Vancomycin levels for Bayesian dose-optimization in critical care: a prospective cohort study

Area of Science:

• Pharmacokinetics and Pharmacodynamics
• Critical Care Medicine
• Infectious Diseases

Background:

• Optimizing vancomycin dosing in critically ill patients is crucial for effective treatment.
• Therapeutic drug monitoring, specifically area under the concentration-time curve/minimum inhibitory concentration (AUC/MIC) ratio, is essential.
• The precise number of vancomycin levels (VLs) required for accurate AUC/MIC estimation remains unclear.

Purpose of the Study:

• To determine the minimum number of vancomycin levels (VLs) needed for accurate AUC/MIC estimation.
• To evaluate the accuracy and bias of different vancomycin level sampling strategies.
• To optimize vancomycin dosing protocols in critically ill populations.

Main Methods:

• A prospective cohort study involving 36 critically ill adult patients.
• Collected vancomycin levels (VLs) at peak, beta, and trough phases.
• Utilized PrecisePK™ Bayesian software to derive five different AUC estimates, comparing them against a reference AUC calculated via the trapezoidal model.

Main Results:

• The AUC estimate using two vancomycin levels (peak and trough) demonstrated higher accuracy and lower bias compared to estimates using fewer or only prior data.
• AUC-3 (peak, trough) showed significantly better accuracy (p=0.042) and lower bias (p=0.036) than AUC-4 (trough) and AUC-5 (prior only).
• Bland-Altman analysis indicated superior agreement between AUC-3 and AUC-2 (beta, trough) with the reference AUC.

Conclusions:

• Bayesian software utilizing two vancomycin levels (VLs) provides a more accurate and less biased AUC/MIC estimation in critically ill patients.
• A strategy employing peak and trough vancomycin levels is sufficient for reliable AUC/MIC estimation.
• This finding can simplify therapeutic drug monitoring for vancomycin in intensive care settings.