Parameter Estimation of Population Pharmacokinetics of Tobramycin in Newborn Infants, Children, and Obese Children, and Its Application to Individualizing Optimal Therapy in Newborn Infants by the Bayesian Method

The population pharmacokinetic parameters of tobramycin were estimated by the computer program NONMEM in 14 newborn infants, 6 children, and 5 obese children with normal renal function. All of them were administered tobramycin with i. v. drip infusion for 30 min. Two models were used to obtain the population pharmacokinetic parameters: in the first model, it is assumed that volume of distribution and clearance are directly proportional to the body weight (linear model), and the other is exponentially related to the body weight (power model). The values of the objective function obtained by the population pharmacokinetic analysis for all of the patients indicated that the power model appeared significantly superior to the linear model. The results were used prospectively to forecast tobramycin concentrations in newborn infants by the Bayesian method. Firstly, the most suitable sampling time for predictive performance of the Bayesian method with single sampling point was determined by compar-ing various sampling times (4, 8, 12 hr). The predictions of serum concentrations using the single point of 12 hr after a single dose were best fitted to the observed data in 7 newborn infants. Secondly, the Bayesian method was applied to the dosage regimen during the multiple dosing in another 5 newborn infants using the population pharmacokinetic parameters estimated by the power model. The predicted serum concentrations of tobramycin were in good agreement with the observed data.