Mtruong,
I would be happy if the "p" price we pay is only 0.005 (0.05 vs 0.045). I believe Nader is misguided in this number, unless he meant 0.00495 (note the zeroes).
So, what is the issue??. Is well depicted in this article:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5052936/
I believe Ohm and CTMedic have already commented on several p-values commonly accepted because there is no consensus in the scientific community. And, obviously, it depends if the interim was planned or unplanned (in our case it was planned).
Our issue is that the number of patients is low (in comparison to other trials), so we will need to pay a larger penalty. One of the most used criteria is the so-called O′Brien-Fleming bounds that provide flexibility to the DSMB in terms of interim(s) spacing and number of patients involved.
This latter criteria would require a 0.0054 at interim look and 0.0492 at the end of trial. So, maybe Nader was referring to the END value but need to account what the interim would look like.
I personally think that we should evaluate how many deaths have occurred already (Nader/AMAREX know this as SAEs are reported immediately) , and based on this number make a decision.
Say, there have been 50 total deaths (25%) I would go for complete interim hoping for the majority being in the SOC arm. This number will give us a good "read" if LL works. If the number of total deaths is less than, say 30, I would not risk it and go the full length with the trial.
Of course, if AMAREX knows what is the interim p-value that will be analyzed for trial stoppage, they can make much more educated decisions in regards to the interim analysis. The above is pure speculations of my part without knowing anything in regards to the pre-agreed trial design.
I would be happy if the "p" price we pay is only 0.005 (0.05 vs 0.045). I believe Nader is misguided in this number, unless he meant 0.00495 (note the zeroes).
So, what is the issue??. Is well depicted in this article:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5052936/
Quote:Every time we look at the data and consider stopping, we introduce the chance of falsely rejecting the null hypothesis. In other words, every time we look at the data, we have the chance of a type 1 error. If we look at the data multiple times, and we use alpha of 0.05 as our criterion for significance, we have a 5% chance of stopping each time. Under the true null hypothesis and just 2 looks at the data, we “approximate” the error rates as: Probability stop at the first look: 0.05, probability stop at the second look: 0.95 × 0.05 = 0.0475, and total probability of stopping is 0.0975.
I believe Ohm and CTMedic have already commented on several p-values commonly accepted because there is no consensus in the scientific community. And, obviously, it depends if the interim was planned or unplanned (in our case it was planned).
Our issue is that the number of patients is low (in comparison to other trials), so we will need to pay a larger penalty. One of the most used criteria is the so-called O′Brien-Fleming bounds that provide flexibility to the DSMB in terms of interim(s) spacing and number of patients involved.
This latter criteria would require a 0.0054 at interim look and 0.0492 at the end of trial. So, maybe Nader was referring to the END value but need to account what the interim would look like.
I personally think that we should evaluate how many deaths have occurred already (Nader/AMAREX know this as SAEs are reported immediately) , and based on this number make a decision.
Say, there have been 50 total deaths (25%) I would go for complete interim hoping for the majority being in the SOC arm. This number will give us a good "read" if LL works. If the number of total deaths is less than, say 30, I would not risk it and go the full length with the trial.
Of course, if AMAREX knows what is the interim p-value that will be analyzed for trial stoppage, they can make much more educated decisions in regards to the interim analysis. The above is pure speculations of my part without knowing anything in regards to the pre-agreed trial design.
10
