Current issues of ACP Journal Club are published in Annals of Internal Medicine


Long-term, low dose warfarin prevented stroke in nonrheumatic atrial fibrillation

ACP J Club. 1991 Mar-April;114:33. doi:10.7326/ACPJC-1991-114-2-033

Source Citation

The Boston Area Anticoagulation Trial for Atrial Fibrillation Investigators. The effect of low-dose warfarin on the risk of stroke in patients with nonrheumatic atrial fibrillation. N Engl J Med. 1990;323:1505-11.



To assess the effectiveness of low-dose warfarin in preventing strokes in patients with nonrheumatic atrial fibrillation.


Randomized single-blinded, controlled trial, with patients followed every 3 months.


32 university-affiliated medical centers in the United States. Most patients were referred by their family physicians.


420 patients with chronic sustained or intermittent atrial fibrillation, and no evidence of mitral stenosis, were randomized. Patients were excluded if they had a clinical indication or contraindication for anticoagulation or if they required aspirin therapy.


212 patients were allocated to warfarin and 208 to the control group. Warfarin doses were determined using a computer-based protocol. The target range for the prothrombin-time ratio was 1.2 to 1.5 times the control value. All patients taking warfarin were told not to take aspirin. Patients in the control group were allowed to take aspirin.

Main outcome measures

The primary end point was ischemic stroke. Secondary end points were major and minor bleeding events and death. Outcome event documentation was reviewed by committees that were blinded to treatment assignment.

Main results

The mean follow-up period was 2.3 years. 2 of 212 patients (1%) in the warfarin group had a stroke compared with 13 of 208 patients (6%) in the control group (P = 0.002). {This absolute risk reduction (ARR) of 5% means that 19 patients would need to be treated (NNT) with warfarin (compared with no warfarin) for a mean of 2.3 years to prevent 1 additional stroke from occurring, 95% CI 10 to 50; the relative risk reduction (RRR) was 85%, CI 41% to 96%}.* Of 13 strokes in the control group, 8 occurred among patients on aspirin. There were 2 major bleeding events in the warfarin group and 1 in the control group. {0.9% vs 0.5%; absolute risk difference 0.4%m CI -3% to 2%, P = 0.6}.* There were 38 minor bleeding events (18%) in the warfarin group, of which 6 led to hospitalization or transfusion, compared with 21 (10%) in the control group, of which 7 led to hospitalization or transfusion. {This absolute risk difference of 8% means that 1 additional patient had bleeding for every 13 patients who received warfarin, CI 7 to 83; the relative risk increase was 78%, CI 9% to 191%}.* 11 patients in the warfarin group died compared with 26 in the control group {5% vs 13%, P = 0.005; NNT 14, CI 8 to 51; RRR 58%, CI 20% to 79%}.*


Long-term, low-dose warfarin therapy was effective in preventing stroke in patients with chronic nonrheumatic atrial fibrillation.

Sources of funding: National Heart, Lung, and Blood Institute; Eliot B. Shoolman Fund; Louise U. Snell; Vera and J.W. Gilliland Fund.

Address for article reprint: Dr. J.P. Kistler, Stroke Service, Massachusetts General Hospital, Boston, MA 02114, USA.

*Numbers calculated from data in article.


This well-designed and well-conducted trial convincingly shows that low-dose warfarin effectively reduces the risk for stroke in a population of mainly elderly patients (85% 60 y or older) with nonrheumatic atrial fibrillation. The results are consistent with other studies of anticoagulation therapy in atrial fibrillation. The absence of double-blinding should not be a drawback in this study because end points were assessed centrally in a blinded fashion. Other issues suggest some caution in applying the results to clinical practice:

1. Early termination of the study because of highly significant beneficial effects in favor of warfarin. Although the direction of the results is unlikely to have been caused by chance, it is possible that the magnitude of benefits may be overestimated.

2. Few end points. Altogether 15 strokes were recorded. Because ischemic stroke was the primary end point, a misclassification of a few end points could potentially influence the results. More events would have led to a greater certainty.

3. Use of aspirin in controls only. This could potentially confound the results and cause confusion to clinicians as to the relative efficacy of aspirin, which has been shown to be effective in reducing stroke related to atrial fibrillation. This study was not designed to compare the effects of warfarin with those of aspirin, and no conclusion should be drawn for such.

Nevertheless, information from this and other similar studies clearly points to the effectiveness and safety of anticoagulation therapy for stroke prophylaxis in patients with sustained or intermittent atrial fibrillation.

Koon K. Teo, MD, PhD
University of Alberta HospitalEdmonton, Alberta, Canada