|
Abstract:
|
Whereas previous works on adaptive design of clinical trials and mid-course sample size re-estimation have focused on conditional power to determine the second-stage sample size, we consider here a new approach that involves choosing the second-stage sample size by mimicking efficient fully-sequential tests using a flexible number of stages. Not only does this approach maintain the prescribed type I error, but it also provides an asymptotically efficient multistage test whose finite-sample performance, measured in terms of the expected sample size and power, is shown to be superior to existing conditional power and adaptive and non-adaptive group sequential designs.
|
