Changes in stroke risk by freedom-from-stroke time in simulated populations with atrial fibrillation: Freedom-from-event effect when event itself is a risk factor

The risk of atrial fibrillation (AF)-related stroke is usually assessed by calculating the CHA(2)DS(2)-VASc score, the components of which are various risk factors, including prior stroke. Although prior stroke is considered the strongest risk factor, the associated risk is actually inferred. Nevertheless, it implies a “freedom-from-event effect” (FEE)—the longer a patient is stroke-free, the lower the stroke risk. Although dynamic prognostication has been applied to cancer, the FEE has been ignored in AF, probably because of methodological difficulties. We conducted a simulation study to evaluate the FEE in the risk of AF-related stroke. We modeled various populations of AF patients and simulated the development of stroke assuming a nonhomogeneous Poisson process, where the hazard depends on age, comorbidities, and individual variability. Parameters were set so that the model respects the CHA(2)DS(2)-VASc scoring scheme and reproduces the 1-year CHA(2)DS(2)-VASc score-wise stroke risk and relative risk conferred by real-world risk factors. We tracked stroke risk over 0 to 15 years of freedom-from-stroke time (FST), both prospective FST (pFST), which begins at the time of diagnosis and continues to the future, and retrospective FST (rFST), which begins at the present and looks backward to the time of diagnosis. The pFST counterbalanced the increase in stroke risk conferred by aging; in patients with a CHA(2)DS(2)-VASc score of 1, the pFST offset 62% of the age-conferred risk increase. The rFST reduced the stroke risk; in patients with a CHA(2)DS(2)-VASc score of 2 and without prior stroke, an rFST of 6.8 years reduced the stroke risk to the midpoint between CHA(2)DS(2)-VASc scores 1 and 2. The study results suggest that the FEE should be considered in evaluating stroke risk in patients with AF. The FEE may be important in other recurrent diseases for which a prior event is a risk factor for a future event.