Pharmacokinetics of amikacin in intensive care unit patients: population analysis.
Silvia Romano, A.C. Falcao Mª. del Mar Fdz. de Gatta, Mª. Victoria Calvo, Mª. Eugenia Méndez, Alfonso Domínguez-Gil, José M. Lanao
Department of Pharmacy and Pharmaceutical Technology, University of Salamanca, Spain, Faculty of Chemistry Sciences. University Aut. of San Luis Potosi, México, University of Coimbra, Portugal, Pharmacy Service of University Hospital of Salamanca, Spain, Pharmacy Service of Rio Hortega Hospital, Valladolid, Spain
The aim of the present study was to analyse the amikacin pharmacokinetic behaviour in intensive care unit (ICU) patients by using a mixed-effect model and sparse data collected during routine clinical care.
The patient population was comprised of 120 medical ICU patients belonging to four different diagnostic groups: sepsis (N = 37), trauma (N = 26), pneumonia (N = 30) and other diagnosis (N = 27). A total of 565 amikacin serum levels were obtained in the whole population. A one-compartment model was used and the following clinical factors were tested for their influence on clearance (Cl) and volume of distribution (Vd): age, weight, creatinine clearance, duration of therapy and clinical diagnosis.
The nonlinear mixed-effect model (NONMEM) was used to assess the pharmacokinetic profile of amikacin in this patient population, an analysis of prediction errors in other ICU patients (N = 38) was carried out (population of validation) to verify the predictive performance of the final model.
In this study, the population final model accounting for amikacin pharmacokinetics in ICU patients was: Cl = 0.934 Clcr *(1 + 0.225 Trauma), Vd = 0.393 BW * (1 + 0.246 Sepsis) where Clcr and BW corresponds to patient’s creatinine clearance and body weight, respectively. The variables "Trauma" and "Sepsis" represents the clinical diagnosis of each patient.
This model was used to predict a total of 129 amikacin serum levels in the population of validation. The predictive performance was adequate for clinical purposes showing the feasibility of our population model to provide reference values used for "a priori" prediction as well as Bayesian approach for individualisation of amikacin therapy in ICU patients.