Objective: To determine optimal intensity of international normalized ratio (INR) of warfarin therapy for the prevention of ischemic events in patients with non-valvular atrial fibrillation (NVAF), we evaluated the risk of severe recurrent stroke, systemic embolism and major hemorrhagic complications according to INR and age.
Methods: We carried out the National Cardiovascular Center (NCVC) NVAF Secondary Prevention Study and analyzed data with those of Japanese Nonvaluvular Atrial Fibrillation-embolism Secondary Prevention Cooperative Study to elucidate relationships of major stroke and hemorrhage with INR and age. In both studies, all patients with cardioembolic stroke were given warfarin, monitored with INR every month, and followed up for primary endpoints of stroke and embolism to other parts of the body, and for secondary endpoints of major hemorrhagic complications requiring blood transfusion or hospitalization. We regarded ischemic stroke with NIH stroke scale (NIHSS) score > or = 10 or systemic embolism as a major ischemic event and ischemic stroke with NIHSS score <10 as a minor ischemic event. There were 203 patients enrolled in total (152 men and 51 women). We investigated the relationship of occurrence of the events with INR and age, and calculated the incidence rates of major and minor ischemic events and major hemorrhagic events.
Results: During the mean follow-up of 653 days, major ischemic stroke and systemic embolism occurred in only 4 patients with INR <1.6, minor ischemic stroke in 10 patients with INR 1.50-2.66, and major hemorrhage in 9 patients with INR 2.30-3.56. Patients with major ischemic or hemorrhagic events were significantly older than those without any events (75+/-4 years vs. 67+/-7 years, p<0.001 unpaired t test). Incidence rates of any events at INR < or = 1.59, 1.60-1.99, 2.00-2.59 and > or = 2.60 were 8.6%, 3.8%, 4.9%, and 25.7%/year, respectively.
Conclusions: Major ischemic or hemorrhagic events occur often in the elderly NVAF patients, in whom an INR value of between 1.6 and 2.6 seems optimal to prevent such events.