A novel multi-arm response adaptive allocation design for multivariate ordinal outcome clinical trials

An innovative and presumably the first of its kind response adaptive design is developed for multi-armed clinical trial producing multivariate ordinal outcomes. The concept of stochastic ordering is extended suitably and to develop a relevant allocation function. Properties of the proposed design are studied both theoretically as well as empirically and practical applicability is further established through redesigning a real clinical trial, where the response is bivariate categorical. It has been found that the proposed design is able to maintain the individual ethics without sacrificing the collective ethics significantly compared to the complete randomization procedure. In other words, the proposed design has the potential to assign more patients to the better performing treatment(s) in an ongoing trial with a good discriminating power to detect the differences among the treatment effects.