Background: Patients with atrial flutter are at an increased risk of stroke. Many physicians are treating atrial flutter patients with warfarin anticoagulation to reduce the stroke risk, with the attendant increased risk of hemorrhagic complications. No randomized controlled trials or decision analysis models exist to direct this therapeutic decision.

Purpose: A decision analysis model to compare the effectiveness of warfarin, aspirin and no-therapy, in patients with chronic atrial flutter for stroke prevention.

Methods: We developed a semi-Markov model to estimate quality-adjusted survival in patients with atrial flutter treated with no-therapy, aspirin or warfarin. All treatment options included warfarin treatment if the atrial flutter converted to atrial fibrillation. No-therapy patients were prescribed neither aspirin nor warfarin; warfarin was initiated for ischemic events. Aspirin patients were initially treated with aspirin, with conversion to warfarin for ischemic events. The warfarin protocol initiated anticoagulation in all patients with atrial flutter, with conversion to aspirin for hemorrhagic complications. The risk of stroke in atrial flutter was estimated from observational studies. Probabilities of major hemorrhage, stroke severity, death rates and quality of life estimates were extrapolated from the atrial fibrillation literature. The base case was a 65 years old warfarin eligible patient with atrial flutter, without prior stroke with an annual stroke rate of 2.7%.

Results: In the base case warfarin treatment provided 14.6 QALYs, 13.6 QALYs with aspirin and 13.2 QALYs with no-therapy. The model was sensitive to the stroke rate, major hemorrhage rate on warfarin, utility of the well state on warfarin, the mortality risk reduction and its duration when treated with warfarin, relative risk of dying with atrial flutter, efficacy of treatment with warfarin and relative risk of stroke with age. A three-way sensitivity analysis varying stroke rate, hemorrhage rate on warfarin and utility of warfarin treatment, showed changing treatment strategies in the range of interest.

Conclusions: In our model, treatment with warfarin increases QALYs compared to no-therapy and aspirin in patients with atrial flutter and without other risk factors for stroke. Since the model is sensitive to many key parameters, which cannot be accurately estimated, randomized control trials should be performed to make definitive recommendations about anticoagulation therapy for stroke prevention patients with atrial flutter.