A Statistical Issue Regarding Nonlinear Least Squares Approaches with Observation Weighting for Pharmacokinetic Analysis

Abstract

Nonlinear least squares methods with observation weighting are often used for applying pharmacokinetic models to individual pharmacokinetic profiles. The observation weighting, however, causes biased and inconsistent estimates of pharmacokinetic parameters. This paper describes the statistical issue regarding nonlinear least squares methods with observation weighting, and presents a simulation study using a onecompartment model with bolus intravenous injection.