Stroke-free duration and stroke risk in patients with atrial fibrillation: simulation using a Bayesian inference

Background. Prior stroke or transient ischemic attack (TIA) is a major predictor of stroke in patients with atrial fibrillation (AF). It may be regarded as a Bayesian predictor by which some underlying causes can be inferred from the result. Similarly, stroke-free duration may be a predictor of freedom from stroke, but this subject has remained unaddressed. Objective: To evaluate the above hypothesis by simulating how stroke-free duration can affect estimated stroke risk in patients with AF using a Bayesian inference. Methods: The development of stroke was modeled as a yearly Bernoulli trial. A Bayesian inference method was applied to the model with correction for the risk increase by aging. The model was composed of two risk parameters: prior stroke (or TIA) and aging. The parameters were treated as random variables. Their distributions were determined to agree with the clinical data reported in the meta-analysis by Stroke Risk in Atrial Fibrillation Working Group. In the simulation, we noted the stroke risk after 1,.., 10 stroke-free years for initial risk 1,..,8%/year. Results: Increasing stroke-free duration reduced the estimated stroke risk, offsetting the risk increase by aging. In patients whose stroke (including TIA) risk was 3%/year or less, the estimated risk remained almost constant with aging if they were stroke-free. In patients whose risk was 4%/year or higher, the risk decreased if they were stroke-free. Conclusion: Our analysis predicted that stroke-free duration in patients with AF is an important predictor of freedom from stroke. Increasing stroke-free duration-will reduce the estimated stroke-risk, which may influence our decision to start anticoagulation. Stroke-free duration should be considered when estimating stroke risk in patients with AF.