Predictive ability of creatinine clearance versus estimated glomerular filtration rate for outcomes in patients with non-valvular atrial fibrillation: Subanalysis of the J-RHYTHM Registry.

Background Renal impairment is a risk factor for various adverse events, especially for death. In general, creatinine clearance (CrCl) is used for dose-adjustments of many drugs including oral anticoagulants, and estimated glomerular filtration rate (eGFR) is adopted for the diagnosis of chronic kidney disease. Predictive ability of CrCl versus eGFR for outcomes in patients with non-valvular atrial fibrillation (NVAF) remains controversial; therefore, this was compared using data from the J-RHYTHM Registry. Methods Out of 7406 outpatients with NVAF from 158 institutions, 6004 (age, 69.7 ± 9.9 years; men, 71.2%) having data of CrCl (mL/min, by the Cockcroft-Gault formula), eGFR (mL/min/1.73 m2, by the equations of the Japanese Society of Nephrology), and body surface area (BSA) were analyzed. C-statistics (area under the receiver-operating characteristic curve) of CrCl and eGFR for events were compared by DeLong's test. Results Thromboembolism, major hemorrhage, and all-cause death occurred in 107 (1.8%), 117 (1.9%), and 154 (2.6%) patients during the 2-year follow-up period. C-statistics of CrCl for each event were 0.609 (95% confidence interval, 0.559-0.658), 0.599 (0.548-0.657), and 0.746 (0.706-0.786); and those of eGFR were 0.542 (0.487-0.597), 0.573 (0.519-0.626), and 0.677 (0.631-0.723), respectively. C-statistics of CrCl for thromboembolism and all-cause death were significantly higher than those of eGFR (P < 0.001 for both). These results were consistent when BSA-unadjusted eGFR was used instead of eGFR (P = 0.002 for thromboembolism and P < 0.001 for all-cause death). Conclusions CrCl was superior to eGFR in the prediction of adverse outcomes, i.e., thromboembolism and all-cause death in Japanese patients with NVAF.