$CYDY @XW013K6, I posted this on YMB earlier: A study in JAMA from August 2020 (jamanetwork.com/abusenotice ) showed the national mortality rate for ICU patients was at or above 35%. Since we know that there were 45 deaths at the halfway point, applying that mortality rate would mean that 23/65 placebo patients died leaving 22/130 deaths in the treatment arm. If you assume an even split from that point forward between deaths in placebo and deaths in treatment arm, you end up with 44/130 in placebo arm and 43/260 in treatment arm. That leaves you with p<0.001. Treatment may have improved over the summer and the fall, so it is no longer 35% in placebo. Even if that is true, you would need for 12 of the 42 patients in the second half of the study to be moved from the placebo to the treatment arm in order for the study to exceed a p value of 0.05. That would mean mortality went from 17% to 25% in treatment group and from 35% to 14% in placebo group in the second half of the study. Not likely.
  • 10