Performance of creatinine-based equations for estimating glomerular filtration rate changes over time

BACKGROUND: Glomerular filtration rate (GFR) is commonly used to monitor chronic kidney disease (CKD) progression, but its validity for evaluating kidney function changes over time has not been comprehensively evaluated. We assessed the performance of creatinine-based equations for estimating GFR slope according to patient characteristics and specific CKD diagnosis. METHODS: In the NephroTest cohort study, we measured GFR 5324 times by chromium 51-labeled ethylenediamine tetraacetic acid renal clearance in 1955 adult patients with CKD Stages 1-4 referred to nephrologists (Stages 1-2, 19%) and simultaneously estimated GFR with both the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) and Modification of Diet in Renal Disease (MDRD) equations for isotope dilution mass spectrometry traceable creatinine; absolute and relative GFR slopes were calculated using a linear mixed model. RESULTS: Over a median follow-up of 3.4 [interquartile range (IQR) 2.0-5.6] years, the decline in mean absolute and relative measured GFR (mGFR) and CKD-EPI and MDRD estimated GFR (eGFR) was 1.6 ± 1.2, 1.5 ± 1.4 and 1.3 ± 1.3 mL/min/1.73 m2/year and 5.9 ± 5.3, 5.3 ± 5.3 and 4.8 ± 5.2%/year, respectively; 52% and 55% of the patients had MDRD and CKD-EPI eGFR slopes within 30% of mGFR slopes. Both equations tended to overestimate the GFR slope in the youngest patients and underestimate it in the oldest, thus producing inverse associations between age and mGFR versus eGFR slope. Other patient characteristics and specific CKD diagnoses had little effect on the performance of the equations in estimating associations. CONCLUSIONS: This study shows little bias, but poor precision in GFR slope estimation for both MDRD and CKD-EPI equations. Importantly, bias strongly varied with age, possibly due to variations in muscle mass over time, with implications for clinical care and research.