Evaluating and Improving Neonatal Gentamicin Pharmacokinetic Models Using Aggregated Routine Clinical Care Data

Area of Science:

• Pharmacology
• Neonatal Medicine
• Pharmacometrics

Background:

• Model-informed precision dosing (MIPD) is crucial for drugs like gentamicin with high variability and narrow therapeutic windows.
• MIPD in neonates is complex due to rapid physiological changes and limited blood volume for sampling.

Purpose of the Study:

• To evaluate the performance of existing neonatal gentamicin population pharmacokinetic (PK) models for routine therapeutic drug monitoring.
• To assess the impact of refitting models with and without Bayesian priors on predictive accuracy.

Main Methods:

• Assessed six published neonatal gentamicin PK models using data from 475 patients across nine US sites.
• Refitted four validated models using Bayesian priors and compared performance against a validation dataset.

Main Results:

• Four out of six models demonstrated acceptable error and bias for clinical gentamicin dosing in neonates.
• Refitted models generally improved predictive accuracy, though the benefit of informative priors varied.
• Model performance was linked to the inclusion of key covariates and similarity to the development population.

Conclusions:

• Validated neonatal gentamicin PK models can reliably support MIPD in clinical practice.
• Continuous learning and model refitting can enhance dosing accuracy, with careful consideration of prior information.
• Findings provide guidance for clinicians and pharmacometricians implementing and developing MIPD strategies for neonatal gentamicin therapy.