Standardization of serum creatinine is essential for accurate use of unbiased estimated GFR equations: evidence from three cohorts matched on renal function

Background Differences in the performance of estimated glomerular filtration rate (eGFR) equations have been attributed to the mathematical form of the equations and to differences between patient demographics and measurement methods. We evaluated differences in serum creatinine (SCr) and eGFR in cohorts matched for age, sex, body mass index (BMI) and measured GFR (mGFR). Methods White North Americans from Minnesota (n = 1093) and the Chronic Renal Insufficiency Cohort (CRIC) (n = 1548) and White subjects from the European Kidney Function Consortium (EKFC) cohort (n = 7727) were matched for demographic patient characteristics (sex, age ± 3 years, BMI ± 2.5 kg/m2) and renal function (mGFR ± 3 ml/min/1.73 m2). SCr was measured with isotope dilution mass spectrometry (IDMS)-traceable assays in the Minnesota and EKFC cohorts and with non-standardized SCr assays recalculated to IDMS in the CRIC. The Minnesota cohort and CRIC shared a common method to measure GFR (renal clearance of iothalamate), while the EKFC cohort used a variety of exogenous markers and methods, all with recognized sufficient accuracy. We compared the SCr levels and eGFR predictions [for Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) and EKFC equations] of patients fulfilling these matching criteria. Results For 305 matched individuals, mean SCr (mg/dL) was not different between the Minnesota and EKFC cohorts (females 0.83 ± 0.20 versus 0.86 ± 0.23, males 1.06 ± 0.23 versus 1.12 ± 0.37; P > .05) but significantly different from the CRIC [females 1.13 ± 0.23 (P