300 patients, 30 hospitals, 10 patients per hospital. Maybe 5...

300 patients, 30 hospitals, 10 patients per hospital.
Maybe 5 given treatment, 5 not as controls. Not sure uf better statistics this way to counter any inter hospital differences in treatment. Rather than totally random where potentially one hospital could have 8 of their 10 receive the active product and another have 7 or 9 controls. “Good” or “bad” hospitals or ICU units ??

So if 5 treated 5 control, results may end up with 5 patients getting better and 5 not.
4/ 5 of treatment group, 1/5 of control group. (Maybe actually 4.5 of treated group and 0.5 of control group averaged out.)
And hopefully 4 (4.5) of the 5 who get better get much better, compared with the 1 (0.5) from the control group.

And each hospital will only have 5 or 6 patients who get better, which will not appear to be too dramatic and might help keep the lid on the results for a while. Especially if they only have 3 patients each month over the 3 months of the trial. And presumably those 10 trial patients are randomised out double blind over the 3 months eithin each hospital.
So don't expect any leaks or good C-19 trial news early if I am correct in my suppositions.