Interesting new article from the BMJ comparing warfarin with NOACs in patients with non-valvular atrial fibrillation, using data from large-scale Danish databases. The findings were:
- Rivaroxaban is marginally more effective than warfarin at preventing ischaemic strokes or systemic embolism, but the AR is 0.3%, which would translate into a NNT of around 333, if my arithmetic is correct. This seems to be a combined outcome, since the difference for ischaemic stroke alone was not statistically significantly different (which is important, since many patients are going to be more bothered by the thought of stroke than embolism, so if embolism where you’re seeing the marginal difference then it may not be as clinically significant).
- Dabigatran and apixaban were not statistically significantly better than warfarin at preventing strokes/embolisms, but…
- ….dabigatran and apixaban were safer. The annual risk of death was lower for either of those two than for warfarin (attributable benefit 3.3% for apixaban and 4.8% for dabigatran), as was the annual risk of any bleeding (AB 1.7% for apixaban and 2.6% for dabigatran). These benefits were not seen when comparing rivaroxaban to warfarin.
This is useful. Based on this, it looks as though dabigatran is the safest choice overall – at least until any further data come in.