Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Aspirin or anticoagulant in nonvalvular AF

Nonvalvular atrial fibrillation increases the risk of stroke by about four times. The issue is not so much whether to do anything, but rather what to do. Should treatment be with oral anticoagulants like warfarin, or with aspirin? Use of warfarin implies intermittent measurement of INR, and perhaps an increased risk of bleeding.

A number of reviews have addressed this, and generally they conclude that warfarin is more effective. A new meta-analysis has gone a step further, and obtained information on each individual patient [1].


There were six randomised trials with 4,052 patients. Patients were assigned to full-dose oral anticoagulant (target INR 2.5-4.0), or aspirin (75-325 mg/day) and (in some patients) low dose warfarin (median 2.0 mg/day). Low dose warfarin patients were included because there was no difference between them and because including such patients would be likely to minimise differences between treatments.

Patients were stratified according to risk of stroke:
Six outcomes were sought:
Baseline features for each individual patient were provided by investigators. These included previous stroke or transient ischaemic attack, hypertension, congestive heart failure, diabetes and coronary artery disease. Mean follow up was for 1.9 years. Strokes were assessed by neuroimaging in 97% of cases.


Oral anticoagulant was given to 1,939 patients and aspirin with or without low dose anticoagulant in 2,113. The groups were well matched at baseline after pooling data from six studies. Their mean age was 72 years and 60% were men. There was a high risk of stroke in 65%, moderate risk in 27% and low risk in 8%.

The main results are shown in Table 1. Oral anticoagulants were associated with reduced rates of all strokes, ischaemic strokes and cardiovascular events compared with aspirin. There was a significantly increased rate of major bleeding.

Table 1: Outcomes in trials of aspirin versus oral anticoagulant for AF.

Hazard ratios below 1 favour oral anticoagulant


Events/100 patient-years



Oral anticoagulant


Hazard ratio
(95% CI)

All stroke 2.4 4.5 0.6 (0.4 to 0.7)
Ischaemic stroke 2.0 4.3 0.5 (0.4 to 0.6)
Haemorrhagic stroke 0.5 0.3 1.8 (0.9 to 3.9)
All cardiovascular events 5.5 7.8 0.7 (0.6 to 0.9)
Myocardial infarction 0.7 1.0 0.6 (0.4 to 1.0)
Systemic emboli 0.2 0.3 0.7 (0.3 to 1.7)
Vascular death 3.1 3.2 1.0 (0.8 to 1.2)
Major bleeding 2.2 1.3 1.7 (1.2 to 2.4)
Lethal bleeding 0.4 0.2 2.2 (0.9 to 5.3)
All-cause death 4.9 5.2 0.9 (0.8 to 1.1)
Rows in bold show significant differences

Across different subgoups the lower rate of ischaemic stroke and higher rate of major bleeds was consistent (age more or less than 75 years, men versus women, prior stroke or transient ischaemic attack, presence or absence of hypertension, congestive heart failure, diabetes, and different stroke risk).

Treating 1000 patients for one year with oral anticoagulants rather than aspirin would prevent 23 ischaemic strokes while causing nine additional major bleeding episodes.


This is a thorough review, and unless or until new trials are undertaken, it is likely to be the last and best word. The authors go to some length to explain any differences between this and previous meta-analyses, especially some differences in trial inclusion. Because of availability of data at the level of the individual patient rather than from published reports, several more trials could be included in the analysis.

They also explore how acceptable the magnitude of the effect might be to patients. They review studies that have looked at how patients view stroke. Patients have a strong aversion to stroke (the think it almost as bad as death), and would be willing to take warfarin even if the absolute risk reduction was 1% per year. Oral anticoagulants hit that 1% target.

A particular strength of the paper is that it is written to be read by people, not computers. A must-read paper.


  1. C van Walraven et al. Oral anticoagulants vs aspirin in nonvalvular atrial fibrillation. An individual patient meta-analysis. JAMA 2002 288: 2441-2448.
previous or next story in this issue