Abstract:
|
Non-comparative two-stage designs based on a binary response variable are typically used in phase II clinical trials. In these designs the response rate of the experimental treatment is compared with a fixed target value that should represent the true response rate for the standard agent. Usually this target value is estimated by exploiting historical data, but no variability is taken into account. In this work we address one of the major concerns about the use of historical controls - i.e. the potential bias in the comparison - by proposing Bayesian single-arm two-stage designs, that allow to incorporate uncertainty in the response rate of the standard treatment. The two-stage sample sizes are selected by ensuring a large posterior probability of declaring the new treatment more effective than the standard one, under the assumption that the new treatment is indeed better in efficacy.
|