ABSTRACT
In dose-finding clinical trials, patients are typically assessed several times during the course of the trial. Longitudinal dose–response models use data at all timepoints to describe the time-changing effect of a treatment regimen. These models are more efficient than dose–response models that focus on data at one specific timepoint. However, longitudinal dose–response models are also more complex and diverse than dose–response models. We describe here the most advanced longitudinal dose–response models, the so-called dose–time–response models. These have mainly been developed by pharmacologists and are closely related to pharmacokinetic/pharmacodynamic (PK/PD) models. We also discuss the value of dose–time–response models for the design of clinical trials, in particular for the choice of dosing regimens. Two applications illustrate dose–time–response models: an analysis of a dose-ranging phase II trial, and a joint analysis of three phase II/III trials. A discussion summarizes the main advantages and challenges of longitudinal dose–response modeling.
