Risk of Recurrent Bleeding Events in Nonvalvular Atrial Fibrillation Treated with Vitamin K Antagonists: A Clinical Practice Research Datalink Study.

 There is little evidence on how the occurrence of a bleed in individuals on vitamin K antagonists (VKAs) impacts the risk of subsequent bleeds, and thromboembolic and ischemic events. Such information would help to inform treatment decisions following bleeds.  To estimate the impact of bleeding events on the risk of subsequent bleeds, venous thromboembolism (VTE), stroke, and myocardial infarction (MI) among patients initiating VKA treatment for new-onset nonvalvular atrial fibrillation (NVAF).  We conducted an observational cohort study using a linked Clinical Practice Research Datalink-Hospital Episode Statistics dataset. Among a cohort of individuals with NVAF, the risk of clinically relevant bleeding, VTE, stroke, and MI was compared between the period prior to the first bleed and the periods following each subsequent bleed. The rate and cost of general practitioner (GP) consultations, prescriptions, and hospitalizations were also compared across these periods.  The risk of clinically relevant bleeding events was observed to be elevated at least twofold in all periods following the first bleeding event. The risk of VTE, stroke, and MI was not found to differ according to the number of clinically relevant bleeding events. The rate and cost of GP consultations, GP prescriptions, and hospitalizations were increased in all periods relative to the period prior to the first bleed.  The doubling in the risk of bleeding following the first bleed, taken alongside the stable risk of MI, VTE, and stroke, suggests that the risk-benefit balance for VKA treatment should be reconsidered following the first clinically relevant bleed.