Vancomycin AUC₂₄ Estimation for Children with Bayesian Forecasting and First-Order Equations to Update Dose...

Objectives: This retrospective study aimed to estimate vancomycin area under the curve over 24 h (AUC₂₄) at steady state for pediatric patients from 1 month to 18 years to update the dose recommendations in a tertiary hospital. Methods: A total of 46 vancomycin treatments from 42 patients in the years 2014-2024 were analyzed. AUC₂₄ was estimated for the initial dose regimen using Bayesian forecasting and a trough concentration (n = 32) or first-order pharmacokinetic (PK) equations and a pair of practical peak and trough concentrations at steady state (n = 14). Three published population PK models (Colin, Kloprogge, and Le models) were compared using typical parameter values (a priori predictions) to find a suitable model for Bayesian estimation. For the Le model, the effect of converting plasma creatinine (PCr) measured with IDMS-traceable enzymatic assay to Jaffe method value was evaluated using published equations. Results: Le model without enzymatic-to-Jaffe PCr conversion and the Colin model were suitable for Bayesian forecasting. These models gave similar Bayesian AUC₂₄ estimates (a posteriori). In all 46 treatments, estimated AUC₂₄ was within the currently recommended range of 400-600 mg·h/L in 8 (17%) treatments, below this range in 36 (78%) treatments, and above this range in 2 (4%) treatments, respectively. Estimated AUC₂₄ was below 300 mg·h/L in 46% of treatments. Conclusions: Le and Colin models were suitable for our pediatric patients for Bayesian forecasting. AUC₂₄ estimates for the initial dose regimen were lower than commonly recommended. A more uniform dosing policy is needed in our hospital, and higher initial doses should be considered.