EVALUATION OF AMINOGLYCOSIDE PHARMACOKINETICS IN POSTPARTUM PATIENTS USING BAYESIAN FORECASTING

Many postpartum women have suboptimal serum concentrations after stand ard doses of aminoglycosides. The purpose of this study was to charact erize the pharmacokinetics of aminoglycosides in postpartum patients t hrough the use of Bayesian forecasting and to test the ability of thes e subpopulation parameters to predict actual aminoglycoside serum conc entrations. In phase I, 28 postpartum patients who received empiric ge ntamicin therapy were identified and Bayesian subpopulation parameters generated, In phase II, additional gentamicin concentrations (peaks a nd troughs) were evaluated to test bias and precision of Bayesian subp opulation versus traditional estimates in predicting actual aminoglyco side serum concentrations, In phase I, 56 gentamicin serum concentrati ons in 28 patients, (age, 26 +/- 7 years; actual body weight [ABW], 84 .3 +/- 18.4 kg: ideal body weight [IBW], 54.6 +/- 5.1 kg; dosing weigh t [DW]. 66.3 +/- 9.4 kg: creatinine clearance [Cl-cr], 140.4 +/- 34.0 ml/min/1.73 m(2)), were evaluated to calculate subpopulation pharmacok inetic parameters of volume of distribution (Vd) 0.29 +/- 0.07 l/kg (D W); elimination rate constant (k(e)) 0.29 +/- 0.05 h(-1) and half-life (t(1/2)) 2.5 +/- 0.5 hours. In phase II, 50 gentamicin serum concentr ations in 25 patients (age, 23 +/- 4 years; ABW 79.4 +/- 17.5 kg; IBW 55.0 +/- 7.3 kg; DW 64.8 +/- 9.6 kg; Cl-cr 139.7 +/- 29.3 ml/min/1.73 m(2)) were evaluated to calculate subpopulation pharmacokinetic parame ters of Vd 0.30 +/- 0.04 l/kg (DW); k(e) 0.27 +/- 0.06 h(-1); and t(1/ 2) 2.9 +/- 0.8 hours. Predictive performance tests (95% confidence int ervals demonstrate that subpopulation postpartum Bayesian parameters s how greater precision for peak concentrations and less bias for trough concentrations than do traditional population estimates (p < 0.05). D efinition of the Bayesian subpopulation parameters will allow us to do se aminoglycosides optimally in postpartum patients who have fragmente d data.