Comparison of estimated glomerular filtration rate equations on prediction of mortality, kidney failure, and acute kidney injury

Abstract

Background The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI)ASR-NB2009 estimated glomerular filtration rate (eGFR) equation has shown substantial overestimation of GFR in Europeans, hence new equations have been developed. We examined the effect of introducing the European Kidney Function Consortium (EKFC) or Lund-Malmö revised (LMR) eGFR equations on KDIGO eGFR category classification in a large cohort. We compared the EKFC and LMR equations with the CKD-EPIASR-NB2009 formula in view of discriminative ability of all-cause mortality, kidney failure with replacement therapy (KFRT) and acute kidney injury (AKI) risks across eGFR categories. Methods Individuals aged ≥18 years with a serum creatinine measurement (December 2006–July 2024) at University Medical Center Utrecht, were included. Hazard ratios (HRs) were analysed for all outcomes per eGFR category, per equation. Harrell's Concordance index (C-index) was used to assess the ability of risk discrimination across eGFR categories. Whether reclassification between eGFR categories was justified by the occurrence of events, was assessed with net reclassification improvement analysis. Results In total, 285686 individuals were included. Compared with the CKD-EPIASR-NB2009 equation, the EKFC and LMR estimated GFR lower [mean –6.3 (standard deviation, SD 5.3) and –10.7 (SD 6.5) mL/min/1.732, respectively]. The number of individuals with eGFR <60 mL/min/1.73 m2 increased 29.0% (EKFC) and 36.4% (LMR). The EKFC predominantly reclassified older individuals, and the LMR older men, to worse eGFR categories. HRs of reclassified individuals to worse eGFR categories were mainly higher compared with the non-reclassified. The EKFC and LMR equations showed equal/improved C-index for mortality (EKFC 0.584/LMR 0.588/CKD-EPIASR-NB2009 0.570), KFRT (0.895/0.900/0.897) and AKI (0.606/0.609/0.599). The LMR equation reclassified more individuals without an event to worse eGFR categories. Conclusion eGFR category classification was substantially different when using the EKFC or LMR equation compared with the CKD-EPIASR-NB2009 formula. Both equations showed equal to improved ability of risk stratification across eGFR categories. Shifts in eGFR category classification might significantly impact clinical decisions. Given that we have identified variation between equations, a careful consideration of the advantages and disadvantages of different eGFR equations is essential.