Abstract:
|
Examples are given of the design and analysis of multi-period studies, typically cross-overs, in early phase studies in the pharmaceutical industry. The aim is to explore the usefulness of the inclusion of the between-subject sequence term typical in the linear models used for analysis and the effect this has on design. Starting with the simple two-period, two-treatment, two-sequence design, various multi-period studies are introduced and alternative designs and analyses considered. The usefulness for inference of different designs and analyses using models with and without sequence are contrasted. It is hoped to be able to show that omitting the sequence term from analysis models allows a broader range of designs, more useful models, and can give more information with the same experimental resources.
|