Bayesian individualization of pharmacokinetics: simple implementation and comparison with non-Bayesian methods

Abstract

One may attempt to individualize drug dosage by estimating an individual's pharmacokinetic parameters. Information useful for this purpose consists of certain population pharmacokinetic parameters (notably those describing the typical relationship between dosage and drug concentrations) and also measured drug concentrations from the individual of concern. Both types of information should be used. A (Bayesian) method that does so has been described in the pharmacokinetic literature. In this report an implementation of the Bayesian method that is readily adapted to a microcomputer is presented. Using simulated data it is compared with two other methods proposed by others, for estimating individual theophylline clearances. Both previously suggested methods are shown to be less precise than the Bayesian method: their typical error magnitudes are 20-70% larger.