COVID ARDS interim analysis, page-6

Reduced mortality may be statistically significant at the initial readout with p<0.05 (less than 5% chance difference between treatment and placebo group is due to random chance). But, I wonder if overwhelming efficacy requires a P of <0.01, ie less than 1% chance difference due to random chance. As once you stop for overwhelming efficacy, you can effectively never trial versus a placebo group in the same setting. So if trial does not stop due to overwhelming efficacy, results may still be statistically significant at p<0.05 level. For non-medical readers p<0.05, is the number that researchers are comfortable indicates an actual difference between placebo and treatment groups.

So I would not expect the trial to be halted for overwhelming efficacy, even if results are indicative of significant reduction in mortality even with a small sample size.

Even is the trial shows marginal statistical significance over the whole trial group, as the study is watered down by patient's with less severe ARDS, having low mortality due to improved standard of care. I would expect the subgroup of more severe patient's to have a statistically significant mortality benefit. Similar to remestemcel-L in adults with acute graft versus host. This would suit nicely given the expense of the treatment, and may make FDA on label approval easier. Given cost to healthcare providers/ insurance companies, who may be more likely to push back against rapid on label approval for a larger pool of patients.