Discrepancies Between Bayesian Vancomycin Models Can Affect Clinical Decisions in the Critically Ill.

Area of Science:

• Pharmacokinetics and Pharmacodynamics
• Critical Care Medicine
• Infectious Diseases

Background:

• Vancomycin is a critical antibiotic for treating serious Gram-positive infections, especially in intensive care units (ICUs).
• Accurate therapeutic drug monitoring of vancomycin is essential for efficacy and safety, particularly in critically ill patients with altered pharmacokinetics.
• Population pharmacokinetic (PPK) models are widely used to estimate vancomycin exposure, such as the 24-hour area under the curve (AUC24).

Purpose of the Study:

• To evaluate the agreement of vancomycin AUC24 estimates derived from three commonly employed PPK models in critically ill patients.
• To determine the extent of concordance and discordance in AUC24 categorization (subtherapeutic, therapeutic, toxic) across these models.

Main Methods:

• A retrospective analysis of vancomycin serum concentrations from 188 critically ill patients was conducted.
• AUC24 values were calculated using the Tucuxi software with three distinct PPK models: Goti 2018, Colin 2019, and Thomson 2009.
• Agreement in AUC24 categorization between the models was assessed using percent agreement.

Main Results:

• A total of 466 AUC24 values were analyzed.
• Therapeutic AUC24 percentages varied by model: 52% (Goti), 42% (Colin), and 47% (Thomson).
• Overall agreement between all three models was 48%, with pairwise agreements ranging from 59% to 68%.

Conclusions:

• Significant discordance exists in vancomycin AUC24 estimates generated by different PPK models in critically ill populations.
• This variability may lead to inconsistent vancomycin dosing decisions and impact patient outcomes.
• The selection of an appropriate and validated pharmacokinetic model is paramount for optimizing vancomycin therapy in critical care settings.