Background: Although recommended, both the Cockcroft and Gault formula (CG) and the modification of diet in renal disease (MDRD) equation are not ideally predictive of glomerular filtration rate (GFR) in diabetic subjects; we tested whether the new Mayo Clinic Quadratic (MCQ) equation performed better.
Methods: In 200 diabetic subjects with a wide range of renal function, GFR was measured by 51Cr-EDTA clearance, and compared with the results of the three predictive equations by regression analysis and Bland and Altman procedures. The correlations with body mass index, age and albumin excretion rates were tested. The precisions (absolute difference as percentage), diagnostic accuracies [receiver operating characteristic (ROC) curves for the diagnosis of moderate and severe chronic kidney disease (CKD)], and the results of stratification according to the KDOQ classification were compared.
Results: The CG and MCQ overestimated mean GFR, whereas the MDRD underestimated it. Correlation coefficients and areas under the ROC curves were better for the MDRD and the MCQ as compared with the CG, which was biased by body weight (+30% overestimation in obese diabetic subjects). The absolute differences with true GFR were slightly lower for the MDRD than the MCQ, and both better than the CG. Both the MDRD and MCQ correctly stratified 65% of the subjects (CG: 55%, P<0.05). In contrast with the MDRD, the MCQ did not underestimate normal GFR, and its performance for stratification was uniformly good over a wide GFR range.
Conclusions: In diabetic subjects, the MCQ has a similar diagnostic performance to the MDRD, but it does not underestimate normal GFR, which is an important advantage.