I thought the "big issue" with the FDA on NI trials was that you MUST establish that the drug has some effect. If the NI margin was such that a placebo has a good chance of passing, you have a real problem.
Therfore the delta (agreed to in FDA backrooms) must, in some way, be based on historical placebo (or SOC) data.
Let's take this trial as an example. If sDVT occurs at a rate of 30% w/o AT, then it's obvious that a NI delta of 10-15% is "reasonable". BUT if sDVT only occurs at a rate of 5%, then a NI delta over 3-4% is absurd.
I have no problem with your statement that the only thing that matters is what the sponsor and FDA agree on. I have 0 clue on that process vs. a SPA.