Evaluation of a Bayesian method of amikacin dosing in intensive care unit patients with normal or impaired renal function.

Our study was designed to determine the population pharmacokinetic parameters of amikacin in intensive care unit patients and to develop a Bayesian method allowing individual estimation of pharmacokinetic parameters. A two-stage method was used for estimating the population characteristics of the pharmacokinetic parameters. Calculations of optimum doses and dosing intervals were based on individual parameters. Our results indicate that the Bayesian method is capable of estimating the individual pharmacokinetic parameters with no significant bias and good precision. Individualization of amikacin dosage was assessed 70 times in 52 patients. To determine the predictive performance of the method, observed peak and trough levels were compared with predicted values by computing precision, bias, and correlation. The amikacin dosing method was unbiased and showed a high correlation coefficient (r = 0.962) between measured and predicted drug serum concentrations. No significant differences were found between the predicted and observed peak (17.3 +/- 3.5 and 17.3 +/- 3.8 micrograms/ml, respectively) and trough (2.86 +/- 0.93 and 3.08 +/- 1.41 micrograms/ml, respectively) amikacin serum concentrations. Among the 52 patients, wide variations were observed in the pharmacokinetic parameters (Vd = 0.21-0.50 L/kg; t 1/2 = 1.1-22 h) and the daily doses (2.8-42 mg/kg/day).