Atrial fibrillation (AF) predisposes to stroke, particularly in patients with rheumatic heart disease, congestive heart failure, arterial hypertension, diabetes mellitus or uncontrolled thyrotoxicosis.1 In those with rheumatic heart disease it is usual to give warfarin to reduce the incidence of stroke, although there has been no randomised controlled trial on which to base this approach. Whether patients with nonrheumatic AF should be anticoagulated was unclear 6 years ago.2 This article reviews the evidence from recent randomised controlled trials and considers whether anticoagulation with warfarin, or antiplatelet therapy with aspirin, should now be routine for patients with nonrheumatic AF.

The incidence of ischaemic stroke (embolic or thrombotic) is higher in patients with persistent AF than in patients with normal rhythm. The risk of stroke persists as long as AF is present but appears to be greatest in the first 1-2 years after onset of the arrhythmia.1,3 Up to 20% of those with AF who suffer a first stroke have a further episode within one year.4

Nonrheumatic AF occurs in 1.7% of people aged 60-64 years, increasing to 11.6% in those over 75 years old.5 People with AF without underlying rheumatic heart disease are 5-6 times more likely to have a first stroke or transient ischaemic attack (TIA) compared with a control population of matched age and blood pressure6; in those with AF and rheumatic heart disease the risk is 17 times greater than in controls.6 The absolute annual risk of stroke in patients with AF of all causes is around 5%3 but varies with age and the presence of other risk factors including hypertension, diabetes mellitus, congestive heart failure and thyrotoxicosis.

Role of warfarin
Five large randomised controlled trials (3 open7,8,9, 2 double blind10,11) have investigated anticoagulation with warfarin for the prevention of a first stroke (primary prevention) in patients with nonrheumatic AF. Another study investigated whether warfarin reduced the risk of further stroke in patients who had already had a minor ischaemic stroke or a TIA (secondary prevention).12

Primary prevention
Most patients in the 5 primary prevention trials had no previous history of stroke; about 6% had had an ischaemic stroke or TIA, usually at least 6 months before the study began, so for them prevention was secondary. The mean age of patients was 69 years, and in over two thirds AF had been present for more than a year. The average length of follow up ranged from 1-3 years.

Data from these 5 studies have been pooled in a meta analysis.13 1225 patients with nonrheumatic AF (1889 patient years) were anticoagulated; the control groups contained 1226 patients (1802 patient years). Twenty seven anticoagulated patients (1.4% per year) had a stroke (either ischaemic or haemorrhagic but excluding TIA) whilst 81 strokes occurred in the control patients (4.5% per year), a reduction in risk of 68% in warfarin treated patients compared with controls (95% confidence interval 50-79%). Benefit was seen in both men and women. For 12 patients on warfarin (0.6% per year) and 36 control patients (2% per year) the stroke was fatal or caused severe disability. Risk of stroke was higher in older patients, in those with diabetes mellitus or a history of hypertension and in the few patients who had suffered a previous stroke or TIA.13 Patients under 65 years old without any of these risk factors (15% of all patients in the studies) were at low risk of stroke (annual rate 1%) even when not treated.

The annual rate of proven cerebral haemorrhage was uniformly low in the trials ­ 0.3% in warfarin treated patients and 0.1% in controls.13 1.3% of patients receiving warfarin had an intracranial or other major haemorrhage (chiefly gastrointestinal) requiring admission to hospital or transfusion, compared with 1.0% in the control group.

These results suggest that in patients with nonrheumatic AF warfarin prevents 31 strokes for every 1000 patients treated annually, at a cost of 3 major haemorrhages. Treating 71 patients for a year avoids one severe or fatal stroke.

In these studies patients were excluded if they had an existing indication for anticoagulant treatment (e.g. a prosthetic heart valve), an absolute contraindication such as uncontrolled hypertension or any of a wide range of relative contraindications such as liver disease or alcohol abuse. Some studies also excluded patients with hyperthyroidism or certain cardiovascular disorders such as unstable angina, recent heart attack or severe heart failure. As a result, 3 studies included fewer than 10% of screened patients with AF.10,11,14 Dropout rates were high in anticoagulated patients (up to 38%), often at the patient's personal choice or because compliance was poor. These limitations mean that the results should be translated with caution into wider clinical practice, though many of the reasons for exclusion would also apply in routine practice.

