Atrial fibrillation: warfarin prevents stroke more effectively than aspirin, but causes more major bleeds

Clinical bottom line (level 1a)

  1. Patients with atrial fibrillation who take adjusted-dose warfarin compared with placebo are less likely to have a stroke (NNT = 18 at 1.6 years) or die (NNT = 68 at 1.6 years) , but more likely to have a major extracranial bleed (NNT = 33 at 1.6 years) .
  2. Patients with atrial fibrillation who take antiplatelets compared with placebo are less likely to have a stroke (NNT = 39 at 1.5 years) , but not clearly less likely to die.
  3. Patients with atrial fibrillation who take adjusted-dose warfarin compared with aspirin are less likely to have a stroke (NNT = 34 at 2.2 years) , but not clearly less likely to die.
  4. Patients with atrial fibrillation who take adjusted-dose warfarin compared with low-dose or fixed-dose warfarin are not clearly less likely to have a stroke or die, but are more likely to have a major extracranial bleed (NNT = 500 at 1.7 years) .
Hart et al: Annals of Internal Medicine 1999; 131 (7): 492-501
Expires November 2003

The study

Systematic review of all randomised controlled trials of
  • Patients: atrial fibrillation
  • Intervention: adjusted dose warfarin, antiplatelets
  • Outcome: stroke, death, intracranial bleeding, major extracranial bleeding


    • Articles found in all languages using OVID/Medline, 1966 to 1999 (search terms: detailed in Cochrane Library ) and enquiries to the Cochrane Collaboration Stroke Review Group and the Antithrombotic Trialists Collaboration.

    Selection criteria: by 2 independent reviewers
    Appraisal criteria: detailed in Cochrane Library
    Articles excluded if:
    • atrial fibrillation associated with prosthetic cardiac valves or mitral stenosis
    • data on patients with atrial fibrillation not reported separately


    16 studies found involving 9874 patients followed for a mean of 1.7 years
    Studies were not found to be significantly heterogeneous.

    The evidence

    Outcome Time to outcome CER OR
    (95% CI)    		 NNT
    (95% CI)
    stroke: adjusted-dose warfarin v. placebo 1.6 years 133/1450
    (9.2%)    		 0.38
    (0.28 to 0.52)     		18
    (16 to 24)
    death: adjusted-dose warfarin v. placebo 1.6 years /
    (%)    		 0.74
    (0.57 to 0.96)     		63
    ( to )
    major extracranial haemorrhage: adjusted-dose warfarin v. placebo 1.6 years /
    (%)    		 2.4
    (1.2 to 4.6)     		-33
    ( to )
    stroke: antiplatelet agents v. placebo 1.5 years 236/1815
    (13.0%)    		 0.78
    (0.62 to 0.98)     		39
    (22 to 440)
    death: antiplatelet agents v. placebo 1.5 years /
    (%)    		 0.84
    (0.67 to 1.05)
    stroke: adjusted-dose warfarin v. aspirin 2.2 years 123/1421
    (8.7%)    		 0.64
    (0.40 to 0.73)     		34
    (20 to 46)
    death: adjusted-dose warfarin v. aspirin 2.2 years /
    (%)    		 0.92
    (0.70 to 1.21)
    stroke: adjusted-dose warfarin v. low or fixed dose warfarin 1.7 years /
    (%)    		 0.62
    (0.32 to 1.20)
    major extracranial haemorrhage: adjusted-dose warfarin v. low or fixed dose warfarin 1.7 years /
    (%)    		 2.0
    (1.2 to 3.4)     		-500
    ( to )
    major intracranial haemorrhage: adjusted-dose warfarin v. low or fixed dose warfarin 1.7 years /
    (%)    		 2.1
    (1.0 to 4.6)

    Citation

    1. Hart RG, Benavente O, McBride R, et al: antithrombotic therapy to prevent stroke in patients with atrial fibrillation: a meta-analysis. Annals of Internal Medicine 1999; 131 (7): 492-501
    Search Terms: hand-search
    Contributor: Chris Ball and Clare Wotton, November 1999
    Reviewer:

    Clinical Question.
    Patient atrial fibrillation
    Intervention or Exposure anticoagulants or anti-platelets
    Outcome death, stroke