Intended for healthcare professionals


Stroke prevention in atrial fibrillation

BMJ 1997; 314 doi: (Published 31 May 1997) Cite this as: BMJ 1997;314:1563

Warfarin is most effective when the INR lies between 2.0 and 4.0

  1. Tim Lancaster, General practitioner (Tim.Lancaster{at},
  2. Jonathan Mant, Clinical lecturer in public health medicinea,
  3. Daniel E Singer, Associate professor of medicineb
  1. a Division of Public Health and Primary Care, Radcliffe Infirmary, Oxford, OX2 6HE
  2. b Harvard Medical School, General Internal Medicine Unit, Massachusetts General Hospital, Boston 02114, MA, USA

    Epidemiological research has established that non-rheumatic atrial fibrillation is an important risk factor for stroke. It increases the risk about fivefold and is particularly important in elderly people, in whom the prevalence of atrial fibrillation is high.1 Randomised trials have shown that this risk is largely reversed by anticoagulation.2 3 The prescription of warfarin for stroke prevention has increased, but concerns about the risks of bleeding continue to dampen enthusiasm for its wider use.4 Targeting those with a higher risk of stroke ensures that only those with most to gain from anticoagulation are exposed to its risks.2 However, reducing the risks will require safer anticoagulation strategies. Three have been tested in recent research: very low intensity warfarin, aspirin, and a combination of these two treatments.

    Information about the efficacy of these strategies is now available from both observational and experimental studies. In a case-control analysis 74 patients with non-rheumatic atrial fibrillation who suffered a stroke while taking warfarin were compared with 222 patients with atrial fibrillation who had not had a stroke while anticoagulated.5 In these patients, the risk of ischaemic stroke, adjusted for other determinants of stroke, increased the further the international normalised ratio fell below 2.0. At a ratio of 1.7 the risk of stroke was twice as high as at a ratio of 2.0. At a ratio of 1.3 it was six times higher. Maximum benefit was achieved at a ratio of 2.0, with no increase in protection against stroke at higher ratios.

    Adding aspirin does not compensate for the reduced efficacy of low intensity anticoagulation. The SPAF (stroke prevention in atrial fibrillation) III randomised trial compared adjusted dose warfarin (international normalised ratio 2.0-3.0) to a combination of fixed dose warfarin (ratio 1.2-1.5 for initial dose adjustment) and aspirin (325 mg/day).6 All the patients had at least one other risk factor for stroke besides atrial fibrillation. The trial stopped early because of a significantly higher rate of ischaemic stroke in those taking the combination of fixed dose warfarin and aspirin: the rate of stroke in the patients taking the combination treatment was 7.9% a year compared with 1.9% a year in those taking adjusted dose warfarin (absolute rate difference 6.0%, relative risk reduction 74%, number needed to treat with adjusted dose warfarin 17). There was no significant difference in rates of major haemorrhage between the two groups.

    In a recent pooled analysis of the early randomised trials, aspirin alone reduced the risk of stroke by 21%.7 This estimate was barely significant (95% confidence interval 0 to 38%). The SPAF III study makes clear that adjusted dose warfarin is much more effective than aspirin. This may seem at odds with the earlier SPAF II randomised trial, which compared warfarin with aspirin alone and included patients without important risk factors in addition to atrial fibrillation.8 In that study warfarin was more effective than aspirin for preventing ischaemic stroke, but in older patients its benefits were cancelled by a higher rate of intracranial haemorrhage. However, the target international normalised ratio for patients treated with warfarin in SPAF II was 2.0-4.5.

    There is increasing evidence that the risk of serious haemorrhage rises at ratios above 4.0.9 10 Provided the international normalised ratio can be kept below this level, warfarin should be preferred to aspirin except in patients with a low risk of stroke or a high risk of haemorrhage. Patients with a low risk of stroke are those aged under 65 and without hypertension, diabetes, or a history of stroke or transient ischaemic attack.2 Patients with a high risk of haemorrhage include those with serious comorbidity and those in whom control of anticoagulation is difficult.9 10 If the risk of haemorrhage is high, then aspirin offers a safer but less effective alternative to anticoagulation.

    For those at higher risk, including older patients, adjusted dose warfarin is the treatment of choice. Existing guidelines recommend a target international normalised ratio for warfarin anticoagulation of 2.0-3.0.11 To reduce the risk of bleeding, some doctors have pragmatically opted for less intensive targets.12 This strategy can no longer be justified. The aim of treatment should be to prevent the international normalised ratio falling below 2.0 or rising above 4.0.5 13 To achieve this with margin for error at both ends of the range, Rosendaal has suggested a target of 2.5-3.5.14 Cautious doctors may prefer to aim for the slightly lower target of 2.0-3.0. The SPAF III trial confirms that this intensity of anticoagulation is highly effective.

    The search to find an effective dosing range below an international normalised ratio of 2.0 seems to be over. Future research in this subject should therefore focus on three areas. Firstly, we need to improve methods for safe management of anticoagulation in routine practice. It is important to know whether specialised anticoagulation units offer a safer service than management by individual general practices. Technical advances such as decision support for dosing and self monitoring by patients also require evaluation.15 Secondly, we need to further define the risk of stroke in atrial fibrillation, perhaps using laboratory markers to supplement risk stratification based on clinical data. Thirdly, we need to develop and test safer antithrombotic drugs.


    • TL is supported by the Imperial Cancer Research Fund. DES has received funding from Dupont Merck Pharmaceuticals and honoraria from Boehringer Mannheim Corporation.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    14. 14.
    15. 15.
    View Abstract