Mixed models for ordinal data: a pharmacokinetic study on the effectiveness of drug for the reduction of epileptic seizures.

Population pharmacokinetics (PK)/pharmacodynamics analysis is considered as an important component of drug development process in these days. Owing to different standardization techniques in different laboratories, the drug concentrations in blood are reported in intervals in which they lie. For obvious reason the outcomes are then recorded as multiple discrete indicators with natural ordering. But in view of the reduced amount of information, the modeling of such data becomes intrinsically more difficult than that of continuous data. The primary concern in this study is to analyze a one-compartmental PK model for ordinal data that can be used to describe the change in the concentration of drug over time. A flexible Monte Carlo EM-based two-step estimation method is proposed. Such a method is, in general, capable of analyzing a reasonably wide class of nonlinear mixed models. The modeling strategy is applied to a Phase I study of the drug Divalproex and its metabolite, developed for the treatment of epilepsy. The model's performance is evaluated through simulation.