|
Abstract:
|
Randomized discontinuation trials may be viewed as a special class of population enrichment trials. They usually consist of two-three stages and their major objective is the evaluation of placebo effects and the selection of a sub-population that respond to an experimental treatment. The statistically valid population partitioning to placebo-responder, non-responder and treatment-responder subgroups paves a road to precision medicine. The first randomized discontinuation trials were used exclusively to test long-term, non-curative therapies for chronic or slow evolving diseases. Development of stochastic longitudinal models combined with the Bayesian ideas and increase of computing power led to their acceptance in various therapeutic areas, most notably in oncology. In this presentation, the Amery-Dony designs and in particular their new symmetric versions are compared with the classical randomized trial designs. It is shown that for binary end-points and in the absence of false classifications the symmetric Amery-Dony designs are always superior to the classical randomized trial designs both statistically and in the sense of subject exposure to the placebo treatment.
|
