|
Abstract:
|
The statutory requirement of substantial evidence for drug approval is interpreted as generally requiring two adequate and well-controlled clinical investigations and each convincing on its own. Often two Phase III studies are required and each needs to show significant treatment effect at the 1-sided significance level of 0.025. This level of evidence is equivalent to control the error rate of claiming an ineffective drug to be efficacious at the level of 0.000625. When each study consists of more than one endpoint, the multiplicity adjustment methods are often applied within individual studies. That is to control study-wise type I error. Such practices may end up with scenarios that each study is considered as positive but the treatment effect for each endpoint is not replicated. As the results of each endpoint are not replicated in both studies, the drug may not gain approval. A new method is proposed that controls the error rate at the level of 0.000625 and the multiplicity adjustment is applied at the compound level within an submission, without compromising the statutory requirement of the substantial evidence. The gain in power is also illustrated in this presentation.
|
