Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Apr;64(4):585-598.
doi: 10.1007/s40262-025-01486-4. Epub 2025 Mar 18.

Leveraging Clinical Data to Enhance the Performance Evaluation of Ceftriaxone Population Pharmacokinetic Models in Children

Stef Schouwenburg  1   2 , Tim Preijers  3   4 , Alan Abdulla  3   4 , Enno D Wildschut  5 , Birgit C P Koch  3   4 , Matthijs de Hoog  5
Affiliations

Leveraging Clinical Data to Enhance the Performance Evaluation of Ceftriaxone Population Pharmacokinetic Models in Children

Stef Schouwenburg et al. Clin Pharmacokinet. 2025 Apr.

Abstract

Introduction: Sepsis affects approximately 8% of pediatric intensive care unit (PICU) admissions in high-income countries. Ceftriaxone, a broad-spectrum beta-lactam antibiotic, is widely used for treating severe infections and bacterial meningitis in children. Despite its frequent use, limited studies address the population pharmacokinetic (popPK) of ceftriaxone in pediatrics. External validation of popPK models is essential to confirm their suitability for individualized dosing in PICU patients, enabling selection of the model best suited to this population.

Methods: This study used data from the EXPAT Kids study, a prospective pharmacokinetics /pharmacodynamics (PK/PD) study. The included popPK models were implemented in NONMEM, with diagnostic goodness-of-fit and visual predictive check analyses performed to assess model accuracy. Predictive performance was evaluated using the relative prediction error, relative root mean square error, and mean (absolute) percentage error.

Results: The predictive performance of the evaluated models varied widely. The included models showed only modest performance and generally seemed to overpredict ceftriaxone concentrations. Unbound ceftriaxone popPK models did not perform adequately. None of the models met all the predefined thresholds for accuracy and precision.

Conclusions: Our external dataset comprised high ceftriaxone trough concentrations, indicating re-evaluation of current ceftriaxone dosing regimens to minimize the risk of overdosing and prevent toxicity. Future research should focus on the fine dosing balance for ceftriaxone, especially in patients with meningitis, by considering adequate exposure while preventing high trough concentrations. Model-informed precision dosing may enhance the use of the optimal individual dosage for critically ill children. However, our findings highlight the importance of externally evaluating ceftriaxone popPK models in the PICU population.

PubMed Disclaimer

Conflict of interest statement

Declarations. Funding: This work was supported by the MRace grant of the Erasmus University Medical Centre, Rotterdam, The Netherlands. Conflict of interest: Birgit C.P. Koch is an Editorial Board member of Clinical Pharmacokinetics. Birgit C.P. Koch was not involved in the selection of peer reviewers for the manuscript nor any of the subsequent editorial decisions. Stef Schouwenburg, Tim Preijers, Alan Abdulla , Enno D. Wildschut, and Matthijs de Hoog have no conflicts of interest to declare concerning this article. Ethics approval: The EXPAT Kids study was approved by the Erasmus MC Medical Ethical Committee (NL76194.078.21) (NL9326). Consent to participate: Written, informed consent was obtained from all participants. Consent for publication: Not applicable. Availability of data and material: The data and material are accessible upon reasonable request by contacting the corresponding author.. Code availability: The model codes that support the findings of this study are accessible upon reasonable request by contacting the corresponding author.

