Replies to post #96779 on Cytodyn Inc (CYDY)

[bobshmob]

or, more simply, (11/28)/(8/56) = 2.75 => rate of SAEs in placebo arm was 2.75x that of the leronlimab arm. As you point out, that's huge.

Bigger than expected by random chance? Not easy to determine because the SAEs are not independent, but somewhere between 0.06 (using simple counts of patients with SAEs in the two groups via Mantel Hanszel test or logistic regression) and 0.024 assuming complete independence (binomial test on the distribution of SAEs between the groups).

Better would be to score the events and put them on a quantitative or ordinal scale, like is done for the primary outcome (but with measures of clinical status of patients). The ordinal scale will add power to the test and give a lower p-value. Also, I suspect that the leronlimab adverse events were early, before the drug fully kicked in, and that the patients had time to recover by day 14, when the primary outcome is complete. The placebo SAEs are much less likely to have recovered by then (I'm guessing).

The ordinal scale + the recovery time => p-value well below 0.06 for the primary outcome = huge success.