I would say that if the trial took place/finished this past May, the mortality in the placebo might have been 25%, but the outcomes have been improving. The reason for such is multifactorial, partially due to various drug usage(dexamethasone etc) and also, ICU's aren't overwhelmed as often as early on. So, even if the difference in mortality isn't statistically significant, there still is hope with the results of the full trial...and possibly any of the secondary endpoints. What p value do we get if there were four deaths in the control and only one with LL?