Abstract:
|
We present an adaptive sequential design for seamless phase 2/3 combination. The design includes a phase 2 stage and a phase 3 stage. Multiple doses, or endpoints, or sub-populations can be evaluated in the phase 2 stage. At the end of the phase 2 stage, one dose (together with control), or one sub-population, or an endpoint can be selected to proceed to the phase 3 stage. The phase 3 stage can include interim analyses. Data from both stages will be used for the final statistical analysis. Sample size can be modified at any interim analysis. The selection rule is flexible: either play the winner, or incorporating other considerations. Type I error is exact if the selection rule is to play the winner. otherwise, the type I error control is conservative. Medium conservative parameter estimates will be provided.
|