Optimal response-adaptive allocation designs in phase III clinical trials: Incorporating ethics in optimality

Optimal response-adaptive designs have been studied by various authors in the context for allocating patients among two competing treatments in a phase III clinical trial. However, depending on the variabilities of the responses, most of these designs may allocate a smaller number of patients to the better treatment. In the current work, we obtain an optimal target for a general class of responses maintaining a prefixed lower bound for the allocation to the better treatment, in addition to an ensured level of statistical precision. Various properties of the proposed procedure are investigated, and compared with the existing competitors. A data study is also included for the assessment of the proposed procedure in real situation.