Background: The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation seems to correct the overdiagnosis of chronic kidney disease (CKD) provided by Modification of Diet in Renal Disease (MDRD) equation. However, this point has not been tested in some ethnic groups. This study investigated the performance of MDRD and CKD-EPI equations in South Brazilian individuals.
Methods: This cross-sectional study included 354 individuals including healthy volunteers, diabetic and non-diabetic individuals with or without CKD. Glomerular filtration rate (GFR) was measured by the 51Cr-EDTA single-injection method (51Cr-GFR). Accuracy (P30), bias, and Bland-Altman agreement plots were evaluated.
Results: In the group as a whole, 51Cr-GFR was 87±37 (6-187), CKD-EPI eGFR, 82±30 (6-152), and MDRD eGFR, 77±28 (6-156) mL/min/1.73 m2 (p<0.001 for all comparisons). Analyzing the subset of individuals with 51Cr-GFR <60 mL/min/1.73 m2, P30 values were, respectively, 76% and 84% for MDRD and for CKD-EPI (p<0.001) while for 51Cr-GFR ≥60 mL/min/1.73 m2, P30 values were 57.5% for both equations (p=1.000). For MDRD and CKD-EPI, mean bias were negative for GFRs <60 (-11 vs. -12, p=0.221) and positive for values >60 (16 vs. 9, p<0.001). In multivariate analysis, absolute bias was unfavorably influenced by measured GFR >60 (for MDRD) and being diabetic or younger (for CKD-EPI).
Conclusions: CKD-EPI reduces GFR underestimation in individuals with GFRs >60, but still presents a quite low accuracy at this GFR range. Moreover, it tends to overestimate GFR in subjects with GFRs <60 mL/min/1.73 m2. CKD stages 1 and 2, diabetes and young age had a negative influence on the performance of the equations.