Abstract:
|
Popular analysis methods for phase II dose-ranging clinical trials include multiple comparisons among doses, modeling of dose response relationship, and a hybrid approach such as MCPMod (Bretz et al., 2005). These approaches serve well when the data are positive and resource (sample size) is adequate. Given resource constraints but with efficient computer power, Bayesian dose-response modeling method is attractive. The method not only links multiple doses and shares information among doses, but can also incorporate prior information from previous studies or experts' opinions, to obtain more powerful comparisons and more precise estimates. The information-rich era and an increasing trend of clinical data sharing by pharmaceutical companies such as SHARE (GSK, 2013), make it feasible to acquire relevant prior information for any clinical development that reaches Phase II. Therefore informative Bayesian priors can be obtained to help design and analyze Phase II dose-ranging clinical trials. We promote the Bayesian dose response modeling methods and share an example on how to obtain robust informative Bayesian priors using relevant historical data.
|