Predictive performance of Sawchuk-Zaske and Bayesian dosing methods for tobramycin

Area of Science:

• Pharmacokinetics and Pharmacodynamics
• Infectious Diseases
• Clinical Pharmacy

Background:

• Accurate prediction of drug concentrations is crucial for optimizing therapy and minimizing toxicity.
• Tobramycin, an aminoglycoside antibiotic, requires careful dosing to achieve therapeutic efficacy against gram-negative infections while avoiding nephrotoxicity and ototoxicity.

Purpose of the Study:

• To compare the reliability of the Sawchuk-Zaske method and a Bayesian method for predicting steady-state tobramycin concentrations.
• To assess the impact of using only trough concentrations on the Bayesian method's predictive performance.
• To evaluate the accuracy of both methods in predicting peak and trough tobramycin concentrations on days 4 and 10 of therapy.

Main Methods:

• A prospective study involving 30 patients treated for gram-negative infections.
• Pharmacokinetic parameter estimation using the Sawchuk-Zaske method and a Bayesian approach.
• Comparison of predicted versus observed tobramycin concentrations (peak and trough) at specific time points.

Main Results:

• Both methods predicted steady-state trough concentrations with comparable accuracy and bias.
• The Bayesian method, when using limited data (e.g., trough concentrations only), tended to overpredict peak tobramycin concentrations.
• Significant differences were observed in the estimation of tobramycin volume of distribution between the Bayesian and Sawchuk-Zaske methods.

Conclusions:

• The Bayesian method demonstrates potential utility for providing aminoglycoside dosing recommendations.
• While accurate for trough levels, the Bayesian method requires careful consideration when predicting peak concentrations, especially with limited data.
• Further research may refine Bayesian approaches for more precise tobramycin dosing strategies.