Paroxysmal atrial fibrillation The meta analysis included 462 patients with paroxysmal AF. Untreated patients had an annual stroke rate of 5.7%, which was reduced to 1.7% in those anticoagulated.13 The results suggest that risk of stroke is as high in patients with paroxysmal AF as in those with continuous AF, and that warfarin is effective in preventing stroke in both groups.

Lone atrial fibrillation In patients under 60 with 'lone AF' (no evidence of cardiovascular disease, or of an underlying cause for the arrhythmia) the incidence of stroke was 0.5% per year in one retrospective study.15 In the recent meta analysis the annual rate of stroke was zero in untreated patients under 60 considered to have lone AF (n=112) but rose gradually and exceeded 2% in those over 70 years.13 On current evidence warfarin is not indicated for patients under 60 years old with lone AF.

Secondary prevention
In the only large trial reported, warfarin reduced the annual risk of stroke in patients with nonrheumatic AF and a history of minor ischaemic stroke or TIA within the previous 3 months.12 Of 225 patients randomised to open treatment with warfarin for a mean of 2-3 years, 20 had a stroke (annual rate 4%), of which 8 were fatal or caused severe disability; of 214 patients on placebo, 50 had a stroke (annual rate 12%), of which 19 were fatal or severely disabling, a relative risk for stroke of 0.34 for anticoagulated patients compared with controls (95% confidence interval 0.2-0.57). Benefit appeared to last as long as warfarin was continued.

Major bleeding complications were 4 times more common in patients on warfarin (2.8% per year) than in those on placebo (0.7% per year). None of the 13 who bled taking warfarin had a proven cerebral haemorrhage. At the relatively high level of anticoagulation used in the study (International Normalised Ratio [INR] 2.5-4.0) warfarin would prevent 80 strokes for every 1000 patients treated at a cost of 21 major haemorrhages.12

Warfarin or aspirin?
Two of the primary prevention trials7,14 and the secondary prevention study12 also investigated whether antiplatelet therapy with aspirin reduced the risk of stroke compared with placebo.

In one primary prevention study, 15 of 336 (4.5%) patients treated with aspirin 75 mg daily and 16 of 336 (4.8%) on placebo suffered a first stroke during the 2 years of follow up.7 In the second study, the annual rate of stroke or systemic embolism was 3.6% with aspirin 325 mg daily, compared with 6.3% with placebo, a reduction in annual risk of 42% (95% confidence interval 9-63%).14 Secondary analysis of these results suggests that benefit from aspirin is mainly in patients with a history of hypertension; this finding needs confirmation.13

In the secondary prevention study, involving a total of 1007 patients, aspirin 300 mg daily reduced the annual incidence of vascular death, nonfatal stroke, nonfatal heart attack or systemic embolism by 25%.12 This accords with the large meta analysis reported in 1994.16,17 The annual rate of stroke was 10% with aspirin and 12% in controls, which was not in itself a statistically significant reduction.12

Taken overall, the studies suggest that aspirin prevents stroke less effectively than warfarin, but they also showed that aspirin is less likely to cause major haemorrhage.

A more recent primary prevention study, not included in the meta analysis, directly compared aspirin 325 mg daily with warfarin.14 715 patients less than 75 years old were treated for a mean of 3 years and 385 older patients for 2 years. Over half in each age group had other risk factors for stroke (history of hypertension, previous thromboembolism, or heart failure). Compared with aspirin, warfarin reduced the incidence of ischaemic stroke by around 30%. However, only in patients under 75 with other risk factors for stroke did this translate into absolute benefit. Aspirin treated patients under 75 with no risk factors other than AF had a low rate of thromboembolism (0.5% annually), and treatment with warfarin was no more effective. In older patients, benefit from warfarin was offset by the higher risk of major bleeding, including intracerebral haemorrhage (4.2% per year with warfarin, 1.6% with aspirin), which was well above that seen in older patients in the other primary prevention studies. This may have been partly because of target INR (2.0-4.5) was higher than in most of these studies. In all, 36 aspirin treated and 33 warfarin treated patients suffered a fatal or severely disabling ischaemic or haemorrhagic stroke. The results appear to confound some of the findings of earlier trials, and further studies are clearly needed in patients with AF under 75 who have no other known risk factors for stroke.

