DS maker relied on post hoc analyses, which the FTC disputed as being stat sig - which is why AMRN having 30+ SE's as official data points such a great idea, part of the trial design, no ad hoc controversy over planned subgroup analysis (no cherry picking data).