Predictive performance of an antibiotic precision dosing software program in critically ill adults with infection

Area of Science:

• Pharmacokinetics and pharmacodynamics in critical care
• Antibiotic drug development and optimization
• Clinical pharmacology and therapeutic drug monitoring

Background:

• Critically ill patients have altered pharmacokinetics, complicating antibiotic dosing.
• Precision dosing software may enhance therapeutic antibiotic exposures.

Purpose of the Study:

• To evaluate the predictive performance of antibiotic precision dosing software.
• To compare a priori and a posteriori predictions for various antibiotics.

Main Methods:

• Utilized the ID-ODSTM precision dosing program with data from the GUIDE trial.
• Predicted and compared drug concentrations (piperacillin, meropenem, cefepime, flucloxacillin, vancomycin) against observed values.
• Assessed predictive performance using MDPE, MDAPE, F20, and F30 metrics.

Main Results:

• A priori prediction for beta-lactams failed acceptance criteria (MDPE -35%, MDAPE 58%).
• A posteriori prediction for beta-lactams met accuracy criteria, with cefepime meeting precision and F20/F30 criteria.
• A priori vancomycin prediction met all accuracy and precision criteria.
• Approximately one-third of a priori predictions suggested unnecessary dosing actions.

Conclusions:

• A priori vancomycin dosing prediction was acceptable in critically ill adults.
• A priori beta-lactam prediction was not acceptable; a posteriori methods improved performance but only cefepime met all criteria.