[Non-parametric estimation of pharmacokinetic parameters of amikacin in patients with non-insulin-dependent diabetes mellitus].

To establish a reference for MAP Bayesian adaptive control of amikacin therapy in non-insulin-dependent diabetic patients, 30 patients (age: 63.5 +/- 10.1 years) were studied. Weight (84.2 +/- 15.4 kg) and body mass index (28.0 +/- 4.3 kg/m2 for males and 30.5 +/- 6.4 kg/m2 for females) were stable during treatment. Creatinine clearance (CCr) was 70.3 +/- 27.2 ml/min/1.73 m2 before treatment and 69.6 +/- 24.3 ml/min/1.73 m2 (NS) at the end of treatment (2 to 15 days). 129 serum concentrations were drawn (4.8 +/- 2.6 levels per patient). The one-compartment model was parameterized as having Vs (l.kg-1) and Kslope (min/ml.h) for each unit of CCr (Kel = Kintercept + Kslope x CCr). The non-renal Kintercept was fixed at 0.00693 h-1. The NPEM computes the joint probability densities. The mean, median, and SD were respectively: Vs = 0.3574, 0.3654, 0.0825 l.kg-1; Kslope = 0.0026, 0.0027, 0.0007 min/ml.h. For the a priori first doses determination, precision is higher with the new population. No difference in adaptive control was observed. In additive, the full joint density probability should be used to develop stochastic multiple model linear quadratic (MMLQ) adaptive control strategies.