EDITOR,—I share C M Sudlow and colleagues' concern over whether evidence from studies of the prevention of stroke in patients with atrial fibrillation applies to clinical practice because of the highly selected nature of the patients in randomised trials.1 Compliance and safety may be different when anticoagulants are used in clinical practice rather than in the carefully selected, well motivated, and closely monitored participants in studies, and patients in clinical practice may have different risks of thromboembolism and bleeding. In a recent survey of patients with atrial fibrillation, for example, …