Anticoagulation control and prediction of adverse events in patients with atrial fibrillation: a systematic review.

To date, there has been no systematic examination of the relationship between international normalized ratio (INR) control measurements and the prediction of adverse events in patients with atrial fibrillation on oral anticoagulation. We searched MEDLINE, EMBASE, and Cochrane through January 2008 for studies of atrial fibrillation patients receiving vitamin-K antagonists that reported INR control measures (percentage of time in therapeutic range [TTR] and percentage of INRs in range) and major hemorrhage and thromboembolic events. In total, 47 studies were included from 38 published articles. TTR ranged from 29% to 75%; percentage of INRs ranged from 34% to 84%. From studies reporting both measures, TTR significantly correlated with percentage of INRs in range (P<0.001). Randomized controlled trials had better INR control than retrospective studies (64.9% versus 56.4%; P=0.01). TTR negatively correlated with major hemorrhage (r=-0.59; P=0.002) and thromboembolic rates (r=-0.59; P=0.01). This effect was significant in retrospective studies (major hemorrhage, r=-0.78; P=0.006 and thromboembolic rate, r=-0.88; P=0.03) but not in randomized controlled trials (major hemorrhage, r=0.18; P=0.33 and thromboembolic rate, r=-0.61; P=0.07). For retrospective studies, a 6.9% improvement in the TTR significantly reduced major hemorrhage by 1 event per 100 patient-years of treatment (95% CI, 0.29 to 1.71 events). In atrial fibrillation patients receiving orally administered anticoagulation treatment, TTR and percentage of INRs in range effectively predict INR control. Data from retrospective studies support the use of TTR to accurately predict reductions in adverse events.