Short answer:
yes it sounds fairly reasonable, but with the caveats in my post in play:
- we don't know the prior in use (which does affect the result a lot)
- we don't know the treatment effect rate, though I assumed here it's the same as the EUA data)
Loooong answer:
On the exact mortality rate for the control arm, it would have started fairly high in May and if it follows well the rates we see in other papers it would have been slowly decreasing., and we don't have a study that maps exactly to the MSB covid trial inclusion data, but the biggest factor for chance of death is age! So if we have a lot of old people in our trial then they might be dying at a much higher rate than the younger groups. The second factor is comorbitities (like Diabetes) and after the 45% readout the inclusion criteria was changed to exclude morbidly obese people - this could be something that emerged from literature that the trial sponsor wanted to include.
When we get the final stats, depending on how well it mimics the other white papers I've read, the mortality rate in the control arm could be as high as 80% or as low as 35%. It's a mistake to take an average, because most of the studies sit in the 35%-50% band, but MSB is targeting "worst of the worst" so I'd expect a control group mortality rate around 50% but I wouldn't be shocked if it was 35% or 80%.
For the table above, a prior is in use that I took from a white paper on Bayesian stats. If you actually adjust that prior a little then the output from the analysis changes a lot. So a lot depends on the prior distribution they agreed on at the start of the trial. You can think of this kind of like a benchmark that has to be surpassed in the interim analyses for overwhelming efficacy (though it's more like a boss in a video game than a benchmark). Thankfully most of the evidence I can see in white papers suggests that their prior should be somewhat like the one I used, if not exactly the same. It's a difficult "boss" to beat, and is likely to force the trial to completion unless our results are really really good.
Lastly, on the treated group we don't know if the people receiving rem-L are going to perform as good as, better or worse than the 12 person EUA trial. You would assume a little bit worse. However, as SOC has improved, maybe the treated group results will also improve a little bit, so this one I AM averaging out as being exactly like the 12 person trial (just for ease of modelling).
Bottom line:
Nothing in life is certain, but there's enough promising evidence that MSCs are effective against covid to be optimistic. For the 60% readout, buying options or whatever is a massive gamble, but if you're a LTH then the readout at the 60% mark and the 100% mark represent exciting moments full of possibility, but of course with no guarantee of success.
Good to see you DYOR on this though.... as the stock enters the news-heavy month of December, the amount of misinformation on this board is going to skyrocket up, and you have to know how to filter out the nefarious and/or stupid posts.
Expand
(20min delay)