The effect of combining anticoagulant and antiplatelet therapy is being assessed.18 The combination cannot be recommended at present.

Monitoring anticoagulant therapy
The target INR in the recent studies ranged from 1.5-2.78 to 2.7-4.5. Current U.K. guidelines recommend an INR of 2-3.19,20 Monitoring was at least monthly in 3 of the primary prevention studies, which is a shorter interval than most U.K. anticoagulant clinics routinely use. This may in part explain the lower bleeding rates observed in the trials than in normal clinical practice, where annual rates of major bleeding of 5-11% have been reported during anticoagulant therapy.21 The risk of intracranial haemorrhage and subdural haematoma rises substantially if the prothrombin time ratio exceeds 2.0, particularly in the elderly.22 Trials assessing the efficacy of low dose warfarin, either fixed dose (1.25 mg daily) or low INR (1.2-1.6), are in progress.18

Starting and stopping therapy
When to begin anticoagulant therapy or aspirin after suspected thromboembolic stroke is controversial1; some strokes in patients with AF are due to primary intracerebral haemorrhage 23 and in other patients cerebral haemorrhage occurs secondarily to infarction, usually in the first 48 hours.1 It seems prudent to defer antithrombotic treatment for at least 48 hours after any stroke, and until haemorrhage has been excluded by CT scan.24

No definitive data exist on how long patients should remain on antithrombotic therapy. The risk of stroke appears to last as long as AF is present, so it seems sensible to continue treatment indefinitely, unless serious bleeding occurs or a contraindication arises.12

Contraindications and precautions
Pregnancy, active peptic ulcer, uncontrolled hypertension and infective endocarditis are all contraindications for the use of warfarin. Many other conditions increase the hazards of anticoagulant therapy or make its control difficult. They include a history of gastrointestinal bleeding, severe heart failure, liver disease, renal failure, alcoholism, mental impairment, thrombocytopenia or coagulation disorders, anaemia or the need for a drug that interacts with warfarin (e.g. a nonsteroidal anti inflammatory drug, anticonvulsant, rifampicin). In patients with these conditions monitoring should be particularly close.

Aspirin is contraindicated in patients with active peptic ulcer, bleeding disorders, known hypersensitivity to aspirin, or a definite history of aspirin or NSAID induced asthma.17

Cost/benefit considerations
A Swedish study has analysed the medical and economic effects of antithrombotic therapy for the prevention of stroke in patients with persistent nonrheumatic AF, including costs of monitoring therapy and direct and indirect costs of hospital and lifelong community care for stroke sufferers.21 The authors concluded that anticoagulation of patients under 80, and aspirin for older patients and those in whom warfarin was contraindicated, was cost effective. Treatment reduced the stroke rate and the cost of medical care by 4% annually, but only if the rate of bleeding complications was low and the annual rate of cerebral haemorrhage on treatment was less than 1.3%. Effective and efficient monitoring is therefore crucial: even in the carefully monitored primary prevention studies over 25% of measurements of INR fell outside the target range for the study.

Anticoagulation reduces the risk of thromboembolic stroke by two thirds in men and women with persistent or paroxysmal nonrheumatic atrial fibrillation. About 31 patients with AF need to be anticoagulated to prevent one first stroke annually. In patients who have already had a stroke or transient ischaemic attack, anticoagulation prevents one further stroke each year for every 11 patients treated. Achieving full benefit depends on careful anticoagulant control and good compliance with treatment. Unless there are contraindications warfarin should be offered to patients up to the age of 80 after careful explanation of the known benefits and risks. In those over 80 years old the risk of cerebral haemorrhage during treatment with warfarin increases, and for many such patients aspirin may be a better option. Patients under 60 years who have 'lone AF' are at very low risk of stroke and should not be anticoagulated. For patients under 65 with no other risk factors for stroke, such as diabetes mellitus, a history of hypertension, stroke or transient ischaemic attack, benefit from anticoagulation also appears marginal.

Aspirin in a dose of 325 mg daily also prevents stroke in patients with AF but is probably less effective than warfarin. Management is simpler and cheaper than with warfarin, and aspirin is less likely to cause haemorrhage. It should always be considered in AF when warfarin is contraindicated or declined.

