Skip navigation

Increased risk of intracranial haemorrhage with aspirin plus warfarin

 

Clinical bottom line

The addition of aspirin to warfarin results in a small increase in the risk of intracranial haemorrhage, by about 0.7% (NNT 147, 95%CI 79 to 996).


Reference

RG Hart et al. Increased risk of intracranial hemorrhage when aspirin is combined with warfarin: a meta-analysis and hypothesis. Cardiovascular diseases 1999 9: 215-217.

Review

The review sought randomised trials testing aspirin combined with warfarin with warfarin at the same intensity. Trials in which the intensity of warfarin use was different between aspirin/no aspirin arms were not included. The diagnosis of intracranial haemorrhage was on clinical criteria in two trials predating reliable neuroimaging.

Results

There were six studies, in which the aspirin dose varied between 75 mg a day (the largest study) to 1000 mg a day. One trial compared warfarin plus aspirin to warfarin plus dipyridamole. There were 3,874 patients and 31 incidents of intracranial haemorrhage. The rate with warfarin alone was 9/1947 (0.5%) and with warfarin plus aspirin there were 22/1927 (1.2%) (Figure 1).

Figure 1: L'Abbé plot of intracranial haemorrhage with warfarin alone and warfarin plus aspirin

 

The relative risk was 2.4 (95% confidence interval 1.2 to 4.8) and the NNT 147 (79 to 996). This means that for every 147 patients treated with warfarin plus aspirin there will be one additional intracranial haemorrhage that would not occur with warfarin alone.

Comment

The authors are understandable cautious, because the number of events is small and the confidence interval around the relative risk is wide. It would not take many additional intracranial haemorrhages to have been missed with warfarin alone, or to have occurred less frequently because of random chance, for statistical significance to be lost. Perhaps as few as two additional haemorrhages would have accomplished this.

However, unless there were compelling reasons for combining aspirin and warfarin it might be better not to.