RESUMO
PURPOSE: Estimated glomerular filtration rate (GFR) is a useful tool for the detection of chronic kidney disease (CKD). Several methods have been proposed, but findings can vary in specific groups such as patients with diabetes, elderly and high and low body mass index and, also, with the stage of CKD. The objective of this study was comparing the accuracy of the currently used equations for estimating GFR with that of the gold standard technetium-(99m)-diethylene triamine pentaacetic acid (99mTc-DTPA). METHODS: We performed a cross-sectional study of 129 patients with all five CKD stages. GFR was estimated using the following: 24-h urine creatinine clearance, Cockcroft-Gault equation, MDRD equation, CKD-EPI equation, Hoek's cystatin C equation, and isotopic 99mTc-DTPA (as gold standard). We evaluated agreement in the whole study population and according to age, sex, weight, and diabetes. RESULTS: All methods had good agreement. The best agreement was observed with the cystatin C [intraclass coefficient correlation (ICC) 95 % confidence interval (95 % CI), 0.87 (0.82-0.91)], followed by CKD-EPI [ICC 0.83 (0.77-0.88)]. Twenty-four-hour urine creatinine clearance showed the worst agreement in patients older than 65 years [ICC 0.70 (0.56-0.79)]. The Cockcroft-Gault equation showed the worst agreement in younger than 65 years [ICC 0.64 (0.42-0.79)]. The best agreement for classification in the correct CKD stage was with the cystatin C equation [κ = 0.80 (0.74-0.87)]. GFR was overestimated with all methods in CKD stages 4 and 5. CONCLUSIONS: The methods used in clinical practice are adequate for classification of CKD. Cystatin C is the most accurate method, followed by CKD-EPI. The Cockcroft-Gault equation is not accurate in young patients. Twenty-four-hour urine creatinine clearance loses accuracy in patients aged older than 65 years.
