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

Stroke Prevention in AF: Update

Systematic review

Back in Bandolier 108 we reported on a splendid individual patient meta-analysis that compared adjusted dose warfarin (with target INR between 2.5 and 4.0) with low dose (75-325 mg a day) aspirin for the prevention of stroke [1]. Warfarin was better than aspirin at preventing ischaemic stroke, but not much different for other events, like haemorrhagic stroke, myocardial infarction, systemic emboli, vascular or other death. Warfarin was associated with an increased risk of major bleeds.

At the time we thought it was a splendid paper, and commented that until new trials appeared it was likely to be the last and best word. A new systematic review [2] does have new trials of ximelagatran compared with adjusted dose warfarin, but the rest is pretty much as it was in 2002.

Systematic review

The review looked for English language publications published to February 2005 which examined various treatments for stroke prevention, including warfarin (fixed low dose and adjusted dose), aspirin, or ximelagatran. Trials had to be randomised, in patients with nonvalvular atrial fibrillation, and with any additional treatments given equally in each group. Studies could be placebo or active controlled.


Included were 13 trials with 14,423 patients, of whom 7,329 were in two trials comparing adjusted dose warfarin with a fixed dose of 72 mg ximelagatran daily (without INR measurement). Adjusted dose warfarin was compared with placebo in six studies (to 1993), with aspirin in five (to 1996), with fixed low dose warfarin in four (to 1999), and with ximelagatran in two (2003, 2005).

The results for the comparison of adjusted dose warfarin with placebo, aspirin, and low fixed dose warfarin present no surprises, have been analysed before [1], and are not relevant today. The comparison with ximelagatran is relevant, and the results for ischaemic stroke or systemic embolism, death, and major or minor bleed are shown in Table 1. Adjusted dose warfarin and ximelagatran were equivalent for stroke and mortality, but ximelagatran produced significantly fewer major and minor bleeds.

Table 1: Major outcomes in two randomised trials that compared adjusted dose warfarin (INR target 2.0-3.0) and ximelagatran 72 mg daily without INR measurement

Number with
Percent with
adjusted dose warfarin
adjusted dose warfarin
Relative risk
(95% CI)
(95% CI)
Ischaemic stroke or systemic embolism
0.96 (0.72 to 1.3)
not calculated
0.96 (0.79 to 1.2)
not calculated
Major bleed
0.74 (0.57 to 0.96)
110 (60 to 810)
Minor bleed
0.85 (0.80 to 0.91)
18 (13 to 30)
There were two trials lasting an average of 1.6 years. Of the 7,329 patients, 3,663 received 72 mg fixed dose ximelagatran, and 3,665 adjusted dose warfarin, with target INR of 2.0 to 3.0

The event rates for ischaemic stroke, death, or major stroke were similar to those found in the previous individual patient meta-analysis of adjusted dose warfarin and aspirin [1]. The two ximelagatran trials had an average duration of 1.6 years, and the number needed to treat to prevent (NNTp) a major bleed over that time was 110 compared with adjusted dose warfarin. The number needed to treat to prevent one fewer minor bleed was 18.

The authors also calculated annual NNTs for stroke prevention for adjusted dose warfarin, ximelagatran, and aspirin compared with no treatment, at different levels of stroke risk (Figure 1).

Figure 1: Calculated NNTs for stroke prevention over one year compared with placebo, for low, medium, and high annual risk rates


A useful update, possibly providing another treatment option if or when ximelagatran becomes available. There is good evidence from this analysis that ximelagatran has equivalent efficacy to adjusted dose warfarin, with lower bleed rates, and, of course, without the need for frequent INR measurements and dose adjustment. Not everyone is so sanguine, though. Another analysis of the two trials [3] reminds us about a higher withdrawal rate with ximelagatran than with warfarin, and a higher rate of liver enzyme elevations. It is also a useful treatise on what constitutes non-inferiority for pointy-head trials buffs.

The analysis also provides the bases, or a basis, for some thinking on what the costs might be. As of writing, the cost of ximelagatran is not known, but it will be more expensive than warfarin for drug acquisition. That will have to be weighed against potentially less harm, and less hassle.

For 1,000 patients with nonvalvular AF treated for 10 years, there would be about 100 fewer major bleeds, about 500 fewer minor bleeds, and about 10,000 fewer INR measurements with the associated problems about obtaining venous blood samples from older patients, and the stress. We will have to wait for someone to do the sums.


  1. C van Walraven et al. Oral anticoagulants vs aspirin in nonvalvular atrial fibrillation. An individual patient meta-analysis. JAMA 2002 288: 2441-2448.
  2. GY Lip, SJ Edwards. Stroke prevention with aspirin, warfarin and ximelagatran in patients with non-valvular atrial fibrillation: a systematic review and meta-analysis. Thrombosis Research 2005 (epublication ahead of print)
  3. S Kaul et al. Trials and tribulations of non-inferiority. The ximelagatran experience. Journal of the American College of Cardiology 2005 46: 1986-1995.

previous or next story