Discordance Between Serum Creatinine-Based and Combined Serum Creatinine/Cystatin C-Based Estimations of Kidney Function in Acutely Ill, Hospitalized Patients

Introduction: Traditional estimations of kidney function relying on serum creatinine (SCr) have significant limitations, as SCr is affected by many non-kidney determinants, such as body composition, age, nutritional status, volume status, etc. While cystatin C (cysC) is not without its own limitations, support for cysC use as an alternative or additional method of kidney function assessment is developing. The role of cysC-based estimations of kidney function in acutely ill, hospitalized patients has not been fully established. Objective: The objective of this study was to determine the incidence of discordance in estimated glomerular filtration rate (eGFR) between SCr-based calculations and SCr/cysC-based calculations of kidney function. Methods: This was a single-center, retrospective, observational cohort study at an academic medical center including adult inpatients admitted in 2023 with SCr and CysC ordered. The primary outcome was discordance, defined as an occurrence of a 30% or greater difference between Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) eGFR SCr-cysC (eGFRcr-cys) and Cockcroft Gault (CG). Data was collected via chart review. Descriptive statistics were used. Nominal data (%) was analyzed by Fisher’s exact test and continuous data (median [IQR]), including discordance, was analyzed via Wilcoxon matched pair signed ranked sum. For comparison, eGFRcr-cys was individualized for patients’ body surface area. Results: The study included 463 patients (49% female, age 61 [48-72] years, 50.1% critically ill) with SCr 1.14 [0.68-1.82] mg/dL and cysC 2.00 [1.33-2.78] g/dL. Clearance was 60.0 [34.0-100.5] mL/min by CG and 43.0 [24.5-71.0] mL/min by eGFRcr-cys. Discordance rate was 44.3% (205/463). Compared to patients without discordance, patients with discordance were younger (55 years [44-65] vs 65 years [55-74]; P < .001) and had lower median SCr (0.93 mg/dL [0.48-1.58] vs 1.35 mg/dL [0.85-2.10], P < .001). Discordance occurred more often in critically ill patients (52.2%; 121/232 patients) than general medicine/surgical patients (36.4%; 84/231 patients, P < .001). Conclusion: In hospitalized patients, eGFRcr-cys was discordant with CG in 44.3% of the population. Discordance in kidney function estimates likely impacts clinical care, including drug dosing. Further studies are needed to determine optimal use of each estimation.