INR variability in atrial fibrillation: a risk model for cerebrovascular events

Use of vitamin K antagonists (VKAs) for stroke prophylaxis in patients with non-valvular atrial fibrillation (NVAF) necessitates frequent monitoring of the international normalized ratio (INR) to avoid the increased risk of hemorrhage associated with excess anticoagulation, or ischemic stroke due to insufficient anticoagulation. We therefore developed a model to estimate the excess morbidity attributable to inadequate INR control in NVAF populations.

Equations expressing the risk of cerebrovascular events as a function of INR were generated using published data. Additional functions were developed to estimate the excess risk attributable to inferior INR control, using the clinical trial setting as the reference.

The derived risk functions were applied to French NVAF patients receiving anticoagulation in routine medical practice. This population achieved a time in therapeutic range (INR 2.0-3.0) of 59%, compared with 68% time in therapeutic range (TTR) in the SPORTIF III and V clinical trials. However, there was considerable variation in the TTR among patients in routine care, of whom 36% were in range for less than 50% of the time. Among this latter group, the relative risk, compared with the clinical trial setting, was 1.47 for ischemic stroke and 2.68 for intracranial hemorrhage. Conversely, for patients achieving a TTR greater than 50%, the relative risks for ischemic stroke and intracranial hemorrhage were 0.99 and 1.16, respectively.

This model permits estimation of the excess risk attributable to inferior INR control in NVAF populations receiving VKA anticoagulation, and has implications for public health planning and management.