Objective To develop a population pharmacokinetic model for colistin sulfate in Chinese critically ill patients, identify factor multifactorial determinants of pharmacokinetic variability including extracorporeal support modalities, and establish a precision dosing framework guided by quantitative pharmacodynamic targets. Methods This study enrolled 59 critically ill adult patients with microbiological evidence of infection caused by carbapenem‐resistant Gram‐negative organisms and antimicrobial susceptibility or empirical treatment. A nonlinear mixed‐effects modeling (NONMEM) approach incorporating 278 concentration measurements was employed to characterize covariate effects, explicitly evaluating ECMO/CRRT impacts. Model robustness was validated through goodness‐of‐fit (GOF), bootstrap, normalized prediction distribution errors (NPDEs), and external clinical verification. Monte Carlo simulations were conducted to quantify target attainment probabilities for f AUC/MIC ≥ 20 under varying renal function and MIC scenarios (0.5∼2 mg·L −1 ). Results A one‐compartment model with first‐order elimination demonstrated superior fit, identifying creatinine clearance (CrCl) as the primary covariate influencing clearance. The integrated validation paradigm confirmed model reliability. The simulation revealed critical limitations of conventional dosing: 1.0∼1.5 MU a day is insufficient for patients with normal renal function or those infected with pathogens with an MIC ≥ 1.0 mg·L −1 . Therefore, the dosing regimen of colistin sulfate should be tailored according to renal function and the MIC. Conclusion To our knowledge, this study establishes the first PPK model integrating dynamic renal function and extracorporeal support effects for colistin sulfate in Chinese critically ill patients. The developed model‐to‐clinic pipeline, combining Monte Carlo simulation‐derived breakpoints and Bayesian forecasting, provides a paradigm for individualized antimicrobial dosing in complex critical care scenarios.
Zhang et al. (Thu,) studied this question.