Hypothesis testing and control of False positives has received a lot of attention in confirmatory clinical trials. However, corresponding solutions for the equally important and related problem of parameter estimation at the end of the adaptive trial have not been completely satisfactory. In this paper, we propose a new method to estimate the effect size of a treatment arm in a two stage design. This is based on a design that can include treatment selection at the end of the first stage and and confirmatory hypothesis tetsing in the second stage. We also extend this method to incorporate potential Sample Size Re-assessment at the time of the treatment selection.
Results show that compared to the naive CI calculations, this new method provides a very efficient estimate for the treatment effect for both the lower bound of the 95% Confidence interval and the median unbiased estimate of treatment effect. An illustrative example with a real-world clinical trial is also discussed.
|