EDITOR,—C M Sudlow and colleagues, and K G Sweeney and colleagues' commentary on their article, exaggerate the difficulties in acting on the research evidence on the use of anticoagulants in atrial fibrillation.1 They point to the risk of bleeding, the need for data on prevalence, and patients' preferences as areas of uncertainty.

Much of the evidence on the incidence of bleeding as a side effect of anticoagulant treatment derives from data recorded in the 1980s and earlier, since when the technology of monitoring has improved. Nevertheless, bleeds are a reason to consider stopping treatment, on the …