Havasu, Trying to answer your questions: How
Post# of 155356
(Total Views: 641)
Posted On: 10/25/2020 11:34:13 AM
Havasu,
Trying to answer your questions:
Please find below a good article that will get you started with the calculations (some examples provided) and why power is so important.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3148614/
Normally FDA takes more than 80%-90% as adequate.
The power is proportional to the number of patients in the trial. In the original design it is deemed that with 390 patients we will have enough power to get significance together with the p number. During the interim (1/2 of patients) normally this is NOT achieved (that is why the p-value has to be much lower than the end-of-trial one for stoppage). When we go to 2/3 full enrollment this (power) will be higher (and, hopefully, p-value lower
).
Please refer to the document below published by FDA.
https://www.fda.gov/media/78495/download
In general, interim analysis should be pre-specified, however there is flexibility to change the number of interim analysis (under the proper objectives). Also, if the data is unblinded there is a penalty to be paid every time.
Trying to answer your questions:
Quote:
How is the power calculated and what value is needed? All things being equal, how does the power change from 70% at 50% interim to 75% interim?
Please find below a good article that will get you started with the calculations (some examples provided) and why power is so important.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3148614/
Normally FDA takes more than 80%-90% as adequate.
The power is proportional to the number of patients in the trial. In the original design it is deemed that with 390 patients we will have enough power to get significance together with the p number. During the interim (1/2 of patients) normally this is NOT achieved (that is why the p-value has to be much lower than the end-of-trial one for stoppage). When we go to 2/3 full enrollment this (power) will be higher (and, hopefully, p-value lower
). Quote:
But most importantly, upon what authority can the DSMC authorize another interim analysis? Do they just do an interim analysis whenever they want? I thought that interim analyses and to be agreed upon with the FDA before the trial begins.
Please refer to the document below published by FDA.
https://www.fda.gov/media/78495/download
In general, interim analysis should be pre-specified, however there is flexibility to change the number of interim analysis (under the proper objectives). Also, if the data is unblinded there is a penalty to be paid every time.
(4)
(0)