Replies to post #93989 on Biotech Values

[DewDiligence]

The M118 quiz answer comes directly from the Discussion section of the new paper in Circulation:

Importantly, because the event rate* in the control UFH arm was much higher than anticipated (31.1% versus the assumed 8%), the prespecified non-inferiority margin of an absolute 8% difference reflects substantially less than a doubling of the control group. The non-inferiority margin used in the present study, therefore, actually excludes a narrower relative difference than originally assumed.

In other words, meeting the prespecified NI margin of 8.0 percentage points — which the study accomplished for each M118 arm and the composite of all arms (#msg-48837450) — was much more difficult in a relative sense than the trial designers (and the FDA) anticipated.

*The event rate referred to above is the one for the primary endpoint, which was the set union of seven binary variables as shown in #msg-48837695.