A 2-STEP APPROACH TO MEASUREMENT ERROR IN TIME-DEPENDENT COVARIATES IN NONLINEAR MIXED-EFFECTS MODELS, WITH APPLICATION TO IGF-I PHARMACOKINETICS

The usual approach to the analysis of population pharmacokinetic studi es is to represent the concentration-time data by a nonlinear mixed-ef fects model. Primary objectives are to characterize the pattern of dru g disposition in the population and to identify individual-specific co variants associated with pharmacokinetic behavior. We consider data fr om a study of insulin-like growth factor I (IGF-I) administered by int ravenous infusion to patients with severe head trauma. Failure to main tain steady-state levels of IGF-I was thought to be related to the tem poral pattern of several covariates measured in the study, and an anal ysis investigating this issue was of interest. Observations on these p otentially relevant covariates for each subject were made at time poin ts different from those at which IGF-I concentrations were determined; moreover, the covariates themselves were likely subject to measuremen t error. The usual approach to time-dependent covariates in population analysis is to invoke a simple interpolation scheme, such as carrying forward the most recent covariate value, ignoring measurement error; however, for these data the complicated observed covariate pattern mak es this approach suspect. A nonlinear mixed-effects model incorporatin g a model for time-dependent covariates measured with error is used to describe the IGF-I data, and fitting is accomplished by a two-step st rategy implemented using standard software. The performance of the met hod is evaluated via simulation.