Predictive performance of a gentamicin population pharmacokinetic model in two western populations of critically ill patients

Laura H. Bukkems, Claire Roger, Caspar J. Hodiamont, Jean-Yves Lefrant, Nicole P. Juffermans, Jason A. Roberts, Reinier M. van Hest

Research output: Contribution to journalArticleAcademicpeer-review

12 Citations (Scopus)

Abstract

External validation of population pharmacokinetic (PK) models is warranted before they can be clinically applied to aid in antibiotic dose selection. The primary objective of this study was to assess the predictive performance of a gentamicin population PK model in intensive care unit (ICU) patients in two independent western populations of critically ill patients. Methods: Data were collected from the ICU where the model was developed (Academic Medical Centre, Amsterdam [AMC]) and from the Centre Hospitalier Universitaire de Nîmes (CHU Nîmes). Primary endpoints were bias and accuracy. The model was regarded as valid if bias was not significantly different from 0 and accuracy was equal to or less than 2.5 mg/L. Non-linear mixed-effects modelling (NONMEM) was used for data analysis. Results: The AMC validation dataset consisted of 192 samples from 66 ICU patients and the CHU Nîmes dataset of 230 gentamicin samples from 50 ICU patients. The structural model predicted the gentamicin plasma concentrations in the AMC population with a non-significant bias (0.35, 95%CI: -0.11–0.81) and a sufficient accuracy of 2.5 mg/L (95%CI: 2.3–2.8). The gentamicin plasma concentrations were overpredicted in the CHU Nîmes population with a significant bias of 4.8 mg/L (95%CI: 4.00–5.62) and an accuracy of 5.5 mg/L (95%CI: 4.7–6.2). Conclusion: The model is valid for use in the AMC ICU population but not in the CHU Nîmes ICU population. This illustrates that caution is needed when using a population PK model in an external population.
Original languageEnglish
Pages (from-to)218-225
JournalInternational Journal of Antimicrobial Agents
Volume52
Issue number2
DOIs
Publication statusPublished - 2018

Cite this