Phase II clinical trials are typically designed as onearm studies based on a binary response variable, where the efficacy of the experimental therapy is compared with a target value, estimated through historical data. However, a very high failure rate at the end of phase III has been recently registered. It could be due to the use of historical controls. As a consequence several authors suggest the introduction of randomization also in phase II clinical trials. Our purpose is to compare one-arm and randomized designs, using a Bayesian approach. The comparison is based on the predictive ability of each design to declare the new treatment as promising, when it is supposed better than the standard one.
A Bayesian-predictive approach to compare one-arm and randomized design in phase II clinical trials
Domenicano I
Formal Analysis
2015
Abstract
Phase II clinical trials are typically designed as onearm studies based on a binary response variable, where the efficacy of the experimental therapy is compared with a target value, estimated through historical data. However, a very high failure rate at the end of phase III has been recently registered. It could be due to the use of historical controls. As a consequence several authors suggest the introduction of randomization also in phase II clinical trials. Our purpose is to compare one-arm and randomized designs, using a Bayesian approach. The comparison is based on the predictive ability of each design to declare the new treatment as promising, when it is supposed better than the standard one.I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.