Bayesian approach to control of amikacin serum concentrations in critically ill patients with sepsis

OBJECTIVE: To compare the predictive performance of a Bayesian program inco rporating a population model with and without severity of illness covariate s in intensive care unit (ICU) patients with sepsis. DESIGN: The clinical physiologic, and pharmacokinetic data of 62 patients w ith sepsis admitted to a tertiary-care center were analyzed retrospectively . The patients were randomly assigned to a active group and a validation gr oup. The model was developed using a three-step approach involving Bayesian estimation of pharmacokinetic parameters, selection of covariates by princ ipal component analysis, and final selection of covariates by stepwise mult iple linear regression. The predictive performance of this model was tested in patients from the validation group and compared with that of a general population model without covariates. RESULTS: Regression analysis revealed that the Acute Physiologic and Chroni c Health Evaluation (APACHE II) score was the most important determinant fo r amikacin volume of distribution (1.5L/kg, APACHE II; r(2) = 0.77). For am ikacin clearance (Cl-amik), creatinine clearance (Cl-cr), positive end-expi ratory pressure (PEEP), and use of catecholamines (CAT) were the most impor tant predictors (Cl-amik = 44.5 + 0.67 Cl-cr - 1.29 PEEP - 8.34 CAT, R2 = 0 .72). The relative mean error (Delta ME) and root mean-square error (Delta RMSE) (95% Cl) were -062 (- 1.2 to 0.01) and 3.78 (2.3 to 4.8) mg/L, respec tively. Since the 95% Cl for Delta RMSE did not include zero, it appears th at the model with covariates is significantly improved in terms of precisio n. CONCLUSIONS: Our results show that, in ICU patients with amikacin, it is re levant to consider covariates related to pathophysiologic status and therap eutic measures. Application of a Bayesian program allows improved control o f the pharmacokinetic parameters in patients who exhibit rapidly changing p hysiologic conditions.