When equations disagree: the impact of creatinine-based eGFR in CKD diagnosis and reclassification

This study evaluated the classification patterns of three creatinine-based equations for eGFR – CKD-EPI 2009, CKD-EPI 2021, and EKFC across different age groups in a Romanian population. Although no equation can universally meet the needs of all clinical contexts, understanding how they differ in classifying patients into CKD stages is essential for optimizing detection and management strategies.

Main findings of the study revealed (1) that CKD-EPI 2021 equation consistently resulted in up-classification compared to CKD-EPI 2009, especially in individuals over 40 years of age and (2) most patients were classified into stage G3a using EKFC.

In our study, most of the patient’s samples were collected in the emergency department (ED), with a higher proportion among younger patients compared to the elderly. The ED handles a large number of patients, which makes it important for CKD screening [13]. In Nigeria, a study showed that following the ED admission, 74% of patients were firstly diagnosed with CKD [14].

Switching from 2009 CKD-EPI to 2021 CKD-EPI

On the premise that race is a social concept rather than a biological one, the 2021 CKD-EPI equation is the first to remove the race coefficient. While this equation performs better for African-American patients in the U.S. compared to the previous version, it tends to overestimate eGFR in the rest of patients [15]. In Europe, its performance is suboptimal for the Caucasian population, with only minor improvements for black patients [16].

Among the three equations, CKD-EPI 2021 identified the smallest proportion of patients in G3a, G3b, G4 and G5 especially in individuals over 40 years old. The correlation observed between the CKD-EPI 2009 and CKD-EPI 2021 equations in classifying patients into identical glomerular filtration rate (GFR) categories demonstrated strong to very strong agreement across all age groups, with τ coefficients exceeding 0.85.

Despite the strong agreement between the two equations, when transitioning from the 2009 version to the 2021 one, we observe a constant up-classification trend across all age groups in our study. This trend may delay CKD diagnosis and nephrology referral, particularly around the MDL of 60 mL/min/1.73 m² that separates stages G2 and G3a. For example, in our cohort, up to 29% of patients no longer met CKD criteria when reclassified using the 2021 equation. Similarly, among patients over 40 years old initially classified in stage G4, between 11 and 15% of male and female were reclassified to stage G3b, potentially affecting nephrologist referral and renal replacement planning.

These findings align with previous studies conducted in Caucasian populations. In a similar study conducted in Canada on a Caucasian population, 28% of patients were reclassified from G5 to G4 using CKD-EPI 2021 [17], while a Swedish study reported a 17% shift in the same direction with many additional patients (39%) reclassified into stage G2 and therefore no longer considered to have CKD [18]. The reclassified patients had higher risks for all cause death and mortality for cardiovascular disease (CVD) [18]. However, the classification has not been evaluated according to age distribution.

Switching from the 2009 CKD-EPI to EKFC

Similarly to 2021 CKD-EPI, the European Kidney Function Consortium (EKFC) does not include a race coefficient [11]. Developed using data predominantly from European populations [19], the EKFC equation includes a “Q value” representing the mean serum creatinine level in the derivation cohort [11]. This adaptable Q value has demonstrated the equation’s feasibility, even outside Europe, including in African cohorts [20].

In our analysis, agreement between CKD-EPI 2009 and EKFC in classifying patients into similar GFR categories ranged from moderate to strong in adults under 40, with lower correlations observed in female patients (τ = 0.56 and 0.66 for females vs. 0.67 and 0.80 for males). In contrast, in patients over 40, correlations exceeded τ = 0.88 in both sexes, indicating strong concordance. In a study evaluating concordance and discrepancies among creatinine based eGFR equations in Swedish patients over 60 years old, the correlation between CKD-EPI 2009 and EKFC was poor (Cohen K 0.54) [21].

Transitioning from the CKD-EPI 2009 to the EKFC equation in adults, no consistent directional pattern was observed across all age groups. Notable shifts occurred around key MDLs, at the GFR threshold of 60 mL/min/1.73 m², the EKFC equation exhibited a trend towards down-classification, from G2 to the G3a category, leading to an increased number of patients meeting the diagnostic criteria for CKD. The most significant reclassification occurred in patients over 65 years, especially women, with 19% women and 13% men being reclassified from G2 to G3a, and 13% women and 8% men in the same age group being reclassified from G3b to G4, increasing CKD diagnosis and nephrologist referral. These differences may reflect sex-specific patterns in age-related GFR decline, as suggested by previous studies [22]. At the 15 mL/min/1.73 m² MDL, moderate proportions of patients (less than 9% in each age group) were reclassified to milder GFR stages.

Similar findings have been reported in other studies. An analysis conducted in an American population demonstrated increased CKD prevalence when applying the EKFC Eq. [23], while an Italian study reported comparable patterns of down-classification at early CKD stages [24]. In older adults in Sweden the CKD-EPI 2009 equation showed higher estimates of GFR compared to EKFC [21]. In the northern European population, Russel, et al., showed that the reclassified patients had higher risk of all cause mortality [25]. These observations suggest that EKFC may enhance CKD identification in European populations, particularly in older adults, though further evaluation using measured GFR and clinical outcomes is needed.

2021 CKD-EPI vs. EKFC

When stratified by sex, the agreement between EKFC and CKD-EPI 2021 equations showed notable variability. In female patients, Kendall’s Tau values ranged from 0.59 to 0.88, while in males the correlation was somewhat higher and more consistent (τ = 0.70–0.88). In both sexes, the strongest agreement was observed in individuals over 65 years of age. These findings suggest that while the two equations yield increasingly concordant classifications with advancing age, discrepancies are more pronounced in younger populations, particularly among women. Although the EKFC and CKD-EPI 2021 equations demonstrated a strong correlation in terms of patient ranking by eGFR (τ = 0.60–0.88), they yielded notably different absolute estimates. As a result, EKFC classified a significantly higher number of patients into stage G3a, especially in individuals over 65 years of age. This apparent paradox – strong statistical agreement alongside substantial reclassification – reflects the fact that correlation coefficients assess the relative ordering of patients but not the absolute eGFR values or threshold-based categorization. Therefore, even a strong correlation may mask clinically meaningful differences in staging and potential treatment implications.

Accurate assessment of kidney function is essential for precision medicine CKD. The EKFC equation demonstrates improved performance in older adults compared with the CKD-EPI 2009, due to its development EKFC included several older adult cohorts [9, 11]. Its use is therefore preferable in patients aged > 65 years, ensuring more reliable GFR estimation and optimized clinical decision-making.

These findings must also be interpreted in light of the current international recommendations. Due to the varying performance of creatinine-based equations across populations, the latest KDIGO guideline [8] recommends using an equation validated for the target population. In the United States, the race-free 2021 CKD-EPI is now the standard. However, European bodies such as the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) and the European Renal Association (ERA) have expressed reservations regarding its applicability to European populations. While acknowledging the promise of the EKFC equation, both organizations recommend awaiting further evidence before endorsing its widespread clinical use [26,27,28].