An Urn Model and the Odds Ratio-Based Design for Clinical Trials
2008, v.14, Issue 4, 571-582
We study the limiting behaviour of a generalized Polya urn, motivated by adaptive data-dependent allocation designs, which are used in Phase III clinical trials in order to allocate a larger number of patients to the better treatment. We establish rigorous limiting results for the model, including the Central Limit Theorem, thus providing the theoretical background for using the odds ratio-based adaptive designs.
Keywords: adaptive designs,Central Limit Theorem,limiting allocation probability,odds ratio estimates,play-the-winner rule