Practical antimicrobial TDM (PAT) of vancomycin in Japanese emergency and critical care units: Concentration overestimation and correction using scaling factors

Abstract

OBJECTIVES: i) To improve the accuracy of initial dosing design in vancomycin therapeutic drug monitoring (TDM) using a free web application of practical antimicrobial TDM (PAT) for vancomycin, ii) to identify patient subgroups in emergency and critical care settings with overestimated predicted population mean serum vancomycin concentrations (PRED), and iii) to calculate appropriate PRED scaling factors. MATERIALS AND METHODS: Vancomycin serum concentrations in patients treated at the Emergency and Critical Care Centers of the University of Miyazaki Hospital, Japan, were simulated using a population pharmacokinetic model. Patients were classified based on the ratio of observed-to-predicted serum vancomycin concentrations (OBS/PRED). Those with a ratio < 0.67 were assigned to the low-trough group, while those with values ≥ 0.67 were the control group. Changes in patient data in the period from admission to immediately before vancomycin administration were analyzed, and scaling factors minimizing the mean squared error (MSE) between PRED and OBS were determined for subgroups more common in the lowtrough group. MSE and root mean squared error (RMSE) values before and after PRED scaling were compared. RESULTS: A total of 47 patients were analyzed, of which 23 were assigned to the low-trough group and 24 to the control group. The low-trough group showed a significantly higher prevalence of patients with creatinine clearance (CCR) < 15 mL/min, albumin < -0.5 g/dL, and trauma. For these subgroups, PRED scaling factors ranged from 0.53 to 0.64. The optimal scaling factor was 0.53 for ΔCCR > 15 mL/min, reducing the RMSE from 5.1 to 4.0 (μg/mL). CONCLUSION: PRED scaling factors were derived to potentially improve the accuracy of initial vancomycin dosing using PAT, and ΔCCR > 15 mL/min, ΔALB < -0.5 g/dL, and trauma were identified as key factors.