Agreement between two routine methods of estimation of glomerular filtration rate in patients with advanced and terminal chronic renal failure

Background: Estimations of glomerular filtration rate (GFR) obtained either by the modification of diet in renal disease study equation (MDRD-GFR) or by classic 24-hour urine collection-based methods (mean of creatinine and urea clearance (Ccr-ur)) are considered to be equivalent in patients with chronic renal failure (CRF). However, the agreement between both methods has been insufficiently studied in patients during the most advanced stages of CRF. Methods: We compared 615 estimations of GFR performed by both methods simultaneously in adult (> 18 years) patients with advanced (aCRF) (15 - 30 ml/min/1.73m(2)) and preterminal (tCRF) (< 15 ml/min/1.73m(2)) chronic renal failure. We also analyze the influence of some relevant covariables (demographic characteristics, inflammatory and nutritional markers) with respect to the concordance between both methods. Results: In aCRF, mean GFR were 19.7 +/- 5.5 (MDRD-GFR) and 19.3 +/- 3.7 ml/min/1.73m(2) (Ccr-ur) (mean difference 0.4 ml/min/1.73m(2), 95% confidence interval Cl -0.3/1.1, p = 0.26), with an intraclass correlation coefficient of 0.46. In tCRF, mean GFR was 12.5 +/- 4. 2 and 10.4 +/- 2.7 mi/min/1.73m(2), respectively (mean difference 2.1 ml/min/1.73m(2), 95% CI 1.7/2.4, p < 0.0005), with an intraclass correlation coefficient of 0.43. Multivariate analysis identified lean body mass, body mass index, protein nitrogen appearance, proteinuria, gender, age, albumin (aCRF) and prealbumin (tCRF) as variables independently correlated with the difference MDRD-GFR minus Ccr-ur. Lean body mass was by far the strongest predictor of deviations between both methods, both in aCRF (R-2 = 0.66, p < 0.0005) and tCRF (R-2 = 0.49, p < 0.0005). Conclusions: MDRD-GFR and Ccr-ur show an acceptable agreement in advanced stages of chronic renal failure. However, MDRD-GFR produces estimations of GFR systematically higher than those given by the Cer-ur method, in patients with tCRF. Moreover, this overestimation is particularly marked in some high risk subsets, including elderly patients and those presenting markers of a poor nutritional condition. Until this issue is further clarified, GFR should be estimated using Ccr-ur rather than MDRD-GFR in patients with tCRF, as also in older and malnourished patients with aCRF, as this may represent a more conservative and safer approach at the time of planning initiation of renal replacement therapy.