Incidence and predictors of severe bleeding during warfarin treatment

Background Optimal warfarin prescription requires correct, individualized assessment of the warfarin-related bleeding risk, which randomised controlled trials may underestimate . Observational studies have reported a range of bleeding risks that differ 40-fold. This variation may be caused by time trends, variation in bleeding definition and study subject selection. We investigated the incidence of, and risk factors for severe bleeding in un-selected warfarin-treated patients from Sweden. Methods Between 2001 and 2005, 40 centres recruited warfarin-naive patients commencing warfarin therapy and followed them prospectively with continuous registration of clinical data. The primary outcome was severe bleeding, according to the WHO universal definition of severe adverse drug reactions. The influence of potential risk factors was investigated by means of a Cox proportional-hazards model. Result A total of 1523 patients contributed 1276 warfarin-exposed patient-years. The incidence of first-time severe bleeding was 2.3 per 100 patient-years (95% confidence interval 1.4 to 3.1). Male sex and use of drugs potentially interacting with warfarin were the only independent risk factors of severe bleeding, with hazard ratios of 2.8 and 2.3, respectively. Age, target International Normalized Ratio (INR), time spent outside target INR range, and warfarin dose requirement were not significantly associated with bleeding risk. Conclusions The risk of severe bleeding in a large naturalistic, prospective cohort of first-time warfarin users was lower than reported in some previous reports. Male gender was an independent predictor of severe bleeding as was the receipt of warfarin-interacting medications at the onset of anticoagulation therapy. Further studies are required to evaluate the effect these findings may have on the quality of current risk-benefit analysis involved in warfarin prescription.