Abstract
This study evaluates the predictive performance of three glomerular filtration rate (eGFR) estimating equations—the 2009 CKD-EPIcr, the 2021 CKD-EPIcr, and the EKFC—in identifying the risk of Acute Kidney Injury (AKI). In a large-scale East Asian cohort, the 2009 CKD-EPIcr equation demonstrated superior discriminative performance and clinical utility compared to the newer race-free 2021 version and the EKFC equation.
Introduction
Estimated Glomerular Filtration Rate (eGFR) is a critical metric for assessing kidney health and predicting complications like Acute Kidney Injury (AKI). While the 2009 CKD-EPI equation was the long-standing standard, it utilized a race-based coefficient. In 2021, a race-free equation was introduced to promote health equity. However, the impact of this shift on predictive accuracy in non-Black populations, particularly East Asians, requires rigorous evaluation. This study investigates whether the 2021 CKD-EPIcr or the European Kidney Function Consortium (EKFC) equations maintain the predictive power of the 2009 standard for AKI risk stratification.
Methods
Researchers conducted a retrospective observational study at Seoul National University Hospital, analyzing data from 187,139 individuals between 2011 and 2021. The study compared:
- 2009 CKD-EPIcr: The traditional standard with race coefficients.
- 2021 CKD-EPIcr: The new race-free standard.
- EKFC: The European Kidney Function Consortium equation.
AKI was defined using KDIGO criteria (serum creatinine increase ≥ 0.3 mg/dL within 48 hours or ≥ 1.5-fold within 7 days). Performance was measured using the Area Under the Receiver Operating Characteristic curve (AUROC) and Net Reclassification Improvement (NRI).
Key Insights: The Overestimation Trap
A significant finding was the tendency of the 2021 equation to yield higher median baseline eGFR values (96.5 mL/min) compared to the 2009 version (92.2 mL/min). In an Asian population, this "optimistic" view of kidney health can mask the early signs of decline that precede AKI.
Because the 2021 equation shifts the baseline higher, it requires a larger absolute change in creatinine to trigger a "high risk" signal, potentially delaying clinical intervention. This highlights a tension in modern medicine: while removing race from algorithms is a social imperative, the biological reality of different populations means that a "universal" equation may lose precision in groups that weren't the primary focus of the model's development.
Key Results
The 2009 CKD-EPIcr equation was the clear winner across almost all metrics:
- Predictive Power (AUROC): The 2009 model achieved an AUROC of 0.7583, significantly higher than the 2021 model (0.7564) and the EKFC model (0.7577) ($P < 0.001$).
- Clinical Misclassification: Using the 2021 equation resulted in a negative Net Reclassification Improvement (NRI) of -7.24%. In practical terms, the 2021 model misclassified over 1,100 patients who actually developed AKI as being "low risk."
- Subgroup Consistency: The 2009 model's superiority held true across various subgroups, including age, sex, and the presence of diabetes, with the gap most pronounced in patients with higher baseline kidney function (eGFR ≥ 60).
Conclusion
While the 2021 CKD-EPIcr equation represents a step forward for social equity in some regions, its application in East Asian populations may compromise AKI risk stratification. For clinicians, the "old" 2009 equation remains the more reliable tool for identifying patients at risk of sudden kidney failure. Future research should investigate if adding biomarkers like Cystatin C improves the accuracy of race-free models in diverse ethnic groups.
Summary updated