What I think is that the SPA has a very high interim hurdle. Higher than "Standard" OBF. And it makes sense considering that Merck wants to use this single Phase III trial to file for approval. And that they only would want a trial stopped if there were overwhelming efficacy.
Kirkman has said on multiple occasions that OBF is being used as the alpha spend function. But he admittedly has said he has difficulty on the statistics side (which was reinforced on the earnings call when he initially said he didn't understand a question on what HR range was now achievable and its clinical relevance - granted it wasn't phrased the best but still interpretable). Also Kirkman gave out enough info on trial design (90% powered 6 mo MST advantage over control arm MST of 20 mos, HR .76-.77, and overall p value .025) to know that Merck designed START with .025 one sided under assumption of PH.
So when he says that the alphas at the 1st and 2nd interims are "far more stringent" than that allocated for the final, I think ones has to filter what he's saying. Under OBF rules .0015 and .009 are "far more stringent" in Kirkman speak than the .022 for the final, but not some ultra stringent .001 OBF derived boundaries for just the interim looks then reverting to .025 at the final (which wouldn't be OBF anyways).
Think about it -- and this goes to iwfal's point about the real value of an HR of .8 or more. With the massive overpowering in this trial, standard OBF (p < .009 or so) would translate to an HR boundary of about .8 or so at the 2nd interim? You want the trial to stop at that level? To get approval on a single Phase III trial?
I'm not following the logic. Are you saying Merck didn't want to stop at the 2nd interim if the HR is .8 or greater because it's not clinically meaningful, but that it would be OK for the final with just an improvement in p value due to higher n?
AI
