Don't know about "desperate", but it does not inspire confidence.
Lowering the bar of p-values - Seems unlikely, with the possible exception of trials where the patient population is tiny.
Subgroup approval - I could only see this happen if subgroups were defined in advance and part of the original protocol. Allowing retrospective subgroups opens the door to all sorts of evil data mining. Run a large enough trial and you could always find something.
News 
Market Data 
Discover 
