eGFR, CKD-EPI 2021 (creatinine)

Clinical background

Estimation of the glomerular filtration rate from serum creatinine is routine in the assessment and follow-up of renal function. The decisions the equation supports are several: when should chronic kidney disease be diagnosed, when should doses of renally excreted drugs be adjusted, and when should the patient be referred to nephrology? All of these decisions turn on a single figure in mL/min/1.73 m², and small differences in that figure can move the patient across a threshold with clinical consequences.

CKD-EPI 2009 included a race coefficient that gave a higher eGFR for Black patients at the same serum creatinine. Because race is a social and not a biological construct, and because the race coefficient could contribute to systematic differences in care, CKD-EPI 2021 produced a revised equation in which the whole model was refitted without race as an explanatory variable [1]. The NKF-ASN Task Force recommended that all laboratories move immediately to the 2021 equation and replace both MDRD and CKD-EPI 2009 [2].

Calculating eGFR with CKD-EPI 2021 (creatinine)

$$\text{eGFR} = 142 \times \min!\left(\frac{S_{cr}}{\kappa},, 1\right)^{\alpha} \times \max!\left(\frac{S_{cr}}{\kappa},, 1\right)^{-1{.}200} \times 0{.}9938^{\text{age}} \times 1{.}012 ;\text{(if female)}$$

where $\kappa = 0{.}7$ for women and $0{.}9$ for men, and $\alpha = -0{.}241$ for women and $-0{.}302$ for men. $S_{cr}$ is the serum creatinine in mg/dL. For men the factor 1.012 is not applied. The result is given per 1.73 m² of body surface area.

The derivation cohort for the creatinine equation comprised 10 studies with 8,254 participants (31.5% Black), and the external validation cohort comprised 12 studies (7 of them new) with 4,050 participants (14.3% Black) [1]. All participants were 18 years or older. The mean measured GFR in the validation cohort was 76.4 ± 29.6 mL/min/1.73 m². The model relates log-transformed measured GFR to log-transformed serum creatinine, age and sex, with separate slopes for creatinine values above and below $\kappa$, but without race as an explanatory variable.

Interpretation in practice

The eGFR value is assigned to the KDIGO GFR categories, which govern diagnosis and management:

The thresholds of 60, 45 and 30 mL/min/1.73 m² have direct implications for dosing: at 60, dose adjustment of renally excreted drugs is recommended; at 45, nephrology referral and review of high-risk drugs become relevant; at 30, specific dosing instructions for severe renal impairment apply.

Validation and performance

In the external validation cohort, P30 (the proportion of eGFR values within 30% of measured GFR) reached 85–90% for both race groups [1]. Agreement with GFR categories was 59–69%, and lower among Black participants. The new equation underestimated measured GFR among Black participants by a median of 3.6 mL/min/1.73 m² (95% CI 1.8–5.5) and overestimated it among non-Black participants by a median of 3.9 mL/min/1.73 m² (95% CI 3.4–4.4) [1]. The differential bias between the race groups was 7.6 mL/min/1.73 m², greater than for the race-based 2009 equation. This is a direct consequence of removing race: the equation distributes the error more evenly across race groups but increases the total bias for each group separately.

In a validation at an American hospital serving a low-resource population (Parkland, Dallas) with 75,442 Black and White patients, 171 of them with measured GFR (iohexol clearance), CKD-EPI 2021 had the smallest absolute median bias of the three equations compared (CKD-EPI 2009, 2021 and EKFC) [3]. P30 did not differ statistically significantly between CKD-EPI 2021 and EKFC. An American multicentre study with 12,854 participants from nine studies found similar P30 values for CKD-EPI 2021 and EKFC, but less bias for EKFC [4].

Limitations

The equation applies to adults from 18 years of age. Paediatric equations should be used for patients under 18.

Creatinine-based estimates are influenced by muscle mass, diet and protein intake. At extremes of muscle mass, and in sarcopenia, neuromuscular disease or after amputation, the eGFR may be systematically misleading. Pregnant women and patients on dialysis are not covered by the equation.

The race-neutral design entails a trade-off: the equation underestimates GFR in Black patients and overestimates it in non-Black patients [1]. The NKF-ASN therefore recommends cystatin C as a confirmatory test in patients at risk of, or diagnosed with, chronic kidney disease, since combined creatinine–cystatin C equations are more accurate [2].

A Nordic multicentre study (six hospitals in Norway, 5,552 adults) found that CKD-EPI 2021 overestimated GFR in the young (<25 years) and the elderly (≥65 years) and that the EKFC equation performed better in this population [5]. The same study found that the Lund–Malmö equation was comparable to EKFC, which is relevant because local equations have historically been used in Scandinavian laboratories. When implementing CKD-EPI 2021 in a Nordic population, a somewhat higher proportion of patients classified as CKD stage 3–5 than with earlier equations should be expected.

The transition from CKD-EPI 2009 to 2021 means that a substantial proportion of patients are reclassified to a higher CKD stage. In the American cohort, approximately 80% of Black patients and 61% of White patients were reclassified to a higher stage [3], which may lead to more nephrology referrals and dose adjustments. When a laboratory changes equation, the clinical consequences should be reviewed.

References

1. Inker LA et al. New creatinine- and cystatin C-based equations to estimate GFR without race. N Engl J Med 2021;385(19):1737–49. PMID: 34554658
2. Delgado C et al. A unifying approach for GFR estimation: recommendations of the NKF-ASN Task Force on reassessing the inclusion of race in diagnosing kidney disease. Am J Kidney Dis 2022;79(2):268–288.e1. PMID: 34563581
3. Ilori EO et al. Performance of the 2021 estimated glomerular filtration rate CKD-EPI refit and the European Kidney Function Consortium (EKFC) formulas. Diagnostics (Basel) 2025;15(8):1047. PMID: 40310408
4. Delanaye P et al. Performance of the European Kidney Function Consortium (EKFC) creatinine-based equation in United States cohorts. Kidney Int 2024;105(3):629–637. PMID: 38101514
5. Averina M et al. Performance of the European Kidney Function Consortium (EKFC) creatinine-based eGFR equation and other eGFR equations in a north European population. A multicentre study in Norway. Clin Chem Lab Med 2026. PMID: 42343553