Résumé
Introduction and Aims: Creatinine clearance (CrCl), an indicator of renal filtration, is the main factor to define and follow CKD patients, adapt and prepare their treatments. It is measured from timed urine collection and blood samples or estimated from blood samples and clinical parameters using equations. We compared the performances of different equations in predicting CrCl.Methods: Creatinine level was determined in 24h-urine and plasma samples from nephrology outpatients from 2004-2018. CrCl was estimated using the Cockcroft-Gault (CG) formula with observed weight and the simplified MDRD and CKD-EPI formulas after removing the correction for body surface area (BSA) using the Dubois equation (MDRD_BSA; CKD-EPI_BSA). We compared estimated CrCl (eCrCl) to measured CrCl (mCrCl) in terms of bias (average difference “eCrCl - mCrCl”), precision (standard deviation of differences “eCrCl - mCrCl”) and accuracy (P30 = proportion of eCrCl within mCrCl ± 30%). Differences in accuracy across subgroups were tested by logistic regressions adjusting for mClCr.Results: There were 728 participants (62% male, aged 70 ± 13 (mean ± SD) years, BMI 28.7 ± 5.4 kg/m², mCrCl 58 ± 38 mL/min). Bias and precision were -8.0 ± 20.3 mL/min with CG, -2.8 ± 19.1 mL/min with CKD-EPI_BSA and 2.4 ± 19.1 mL/min with MDRD_BSA (table). Bias was positive at low mClCr, near 0 at average mClCr, and negative at high mClCr, with values higher with MDRD_BSA and lower with CG (table). At low mClCr, MDRD_BSA and CKD-EPI_BSA overestimated mClCr more than CG. Accuracy was higher with CKD-EPI_BSA (74%) and lower with CG (66%), table. At low mCrCl, accuracy was lower with CKD-EPI_BSA and MDRD_BSA (P<0.001), and in females with CG; while in males, accuracy of CG was higher (Pint <0.001). Accuracy of CKD-EPI_BSA was higher in all subgroups of mCrCl, except for males with low mClCr (table). Accuracy was lower in younger patients with MDRD_BSA (P=0.02) and CKD-EPI_BSA (P=0.001), with levels similar to that of CG.Conclusions: Overall bias, precision and accuracy of MDRD_BSA and CKD-EPI_BSA were better than CG. Performances varied with mClCr, gender and age. At low mClCr, MDRD_BSA and CKD-EPI_BSA were more positively biased. Still, accuracies of MDRD_BSA and CKD-EPI_BSA were better than CG in most subgroups. The use of eGFR uncorrected for BSA for dose adjustment may be relevant, provided that consequences on drug safety and efficacy are further considered.