Operational Feasibility of A Priori Covariate-Based Meropenem Exposure Prediction Using a Southeast Asian Population Pharmacokinetic Model in Indonesian Critically Ill Patients with Sepsis

Background Meropenem efficacy in critically ill patients depends on 100% free drug time above MIC (fT > MIC), yet pharmacokinetic variability frequently leaves this target unmet. Therapeutic drug monitoring (TDM) is recommended but rarely available in resource-limited ICUs, including Indonesia. A covariate-based a priori dosing workflow was operationalized using an external Southeast Asian population pharmacokinetic (popPK) model. Methods A retrospective pharmacometric cohort study enrolled 69 ICU adults with sepsis. Without measured concentrations, random effects were fixed at zero (η = 0), yielding covariate-based a priori predictions. Pharmacokinetic parameters were derived from the Boonpeng et al. (2022) two-compartment model using creatinine clearance, serum albumin, and shock status. Missing albumin (39.1%) was imputed by MICE. MICs derived from ATLAS Asia-Pacific 2020–2024. Outcomes were examined by bootstrap-validated bivariate analysis, Firth penalised logistic regression, and three-layer sensitivity analyses. Results Fifty-four patients (78.3%; 95% CI 66.7–87.3%) were predicted to attain 100% fT>MIC. All 15 inadequate patients (21.7%) carried resistant pathogens (MIC 32 mg/L), predominantly carbapenem-resistant Acinetobacter baumannii , making attainment status effectively equivalent to susceptibility status. Predicted attainment was not associated with 72-h treatment failure (RR 1.16; 95% CI 0.58–2.29; P = 0.665) or in-hospital mortality (RR 0.74; 95% CI 0.35–1.56; P = 0.535), consistent across all sensitivity analysis. This approach predicted more than a sixfold variation in individual clearance (2.21–13.58 L/h) and identified 21.7% with predicted subtherapeutic exposure, entirely based on routinely available clinical covariate data without requiring therapeutic drug monitoring (TDM). Conclusions A covariate-based dosing workflow using an external Southeast Asian popPK model was operationally feasible in a resource-limited Indonesian ICU, stratifyng patients by predicted exposure using only routinely available covariate data, without requiring TDM infrastructure. This study demonstrates proof of concept for this technical pathway rather than clinical validation. Prospective validation incorporating measured drug concentrations, Bayesian updating, and pathogen-specific MICs is essential before clinical implementation.