Abstract
Chronic atrial fibrillation is generally thought to cause stroke by atrial thrombus formation with subsequent embolization. Rheumatic heart disease previously led to most cases of atrial fibrillation, but recent substantial declines in the incidence of rheumatic fever have changed the distribution of causes of atrial fibrillation. Elderly patients have an appreciable prevalence of chronic atrial fibrillation. The risks and potential benefits of long-term anti-coagulation for elderly patients with atrial fibrillation are discussed and the quality of the evidence assessed. It is concluded that the evidence is sufficiently incomplete and imperfect for a large, well-designed trial to be needed; however, this would be difficult and expensive.