ABSTRACT
In clinical trials, power analysis for sample size calculation is often performed under a valid study design for achieving the intended study objectives with a desired power at a pre-specified level of significance. As discussed in previous chapters, most adaptive trial designs are considered less well-understood designs at which valid statistical methods for assessment of treatment effects might have not yet been developed. In this case, explicit formulas for sample size calculation and power estimation are not trackable. However, required sample size for achieving the desired power can still be obtained through clinical trial simulation. In this chapter, we will develop a set of algorithms for K-stage group sequential design (which is considered a well-understood adaptive trial design) based on well-established statistical methods such as the method of the sum of p-values (MSP), the method of the product of p-values (MPP), and the method of inverse-normal p-valuse (MINP). For other less well-understood adaptive designs, sample size required for achieving the desired power at a prespecified level of significance can be obtained through clinical trial simulation under some framework with clearly specified model and parameter assumptions.
