Related Experiment Videos
Bayesian forecasting of serum vancomycin concentrations with non-steady-state sampling strategies
K A Rodvold1, J C Rotschafer, S S Gilliland
1Department of Pharmacy Practice, College of Pharmacy, University of Illinois at Chicago 60612.
Therapeutic Drug Monitoring
|February 1, 1994
Summary
Bayesian forecasting for vancomycin concentrations using non-steady-state samples offers minimal predictive value for future steady-state levels. Population pharmacokinetic parameters provide the most accurate predictions for both peak and trough concentrations.
Area of Science:
- Pharmacokinetics and Pharmacodynamics
- Therapeutic Drug Monitoring
- Bayesian Analysis
Background:
- Accurate therapeutic drug monitoring is crucial for optimizing antimicrobial efficacy and minimizing toxicity.
- Bayesian forecasting is a valuable tool for individualizing drug dosage regimens.
- Evaluating the utility of non-steady-state drug concentrations in forecasting is essential for refining clinical practice.
Purpose of the Study:
- To retrospectively evaluate the impact of three non-steady-state sampling strategies on the precision and bias of vancomycin pharmacokinetic predictions.
- To compare the effectiveness of fitting three versus five pharmacokinetic parameter estimates in a two-compartment Bayesian model.
- To determine if non-steady-state feedback concentrations improve predictions of future steady-state vancomycin concentrations.
Main Methods:
- Retrospective analysis of vancomycin concentrations in 27 adult patients with stable renal function.
- Application of three sampling strategies: single midpoint, peak and trough, and three serial concentrations.
- Utilized a two-compartment Bayesian forecasting program with three or five pharmacokinetic parameter estimates.
Main Results:
- Population-based parameter estimates yielded the most precise and least-biased predictions of steady-state peak vancomycin concentrations (ME = -0.40).
- Non-steady-state feedback concentrations did not significantly enhance predictions of future steady-state peak concentrations.
- A single midpoint non-steady-state concentration provided the least-biased prediction for steady-state trough concentrations.
- Fitting three parameters resulted in similar outcomes to fitting five parameters.
- Non-steady-state concentrations offered minimal additional information for Bayesian forecasting of future steady-state concentrations.
Conclusions:
- Non-steady-state vancomycin concentrations provide limited value for Bayesian forecasting of future steady-state concentrations.
- Population pharmacokinetic parameters are superior for predicting steady-state vancomycin peaks.
- Clinical decisions regarding vancomycin dosing should prioritize population data over non-steady-state sampling for forecasting purposes.