Predictive Performance of Bayesian Dosing Software for Vancomycin in Intensive Care Unit Patients
Year: 2025
Type: article
Study design: Observational Study
Abstract: BACKGROUND: According to the updated guidelines, Bayesian-derived area under the curve estimation is recommended to guide vancomycin dosing. However, the Bayesian dosing software that facilitates this procedure has not been adequately assessed in intensive care unit (ICU) patients. This study evaluated the performance of 3 commonly used Bayesian software programs in predicting vancomycin concentrations in ICU patients before they could be utilized for personalized dosing in this population. METHODS: Retrospective data from adult ICU patients who were administered vancomycin intravenously were obtained to predict serum concentrations a priori (based solely on patient characteristics) or a posteriori (Bayesian forecasting using measured concentrations). The predictive performance was evaluated via bias and precision using relative bias (rBias) and relative root mean squared error, respectively. RESULTS: Data from 139 patients with 284 vancomycin concentrations were evaluated using 3 software programs: SmartDose (He model), Pharmado (Yasuhara model), and PrecisePK (Rodvald and Goti model). All 3 programs showed clinically acceptable bias with the exception of the Goti model of PrecisePK in an a priori estimation (rBias, 27.44%). A relatively low level of precision in terms of relative root mean squared error was observed in all these programs, but with a marked improvement in the a posteriori estimation (27.69%-37.64%) compared with the a priori situation (45.12%-68.59%). CONCLUSIONS: Bayesian dosing software is a potential tool for vancomycin dose optimization in ICU patients. Patients with different physiological and pathological features may be referred to specific Bayesian programs.
Source: Therapeutic Drug Monitoring
Authors: Gali Bai, Hui Qi, Yaqun Huang, Jiao Zhang, Huiying Zhao, Ruiting Wen, Xiaohong Zhang
Language: English
Subfield: Pharmacology
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W4407679331