Y.Merlé1, J.M. Tré Luyer2 and G.Pons2
1 INSERM U436, 91 Bd De L'hopital 75634 Paris Cedex 13 France; 2 Service De Pharmacologie Perinatale Et Pediatrique Hopital Saint Vincent De Paul 74-82 Avenue Denfert-Rochereau 75674 Paris Cedex 14 France
For many drugs intravenously given to infants and neonates dilution procedures have to be performed before the infusion resulting in errors on the amount of drug really administered. In this work a statistical model for accounting more specifically these dose errors due to dilution procedures in the context of a population pharmacokinetic analysis of netilmicin is proposed. Therapeutic drug monitoring data were retrospectively collected from 186 neonates and 95 infants receiving netilmicin by short intravenous infusion. A one compartment linear model was used for describing the netilmicin pharmacokinetics from our data. An additive zero-mean normal error model was assumed for all measurements with a constant CV of 5 %. The amount of drug really present in the syringe was considered as the sum of the target value and of a random dose error arising from a zero mean normal distribution with a variance evaluated from experimental data and which was found to correspond to a constant CV of about 20 %. Only the dose error on the infusion immediately preceding the measurement was taken into account. The analysis was conducted from two learning samples by NPML and 9 covariates were included. NPML estimated population parameters were consistent with those previously obtained from parametric methods. Population and Bayesian predictions were very satisfactory for the neonate validation group and the mean and variance of the corresponding standardized prediction errors were near to 0 and 1 respectively. No trend was observed when plotting the latter versus the predicted concentrations. For the infant validation group the results were a bit less satisfactory in terms of population predictions. Most of the relationships between parameters and covariates detected were also consistent with those of previously published studies. Therefore our results validate our experimental error model and further studies are now needed to assess the impact of accounting for these dose errors due to dilution procedures in population PK analysis.
Reference: PAGE 8 (1999) Abstr 162 [www.page-meeting.org/?abstract=162]
Poster: poster