How to Best Define Patients with Moderate Chronic Kidney Disease

BACKGROUND: The objective of this study was to identify which formula may best identify moderate chronic kidney disease (CKD) (glomerular filtration rate (GFR) cut-off of 60 ml/min/1.73 m(2)). METHODS: We compared the performances of 14 serum creatinine (S(cr)) and 11 cystatin C (Cys C) estimated GFR equations using inulin clearance (Cl(in)) as the reference test in a stable CKD population of 101 patients. Scatter, coefficient of variation, bias, precision, accuracy within 30% ranges from the reference method, agreements and receiving operating characteristics (ROC) of each test were compared. RESULTS: ROC analysis identified Davis, Salzar, Virga and Cockcroft-Gault as the most sensitive (>or=85%) and the isotope dilution mass spectrometry (IDMS), Edwards, MacIsaac as the most specific (95%) to define the GFR cut-off level of 60 ml/min/1.73 m(2). Area under the ROC curve (AUC) was generally >0.8 (p <or= 0.0001). 2 x 2 contingency tables to define CKD demonstrated sensitivity of 90% for Davis, while the IDMS was the most specific (95%). Among the Cys-C-based equations, Filler was the most sensitive (83%) and MacIsaac was the most specific (95%). CONCLUSION: The current equations lack consistent good performance to define CKD. The MDRD-IDMS equation missed 30% but demonstrated a high specificity to confirm those with moderate CKD. A combination of two equations, one sensitive and another specific, may be required for epidemiological studies.