Jørgen Frøkiær

Background The aim of this study was to compare the ability of renal indicators [serum creatinine (SCr), cystatin C (SCysC)] and glomerular filtration rate (GFR)-models to discriminate normal and reduced renal function. As a single cut-off level will always lead to false classifications, we propose using two cut-off levels, dividing renal function into normal or reduced, with an intermediate "gray zone" of indeterminable results.Methods Glomerular filtration rate was measured by plasma clearance of ⁵¹Cr-EDTA (13.7–147.4 mL/min/1.73 m²) in 119 children (age range 2.3–14.9 years).Reduced renal function was defined as a GFR of<82 mL/min/1.73 m². SCr, SCysC, age-normalized creatinine (SCr-ratio), and eight published GFR-models were compared for their ability to correctly classify renal function as normal or reduced. Cut-off levels were determined so as to give 99 % certainty outside the gray zone.Results The multivariable GFR-models by Schwartz et al. (J Am Soc Nephrol 2009; 20:629–637) and Zappitelli et al. (Am J Kidney Dis 2006; 48:221–230) and two models by Andersen et al. [Am J Kidney Dis 2012; 59(1):50–57: body cell mass (BCM)-model and Weight-model] performed significantly better than all other variables (P<0.01), with the BCM-model performing the best (P<0.05). The SCr-based Schwartz formula and SCr-ratio both performed better than SCr and SCysC.Conclusions Among the 119 children enrolled in this study and the renal indicators tested, the BCM-model had the best diagnostic performance in terms of screeningor normal or reduced renal function, and the SCr-ratio was a superiordiagnostic tool to both SCr and SCysC.