*Purpose: Serum creatinine-based estimated glomerular filtration rate (eGFR) models have been shown to over- or underestimate measured iohexol GFR (mGFR), but it is unclear if combining these estimating equations could more accurately predict mGFR in living kidney donor candidates.

*Methods: Using 533 living kidney donors from August 2008 – November 2010, we compared the Modification of Diet in Renal Disease (MDRD), Cockcroft-Gault (CG), and Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) eGFR models with mGFR by iohexol clearance. As the MDRD model was derived from a population with low mean GFR (40 mL/min), we incorporated a previously published refitted model (MDRD*), developed from a cohort with more similar mean GFR (123 mL/min) as our healthy kidney donors. We assessed the performance of these estimating models against mGFR in: bias, precision, and relative accuracy.

*Results: The mean measured GFR in our population was 110.3 ± 16.51 ml/min/1.73 m² with a mean serum creatinine at donation of 0.83 ± 0.16 mg/dL. Analysis of the individual estimating models showed that eGFR(CG) underestimated mGFR by 4.93 ± 22.33 mL/min/1.73 m² and eGFR(MDRD*) overestimated mGFR by 6.92 ± 15.87 mL/min/1.73 m². Averaging eGFR (MDRD*) and eGFR (CG) provided the least bias model with the best accuracy. It overestimated mGFR by 1 ± 18.51 mL/min/1.73 m² and was within 10% of mGFR 49.9% of the time. Moreover, this combined equation was within 30% of mGFR 91% of the time.

*Conclusions: Averaging the refitted MDRD* and CG equations provided the least biased and most accurate model for estimating GFR in kidney donors prior to transplantation.

« Back to 2019 American Transplant Congress