Impact of Adopting the 2021 CKD-EPI Equation on CKD Detection in a Multiethnic Dutch Population: The HELIUS Study

Background: Until recently, the standard method of estimating GFR for the diagnosis of CKD included an ethnicity coefficient for those of African descent. There has been debate whether or not to exclude the coefficient from the CKD-EPI equation, which resulted in the 2021 CKD-EPI equation. We explored the impact of adopting this formula on eGFR and CKD detection in a multi-ethnic Dutch population. Methods: We analyzed data of 21,617 participants (mean age 44yr, 42% male) of the multi-ethnic HELIUS cohort study (Amsterdam, the Netherlands). Three groups were distinguished: African Surinamese (4,151), Ghanaian (2,339) and Non-corrected (15,127; i.e., those who required no correction for ethnicity. eGFR was calculated using the uncorrected and corrected 2009 and 2021 CKD-EPI equation. CKD prevalence (i.e., eGFR (<60mL/min/1.73m2) and/or ACR (≥3mg/mmol) was calculated. CKD case detection for all three equations were compared in high risk groups (i.e., diabetes mellitus, hypertension or cardiovascular disease). C-statistics for CKD probability were compared. Inter-rater reliability was measured via Cohen’s kappa and consistency calculation. Results: Compared to the Non-corrected group (mean eGFR 102 mL/min/1.73m2), age- and sex-adjusted differences (p<0.001) in the mean eGFR(SE) were 4.6±0.2 vs -8.9±0.2mL/min/1.73m2 in the African Surinamese and 3.2±0.3 vs -10.4±0.3mL/min/1.73m2 in the Ghanaian participants for the 2009 and 2021 CKD-EPI equation, respectively. CKD prevalences were similar for both equations overall and in subgroups (e.g., 10.9 vs 11.6% (p=0.33) among African Surinamese). CKD case detection did not differ between equations. The C-index for CKD probabilities was not influenced by the equation. The internal consistency in detecting CKD of the 2009 and 2021 equation was high (Figure 1). Conclusion: In our cohort, adoption of the 2021 CKD-EPI equation leads to lower eGFR in African Surinamese and Ghanaian participants, but similar overall estimates of CKD prevalence. Our study indicates that discontinuation of the ethnicity-coefficient may have very little impact on CKD detection in a multi-ethnic Dutch population.