Sample size calculations for clinical trials with an adaptive design

When designing a clinical trial one of the most important steps is the estimation of how many patients will be required for it to achieve its objectives. This is an important step for planning as the estimate of the number of patients feeds into the costs and duration of the study being designed. An accurate estimate of the sample size is also important for ethical reasons. If the estimate is wrong then it could mean too few patients or too many patients are being enrolled to answer the scientific question of the study.
Recently a development in clinical trials has been the application of adaptive designs. A traditional fixed design will recruit until the estimated sample size is obtained when it is stopped and the results of the study analysed. In an adaptive design the results of the study are analysed in an ongoing basis with decisions made to continue, curtail or amend the trial design.
The opportunity to stop the trial early with an adaptive design is seen as a main reason for their adoption. This means few patients will be enrolled into the trial, decisions are made quicker and the costs of the research are reduced. The hope therefore when planning an adaptive design is that the sample size will be smaller.
Given that the a reason to undertake an adaptive design is to reduce the sample size when planning a study the sample size calculation in the protocol will usually be bigger than that for a fixed trial design. This is because a maximum sample size is reported. The maximum sample size is the sample size for the study to end recruitment if it has not stopped at an interim analysis stage.
So a study design is proposed to reduce time, costs and patients recruited which paradoxically seems to increase time, costs and patients recruited.
This PhD will review sample size calculations in adaptive designs. It will start by reviewing the sample size calculations in protocols for adaptive designs and how they are reported. Is it appropriate to quote the maximum sample size? Is there a better sample size to plan the study on?
The PhD will then review sample size calculations for adaptive designs and make recommendations as to their application.
The PhD although methodological will be a highly applied project anchored in a real world practical problem