Reprinted with permission from Drug and Therapeutics Bulletin 1994;32:57-60.



  1. Cerebral Embolism Task Force. Cardiogenic brain embolism: the second report of the Cerebral Embolism Task Force. Arch Neurol 1989;46:727-43.
  2. Atrial fibrillation and stroke: do anticoagulants help? Drug Ther Bull 1989;27:45-7.
  3. Cairns JA, Connolly SJ. Nonrheumatic atrial fibrillation: risk of stroke and role of antithrombotic therapy. Circulation 1991;84:469-81.
  4. Sherman DG, Hart RG, Easton JD. The secondary prevention of stroke in patients with atrial fibrillation. Arch Neurol 1986;43:68-70.
  5. Lake FR, Cullen KJ, de Klerk NH, et al. Atrial fibrillation and mortality in an elderly population. Aust NZ J Med 1989;19:321-6.
  6. Wolf PA, Dawber TR, Thomas HE Jr, Kannel WB. Epidemiologic assessment of chronic atrial fibrillation and risk of stroke: the Framingham study. Neurology 1978;28:973-7.
  7. Petersen P, Boysen G, Godtfredsen J, et al. Placebocontrolled randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation. The Copenhagan AFASAK study. Lancet 1989;i:175-9.
  8. 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.
  9. Stroke Prevention in Atrial Fibrillation Investigators. Stroke prevention in atrial fibrillation study: final results. Circulation 1991;84:527-39.
  10. Connolly SJ, Laupacis A, Gent M, et al. Canadian atrial fibrillation anticoagulation (CAFA) study. J Am Coll Cardiol 1991;18:349-55.
  11. Ezekowitz MD, Bridgers SL, James KE, et al. Warfarin in the prevention of stroke associated with nonrheumatic atrial fibrillation. N Engl J Med 1992;327:1406-12.
  12. EAFT (European Atrial Fibrillation Trial) Study Group. Secondary prevention in nonrheumatic atrial fibrillation after transient ischaemic attack or minor stroke. Lancet 1993;342:1255-62.
  13. Atrial Fibrillation Investigators. Risk factors for stroke and efficacy of antithrombotic therapy in atrial fibrillation. Analysis of pooled data from five randomized controlled trials. Arch Intern Med 1994;154:1449-57.
  14. Stroke Prevention in Atrial Fibrillation Investigators. Warfarin versus aspirin for prevention of thromboembolism in atrial fibrillation: the Stroke Prevention in Atrial Fibrillation II Study. Lancet 1994;343:687-91.
  15. Kopecky SL, Gersh BJ, McGoon MD, et al. The natural history of lone atrial fibrillation. A population based study over three decades. N Engl J Med 1987;317:669-74.
  16. Antiplatelet Trialists' Collaborative Group. Collaborative overview of randomised trials of antiplatelet therapy. Part I: prevention of death, myocardial infarction and stroke by prolonged antiplatelet therapy in various categories of patients. Br Med J 1994;308:81-106.
  17. Aspirin to prevent heart attack or stroke. Drug Ther Bull 1994;32:1-3.
  18. Major ongoing stroke trials. Stroke 1994;25:541-5.\r\n
  19. British Society for Haematology. Guidelines on oral anticoagulation: second edition. J Clin Pathol 1990;43:177-83.
  20. How to anticoagulate. Drug Ther Bull 1992;30:77-80.
  21. Gustafsson C, Asplund K, Britton M, et al. Cost effectiveness of primary stroke prevention in atrial fibrillation: Swedish national perspective. Br Med J 1992;305:1457-60.
  22. Hylek EM, Singer DE. Risk factors for intracranial haemorrhage in outpatients taking warfarin. Ann Intern Med 1994;120:897-902.
  23. Sandercock P, Bamford J, Dennis M, et al. Atrial fibrillation and stroke: prevalence in different types of stroke and influence on early and long term prognosis (Oxfordshire community stroke project). Br Med J 1992;305:1460-5. \r\n
  24. Sherman DG, Dyken ML, Fisher M, et al. Antithrombotic therapy for cerebrovascular disorders. Chest 1992;102(4 Suppl):529S-537S.