Figures

Fig. 1
Fig. 1
Total and unbound ceftriaxone concentrations versus time after dose administration. Ceftriaxone concentrations displayed on a log-scale; the dashed line depicts the ceftriaxone concentration of 100 mg/L to illustrate the trough concentration threshold for toxicity
Fig. 2
Fig. 2
Goodness of fit of predicted intravenous total ceftriaxone concentrations. Observed total ceftriaxone concentrations versus population predictions; the black line represents the line of identity and the blue line represents the locally estimated scatterplot smoothing (LOESS) line, which follows the highest density of the measured total ceftriaxone concentration and predictions
Fig. 3
Fig. 3
Goodness of fit of individual predicted total intravenous ceftriaxone concentrations. Observed total ceftriaxone concentrations versus individual predictions; the black line represents the line of identity and the blue line represents the LOESS line (locally estimated scatterplot smoothing), which follows the highest density of the measured total ceftriaxone concentration and predictions
Fig. 4
Fig. 4
Goodness of fit of predicted intravenous unbound ceftriaxone concentrations. Observed unbound ceftriaxone concentrations versus population predictions; the black line represents the line of identity and the blue line represents the LOESS line (locally estimated scatterplot smoothing), which follows the highest density of the observed unbound ceftriaxone concentration and predictions
Fig. 5
Fig. 5
Goodness of fit of individual predicted intravenous unbound ceftriaxone concentrations. Observed unbound ceftriaxone concentrations versus individual predictions; the black line represents the line of identity and the blue line represents the LOESS line (locally estimated scatterplot smoothing), which follows the highest density of the observed unbound ceftriaxone concentration and predictions
Fig. 6
Fig. 6
Visual predictive checks of the included popPK models for predicting intravenous total ceftriaxone concentrations (n = 1000). The solid black line represents the median of the measured total clavulanic acid concentrations. The dashed black lines represent the 10th and 90th percentiles of the total ceftriaxone observations. The dark blue shaded areas represent the interval for the median of the model predictions (with 97.5% confidence intervals for these predictions). The light-blue shaded areas represent the 10th and 90th percentiles for the model predictions (both with 97.5% confidence intervals)
Fig. 7
Fig. 7
Visual predictive checks of the included pharmacokinetic models for intravenous unbound ceftriaxone concentrations (n = 1000). The solid black line represents the median of the measured ceftriaxone concentrations. The dashed black lines represent the 10th and 90th percentiles of the ceftriaxone observations. The dark blue shaded areas represent the interval for the median of the model predictions (with 97.5% confidence intervals for these predictions). The light blue shaded areas represent the 10th and 90th percentiles for the model predictions (both with 97.5% confidence intervals). A Shows the VPCs for Hartman et al., stratified by route of administration, and B shows the VPCs for Tang Girdwood et al. model of unbound ceftriaxone concentrations (mg/L)
Fig. 8
Fig. 8
Comparison of the relative root mean square error (rRMSE), median percent error (MDPE), and mean absolute percent error (MAPE) for total ceftriaxone concentrations across the included popPK models. The orange barplots represent the performance metrics for individual predictions (IPRED) and the gray barplots represent the performance metrics for population predictions (PRED). The dashed lines represent criteria for acceptable model predictions, which are 20% for rRMSE, − 20 to 20% for MDPE, and 30% for MAPE
Fig. 9
Fig. 9
Comparison of the relative root mean square error (rRMSE), median prediction error (MDPE), and median absolute prediction error (MAPE) for unbound ceftriaxone concentrations across the included population pharmacokinetic models. The orange barplots represent the performance metrics for individual predictions (IPRED) and the gray barplots represent the performance metrics for population predictions (PRED). The dashed lines represent criteria for acceptable model predictions, which are 20% for rRMSE, − 20 to 20% for MDPE and 30% for MAPE
See this image and copyright information in PMC

References

    1. Weiss SL, Peters MJ, Alhazzani W, Agus MSD, Flori HR, Inwald DP, et al. Surviving sepsis campaign international guidelines for the management of septic shock and sepsis-associated organ dysfunction in children. Pediat Crit Care Med. 2020;21(2):e52–106. - PubMed
    1. van de Beek D, Cabellos C, Dzupova O, Esposito S, Klein M, Kloek AT, et al. ESCMID guideline: diagnosis and treatment of acute bacterial meningitis. Clin Microbiol Infect. 2016;22(Suppl 3):S37-62. - PubMed
    1. Hartman SJF, Upadhyay PJ, Hagedoorn NN, Mathôt RAA, Moll HA, van der Flier M, et al. current ceftriaxone dose recommendations are adequate for most critically ill children: results of a population pharmacokinetic modeling and simulation study. Clin Pharmacokinet. 2021;60(10):1361–72. - PMC - PubMed
    1. Iida S, Kawanishi T, Hayashi M. Indications for a ceftriaxone dosing regimen in Japanese paediatric patients using population pharmacokinetic/pharmacodynamic analysis and simulation. J Pharm Pharmacol. 2010;63(1):65–72. - PubMed
    1. Tang Girdwood S, Dong M, Tang P, Stoneman E, Jones R, Yunger T, et al. Population pharmacokinetic modeling of total and free ceftriaxone in critically ill children and young adults and Monte Carlo simulations support twice daily dosing for target attainment. Antimicrob Agents Chemother. 2022;66(1): e0142721. - PMC - PubMed

LinkOut - more resources