Accuracy of GFR estimating equations based on creatinine, cystatin C or both in routine care.

Background New equations to estimate glomerular filtration rate based on creatinine (eGFR_cr), cystatin C (eGFR_cys) or both (eGFR_cr-cys) have been developed by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) and the European Kidney Function Consortium (EKFC). There is a need to evaluate the performance of these equations in diverse European settings to inform implementation decisions, especially among people with key comorbid conditions. Methods We performed a cross-sectional study including 6174 adults referred for single-point plasma clearance of iohexol in Stockholm, Sweden, with 9579 concurrent measurements of creatinine and cystatin C. We assessed the performance of the CKD-EPI 2009/2012/2021, EKFC 2021/2023, revised Lund-Malmö (RLM) 2011 and Caucasian, Asian, Pediatric and Adult (CAPA) 2014 equations against measured GFR (mGFR). Results Mean age was 56 years, median mGFR was 62 mL/min/1.73 m² and 40% were female. Comorbid conditions were common: cardiovascular disease (30%), liver disease (28%), diabetes (26%) and cancer (26%). All eGFR_cr-cys equations had small bias and P₃₀ (the percentage of estimated values within 30% of mGFR) close to 90%, and performed better than eGFR_cr or eGFR_cys equations. Among eGFR_cr equations, CKD-EPI 2009 and CKD-EPI 2021 showed larger bias and lower P₃₀ than EKFC 2021 and RLM. There were no meaningful differences in performance across eGFR_cys equations. Findings were consistent across comorbid conditions, and eGFR_cr-cys equations showed good performance in patients with liver disease, cancer and heart failure. Conclusions In conclusion, eGFR_cr-cys equations performed best, with minimal variation among equations in this Swedish cohort. The lower performance of CKD-EPI eGFR_cr equations compared with EKFC and RLM may reflect differences in population characteristics and mGFR methods. Implementing eGFR_cr equations will require a trade-off between accuracy and uniformity across regions. [ABSTRACT FROM AUTHOR]