Comparative Prognostic Assessment of EKFC Equations for Mortality Prediction: A Population-based Cohort Study

Background: The European Kidney Function Consortium (EKFC) has introduced equations for estimating glomerular filtration rate (eGFR) using serum creatinine (Cr) and cystatin C (Cys). While their diagnostic accuracy is established, their comparative prognostic value for long-term mortality in the general population is not well-defined. This study compared the prognostic performance of three EKFC equations—creatinine-based (EKFC-eGFRcr), cystatin C-based (EKFC-eGFRcys), and the combined (EKFC-eGFRcrcys) model—for predicting all-cause and cardiovascular (CV) mortality. Method: We analyzed a population-based cohort of 4519 participants from the National Health and Nutrition Examination Survey (NHANES) 1999–2002. eGFR was calculated using all three EKFC variations (EKFC-eGFRcr, EKFC-eGFRcys, and EKFC-eGFRcrcys). The primary and secondary endpoints were all-cause and CV mortality, respectively, through December 31, 2019. Predictive improvements were evaluated using integrated discrimination improvement (IDI), net reclassification improvement (NRI), and number needed to measure (NNM). Result: The eGFR value calculated by the EKFC-eGFRcys, EKFC-eGFRcr, and EKFC-eGFRcrcys was 78.92±22.98, 84.45±23.21, and 81.68±22.13 ml/min/1.73m2, respectively. Over a median follow-up of 211 months, all three eGFR models were independent predictors of mortality (all P <0.001). Restricted cubic spline regression revealed significant non-linear, inverse associations between all eGFR measures and adjusted mortality risks (all P for non-linearity < 0.001). Compared to EKFC-eGFRcr, EKFC-eGFRcys significantly improved risk reclassification for all-cause mortality (categorical NRI = 15.9%; 95% CI: 14.1–17.8%; NNM = 7). The combined EKFC-eGFRcrcys also outperformed the creatinine-only model (NRI = 8.9%; NNM = 12). However, EKFC-eGFRcys demonstrated modestly better performance than the combined equation (NRI = 7.0%; NNM = 15), suggesting limited incremental value in adding creatinine to cystatin C. The superiority of EKFC-eGFRcys was primarily driven by improved classification of non-events. Conclusion: The EKFC-eGFRcys equation provides superior prognostic value for all-cause and CV mortality compared to both creatinine-based and combined EKFC equations. Notably, the combined model offered no prognostic advantage over cystatin C alone. These findings highlight the clinical importance of cystatin C-based eGFR for accurate risk stratification in the general population.