Improved Prediction of Vancomycin Pharmacokinetics in Patients Receiving Continuous Renal Replacement Therapy Using a Novel Population Model Validated With Real-World Data

BACKGROUND: This study was conducted to develop a novel population pharmacokinetic (popPK) model for vancomycin in hemodialysis patients using a large, real-world data set and compare its predictive performance against published models for use in model-informed precision dosing. METHODS: Routine clinical care data from 2409 adult patients across 91 US sites, including 821 patients undergoing continuous renal replacement therapy (CRRT), were included in this analysis. A two-compartment popPK model was developed using a training set comprising 70% of the data and evaluated against a test set (30%). Predictive performance was evaluated using clinical accuracy, mean percent error, and normalized root mean square error against five published vancomycin popPK models for hemodialysis patients. RESULTS: The new popPK model, which incorporated fat-free mass, CKD-EPI 2021-estimated glomerular filtration rate, and age as covariates, and an additive model for CRRT clearance, demonstrated superior predictive performance. It showed significantly higher accuracy, lower bias, and lower normalized root mean square error compared with published models. The model accurately described vancomycin PK under CRRT, with an estimated additive CRRT clearance of 1.02 L/h. Inclusion of nonhemodialysis patients in model development stabilized parameter estimates without affecting predictive performance in CRRT patients. CONCLUSIONS: We developed a novel vancomycin popPK model for adult CRRT patients that outperformed published models in a large real-world hemodialysis data set. This model improved predictive performance for serum vancomycin prediction, suggesting its potential to improve target attainment based on model-informed precision dosing for hemodialysis patients.