When it comes to multiple endpoints (as opposed to subgroups), PR’s tend to more honest and straightforward about the statistics. For instance, the recent PR from MYOG (#msg-10603310) said:
>> Other secondary endpoints had clinically relevant improvements that achieved p values of less than 0.05 but were not considered statistically significant due to the pre-specified approach for multiple comparisons. <<
In other words, these analyses had p>=0.05/k, where k is defined as in the cited article from NEJM.
Market Data 
Markets 
AI
