Performance of creatinine-based equations for estimating glomerular filtration rate compared to endogenous creatinine clearance

Abstract

Introduction: The guidelines recommend estimating the glomerular filtration rate using serum creatinine-based equations as a predictor of kidney disease, preferably adjusted for local population groups. Methods: Cross-sectional study that evaluated the performance of four equations used for estimating GFR compared to endogenous creatinine clearance (ClCr) in 1,281 participants. Modification of Diet equations in Renal Disease Study Group (MDRD), Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI), CKD-EPI with adjustment for local population (CKD-EPI local) and Full Age Spectrum (FAS) in comparison with endogenous creatinine clearance (ClCr). We used the Quantile Regression to calculate the median bias, interquartile range (IQR), Bland-Altman agreement analysis and 30% margin of error (P30). Results: The mean age of participants was 52.5 ± 16.5 years with 466 women (38%), median ClCr[IQR] of 92.0 [58.0; 122.0] mL/min/1.73 m2, with 320 (25%) participants presenting ClCr < 60 mL/min/1.73 m2. The performance of the local CKD-EPI and FAS equations were superior to MDRD and CKD-EPI in relation to variability (0.92 [0.89; 0.94]) and P30 (90.5% [88.7; 92, 0]). In the group with ClCr < 60 mL/min/1.73 m2, the local CKD-EPI and FAS equations showed less variability than the CKD-EPI and MDRD (0.90 [0.86; 0.98] and 1.05 [0.97; 1.09] vs. 0.63 [0.61; 0.68] and 0.65 [0.62; 0.70], P < 0.01) and best P30 (85.5) % [81.0; 90.0], 88.0% [84.0; 92.0] vs. 52.0% (46.0; 58.0) and 53.0% [47.0; 58 .5], P < 0.01). Conclusion: Local CKD-EPI and FAS equations performed better than CKD-EPI and MDRD when compared to ClCr.