As a blanket statement, the only time that such debate is likely to arise is if FDA, or sponsors (be they pharmaceutical firms or individual investigators), or both have erred during protocol development, protocol evaluation, or interpretation of data analysis. So, it is an exercise of futility. The context of such discussions needs to be explicitly understood. If the context is with respect to approvability, statistical vs clinical is not a relevant discussion – risk/benefit is the appropriate framework of reference and there is no statistic that addresses risk/benefit. If the context is with respect evaluation each of the components of a combined endpoint (the question being does one component that is not as meaningful as another drive the result of the combined endpoint), the problem is with the initial (and faulty) definition of the combined endpoint. More examples follow.