The widely-cited 88% figure (doi:10.1001/jama.2020.6775) is misleading and probably off by a substantial amount. In particular, the figure is based on 1151 severe/critical patients, of whom 252 died, 38 were discharged, and 831 remained in the hospital at the end of the study. The 88% is 252/(38 + 252), which ignores the 72% of patients whose cases were unresolved. The problem is that mortality tends to be fairly quick after hospitalization, so unresolved cases in studies like this tend to skew toward survival. In any case, the sample that the 88% figure is based on is a convenience sample with a 72% non-response rate that very likely introduces a large bias.
A later study at Emory (Auld et al 2020) that followed critically ill patients for longer found the mortality rate for was 67/217 = 30.9% vs. 131/217 = 60.4% discharged and 19/217 = 8.8% unresolved.
Another paper (Savel et al 2020) discuss further the widespread media-driven misperceptions of the mortality rate for severe/critical patients and give several further references to look at.
I'm betting we see 40% mortality for severe/critical in the control group. If we then have less than about 25% in the leronlimab group then, we are looking good!
I did a power analysis based on trial size and mortality rate for the control group, asking what size treatment effect would give a p-value of 0.05 with probability 0.8. If mortality rate is 0.4 for the control, we'd need leronlimab to have a mortality rate of 0.2 to get power of 0.8 with 51 patients. With full enrollment, leronlimab would only need to have a mortality rate of 0.345 to be reasonably sure of showing statistical significance.
You are absolutely right about the slight but real possibility of it being a two-tailed test, but hopefully not....
Your 65% guess on Placebo seems reasonable. Perhaps others here may provide additional insight. I think that anything higher than 50% for placebo would make it easier for Leronimab.