Aims Limited data are available on the impact of renal function on the outcome of patients with atrial fibrillation (AF). Methods and results AMADEUS was a multicentre, randomized, open-label non-inferiority study that compared fixed-dose idraparinux with conventional anticoagulation by dose-adjusted vitamin K antagonists. We performed a post hoc analysis to assess the impact of renal function on the outcomes of anticoagulated AF patients. The primary efficacy outcomewas the composite of stroke/systemic embolism (SE). The principal safety outcome of this analysis was major bleeding. We calculated c-indexes, reflecting the ability for discriminating diseased vs. non-diseased patients, and the net reclassification improvement (NRI, an index of inferior/superior performance of risk estimation scores). Of 4576 patients, 45 strokes and 103 major bleeding events occurred following an average follow-up of 325 +/- 164 days. Patients with CrCl >90 mL/min had an annual stroke/SE rate of 0.6% compared with 0.8% for those with CrCl 60-90 mL/min and 2.2% for those with CrCl <60 mL/min (P < 0.001 for linear association). After adjusting for stroke risk factors, patients with CrCl <60 mL/min had more than two-fold higher risk of stroke/SE and almost 60% higher risk of major bleeding compared with those with CrCl >= 60. In patients with the CHA(2)DS(2)VASc score 1-2, CrCl <60 mL/min was associated with eight-fold higher stroke risk. When added to the CHA(2)DS(2)VASc or CHADS(2) scores CrCl <60 mL/min did not improve the c-indexes for CHADS(2) (P=0.054) or CHA(2)DS(2)VASc (P=0.63) but resulted in significant NRI (0.26, P=0.02) in this anticoagulated trial cohort. Conclusion Renal impairment (CrCl <60 mL/min) doubles the risk of stroke and increased the risk of major bleeding by almost 60% in anticoagulated patients with AF. Renal impairment was additive to stroke risk prediction scores based on a significant NRI, but no significant improvement in discrimination ability (based on c-indexes) for CHA(2)DS(2)VASc or CHADS(2) was observed.
