There are ways to control for multiplicity. I think that in the case were are discussing that is why the p requirement was tightened up from p<0.05 to p<0.025 for both the primary and secondary endpoint.
Having never seen a SAP I'll take her word for it.
Multiplicity is mostly aimed at ad hoc analysis. If the alternate analysis is specified in the SAP it is not an ad hoc analysis. However, as I pointed out above, the p requirement is tightened up in the alternative analysis.
FWIW, it is not true that "the first-listed" measurements are the ones given more weight. There may be several primary endpoints and they don't rank above one another regardless of the order in which they appear in the statistical analysis plan. Of course, the secondary endpoints are considered less important than the primary endpoints. Your friend may have been simply referring to primary v secondary and it just didn't communicate clearly for whatever reason. (Hopefully my understanding of what you wrote was accurate - it may not have been.)
News 
Market Data 
Discover 
