Well, usually trials are only stopped at pre-speci
Post# of 155380
(Total Views: 877)
Posted On: 06/10/2020 5:52:56 AM
Well, usually trials are only stopped at pre-specified interim analyses for three reasons:
(1) Futility
(2) Safety concerns
(3) Overwhelming efficacy: i.e. it is impossible that if the trial was to be finished (390 subjects) that the result would become statistically insignificant. And with a p-value of just 0.03 or 0.05 the trial could still fail afterwards. The efficacy needs to be proven beyond reasonable doubt; a very low p-value is required for that.
Now, given the pandemic, if we get a good/barely stat sig result already at the interim analysis and hopefully succeed in the m/m trial at least on ome secondary endpoints, I'm hopeful that given our stellar safety profile, FDA will grant approval with the requirement to complete the study/ do a post-approval study.
So yes, the most likely outcome is that the s/c trial will continue without modifications: there usually is no definite "failure/success" at an interim analysis with such a low n. If we e.g. get only half the deaths in the leronlimab arm but only 20% mortality in the placebo arm, we don't get a stat sig result at n=51, but that doesn't mean we failed, since with 360 patients we would get p 0.00722 with a 50% mortality reduction.
Barring an extraordinarily high death-rate in the placebo group, we need 0 or only 1 death in the leronlimab arm, best case. Hoping for that best-case scenario!
It would be a great critical question for an analyst to ask at the CC: what is their rationale for an interim analysis at n=51 (vs. an analysis at 75 or 100), given that chances for an overwhelming success are marginal (out of statistical reasons simply due to the low sample size). They are very confident in leronlimab and so am I, and it seems they are thinking a 70-90% mortality reduction possible (?). Don't know.
(1) Futility
(2) Safety concerns
(3) Overwhelming efficacy: i.e. it is impossible that if the trial was to be finished (390 subjects) that the result would become statistically insignificant. And with a p-value of just 0.03 or 0.05 the trial could still fail afterwards. The efficacy needs to be proven beyond reasonable doubt; a very low p-value is required for that.
Now, given the pandemic, if we get a good/barely stat sig result already at the interim analysis and hopefully succeed in the m/m trial at least on ome secondary endpoints, I'm hopeful that given our stellar safety profile, FDA will grant approval with the requirement to complete the study/ do a post-approval study.
So yes, the most likely outcome is that the s/c trial will continue without modifications: there usually is no definite "failure/success" at an interim analysis with such a low n. If we e.g. get only half the deaths in the leronlimab arm but only 20% mortality in the placebo arm, we don't get a stat sig result at n=51, but that doesn't mean we failed, since with 360 patients we would get p 0.00722 with a 50% mortality reduction.
Barring an extraordinarily high death-rate in the placebo group, we need 0 or only 1 death in the leronlimab arm, best case. Hoping for that best-case scenario!
It would be a great critical question for an analyst to ask at the CC: what is their rationale for an interim analysis at n=51 (vs. an analysis at 75 or 100), given that chances for an overwhelming success are marginal (out of statistical reasons simply due to the low sample size). They are very confident in leronlimab and so am I, and it seems they are thinking a 70-90% mortality reduction possible (?). Don't know.
(4)
(0)