Use of aminoglycosides in critically ill patients: individualization of dosage using Bayesian statistics and pharmacokinetic principles.

Dosage prediction methods for optimization of aminoglycoside therapy were evaluated in 19 critically ill patients. Two different three-point methods and a novel Bayesian method were used. The predictive accuracy and precision of all tested methods were clinically satisfactory. The time needed for the necessary calculations was 10-15 min. After administration of predicted dosages, approximately 90% of the observed peak and trough aminoglycoside serum concentrations fell within the therapeutic range. Without systematic use of these prediction methods, only approximately 40% of tobramycin peak serum concentrations and 80% of trough serum concentrations were within the therapeutic range. Daily dosage adjustments particularly in patients with rapid changes of aminoglycoside clearance are greatly facilitated by use of these prediction methods. These findings suggest that all of the dosage prediction methods tested are well suited for optimization of aminoglycoside dosage regimens.