Acute kidney injury (AKI) in critically ill patients is associated with acute morbidity and mortality [1]. Approximately half of intensive care unit (ICU) patients develop AKI [2]. Renal replacement therapy (RRT) is an important treatment modality for patients with severe AKI, with usage rates ranging from 6% in all patients to almost 50% in patients with septic shock [3,4].

Vancomycin is a glycopeptide and is recommended as a first-line treatment for methicillin-resistant Staphylococcus aureus (MRSA) infections [5]. It is also frequently used in ICU patients for the treatment of severe to life‐threatening infections caused by Gram‐positive bacteria, in particular MRSA [5]. The pharmacokinetics (PK) of vancomycin in patients receiving RRT are not only affected by the pathophysiological alterations but also by RRT treatment. When RRT is initiated in these patients, there is an estimated 20% risk of inadequate drug concentrations with standard regimens [6]. Given vancomycin’s narrow therapeutic index, the use of therapeutic drug monitoring (TDM) is essential for precise dosing [[7], [8], [9]]. However, monitoring vancomycin concentrations in patients undergoing RRT poses a significant challenge. This is due to the high prevalence of anemia among critically ill patients [10], coupled with the risk of exacerbating vascular damage through blood sampling procedures. Despite the importance of maintaining therapeutic vancomycin concentrations, specific indices and target values have not been established for patients receiving RRT. Nevertheless, an AUC24 target of 400–600 mg·h/L, which is used in other patient populations, has been recommended for dialysis patients [[8], [11]]. Considering the dual challenges of RRT-induced PK variability and TDM constraints in these patients, what strategies enable reliable simulation of vancomycin regimens that consistently attain therapeutic AUC24 targets?

Population pharmacokinetic (PopPK) analysis, which coupled with Bayesian estimation, is one of the most important methods for individualized therapy [12] and has the potential to enhance and streamline the TDM process [13]. In ICU patients receiving RRT, vancomycin exposure increases due to renal impairment, but decreases from sepsis-induced volume expansion and RRT-enhanced clearance. Therefore, developing the PopPK model in these patients needs to consider these factors. Some studies have established PopPK models of vancomycin for patients undergoing RRT, in or not in the ICU. However, most of them are restricted to the specific RRT modality and settings used in the study concerned. Although a few studies have considered the RRT settings and filter characteristics, with some being externally validated, they cannot be directly extrapolated to different clinical settings without systematic evaluation. The most important issue in model evaluation is whether the established model can predict the PK in a prospective study or can be extrapolated to the patients at other institutions, particularly when the purpose of modeling is a prediction of optimal individual dose [14].

In this study, we systematically evaluated the published vancomycin PopPK models developed mainly in adult patients undergoing RRT, aimed to identify PopPK models for the use of model-informed precision dosing (MIPD) and simulation of the dosing regimens of vancomycin in RRT patients.