardiography; 30% had newly diagnosed atrial fibrillation and the rest were known to have a history of atrial fibrillation. Before the study began, 40% of subjects were taking warfarin and 42% were taking aspirin. We excluded subjects with rheumatic heart disease, active peptic ulcer disease, a history of intracranial hemorrhage, or a history of a major nontraumatic hemorrhage within the past 5 years. We also excluded subjects with a blood pressure >180/110 mm Hg and those whose primary care physicians concluded should or should not be taking warfarin.
Subjects were randomly assigned to receive 75 mg/day of aspirin or to warfarin therapy to achieve a target international normalized ratio (INR) of 2.5 (acceptable range, 2-3). The frequency of INR testing was determined by the local investigator.
The primary end point of the study was fatal or disabling stroke (ischemic or hemorrhagic), other intracranial hemorrhage, or clinically significant arterial embolism. Two independent neurologists, who were blinded to treatment allocation, reviewed clinical details regarding possible primary events. We also evaluated major secondary outcomes, including extracranial hemorrhage, other vascular events, and all-cause mortality. The primary analysis was intention to treat.
We identified 4639 subjects aged 75 years or older with atrial fibrillation, of whom 973 (21%) were randomly assigned to one of the arms of the study. Subjects were most commonly excluded because the primary care physician determined that the subject should be taking warfarin (1570 cases; 43% of exclusions). In an additional 905 instances (25% of exclusions), the decision was made by the subject. In a few cases, subjects were excluded because of a risk of hemorrhage, either as an absolute contraindication or because of the decision of the primary care physician (n = 528; 14%).
The average follow-up period was 2.7 years. Follow-up was 99% complete, with 4 subjects lost to follow-up in each arm. By the end of the study, 67% of those receiving warfarin and 76% of those receiving aspirin were still taking the assigned therapy. Most of the subjects receiving aspirin who were not taking it at the end of the study (n = 82; 70%) were switched to warfarin. Conversely, most of the subjects receiving warfarin who were not taking it by the end of the study (n = 127; 78%) were switched to an antiplatelet agent (the agent was aspirin in 124 cases).
Subjects taking warfarin had INR values in the therapeutic range 67% of the time, with a mean INR of 2.4. The use of other medications to lower cardiovascular risk was similar in both arms of the trial. Two years after randomization, 25% of subjects receiving warfarin and 23% of those receiving aspirin were taking a lipid-lowering agent. The mean systolic blood pressure at this time was the same in both groups (137 mm Hg).
Table 1. Baseline characteristics for subjects with and without diabetes before and after
propensity score matching.
The risk of a primary end point was significantly lower in subjects taking warfarin compared with those taking aspirin: 1.8% per year versus 3.8% per year, respectively (Table 1). There was a 70% reduction in the risk of ischemic stroke and no significant difference in the risk of hemorrhagic stroke or other intracranial hemorrhage. We assessed the risk of the primary end point by treatment allocation in different patient subgroups (Table 2). The risk of a primary event increased with age, tended to be higher in subjects already known to have atrial fibrillation compared with those found to have the disorder during an opportunistic screening, and was particularly high in subjects with a history of stroke or transient ischemic attack.
Table 2. Warfarin versus aspirin in different subgroups in
the BAFTA study.
There was no evidence that warfarin was less effective in any of these subgroups. Of particular note, we found warfarin to be as effective in subjects aged 85 years and older as in younger subjects. With regard to secondary outcomes, there were no differences in overall mortality, nonstroke vascular events, or the risk of hemorrhage between the warfarin and aspirin groups (Table 3).
We also evaluated the risk of major hemorrhage by subgroup (Table 2). Major hemorrhage in this context was a combination of the intracranial events (hemorrhagic stroke and subdural hemorrhage) that formed part of the primary end point together with extracra-nial hemorrhages that were fatal or resulted in the need for surgery or transfusion. There were only 25 such events in each arm of the trial; thus, power to detect
differences was limited. Nevertheless, there was no evidence in any subgroup that warfarin was associated with more major hemorrhages than aspirin. For example, the risk of bleeding increased by similar amounts with age in both groups, and warfarin did not appear to be any more dangerous than aspirin in the older study participants than in the younger ones.
We found that warfarin was substantially more effective than aspirin in preventing stroke and arterial embolism in an elderly population with atrial fibrillation, and there was no evidence that it led to a greater risk than aspirin. We likely underestimated the benefits of warfarin because of treatment crossovers that occurred during the course of the trial and because a substantial proportion of subjects were excluded from the trial when their primary care physician determined that they should be taking warfarin. It is possible that the treatment crossovers also led to underestimation of the risks of bleeding while taking warfarin, although in that respect, it is reassuring that the on-treatment analysis that we performed also showed no difference in the risk of major hemorrhage between warfarin and aspirin therapy (relative risk, 0.88; 95% confidence interval, 0.46-1.63).
Table 3. Secondary outcomes in the BAFTA study.
Compared with previous trial data, our results are consistent with estimates of the efficacy of warfarin compared with aspirin; however, we observed a different result with regard to major hemorrhage risk. Van Walraven and colleagues found a doubling of risk among patients taking warfarin compared with aspirin.4 Possible explanations for this include chance, higher INR ranges used in earlier trials,7-9 prior exposure to warfarin among 40% of the study population,10 and better control of blood pressure.
Several studies have shown that fewer than half of patients with atrial fibrillation who are older than 75 years are receiving anticoagulation therapy, whether they are being treated in the primary care or hospital setting.11,12 Our results show that warfarin could be safely used much more widely in this age group.
Anticoagulation treatment to achieve a target INR of 2 to 3 should be the preferred treatment for all patients who have atrial fibrillation and are older than 75 years of age, unless there are contraindications or the patient thinks that the benefit of this therapy is not worth the inconvenience of the associated monitoring. Age by itself should not be regarded as a contraindication to anticoagulation therapy.
DISCLOSURE The BAFTA study was funded by the UK Medical Research Council.