Repeats [Study As­sess­ment]

posted by ElMaestro  – Denmark, 2013-07-27 13:01 (4714 d 02:29 ago) – Posting: # 11068
Views: 25,256

Hi Sam,

❝ (...) the Fasting pivotal data has unexpected outcome. So we wanted to repeat the whole study (...)


This could be seen as bad and unethical science, and depending on where you submit for approval you may get a thumbs down.
You should not blindly repeat a study just because it had the 'wrong' outcome. There are situations where you can justify a repeat, notably:
  1. If you have a reason to believe the study conduct was flawed and results not trustworthy, then do an audit and investigate. If your audit truly confirms the presence of a problem then you may be able to repeat the study.

  2. If the study failed because of potential lack of power (wide CIs) then you may be able to repeat with a higher sample size. But this is a tricky matter because you will have to decide which PE to use for the calculation of sample size. If you observed a PE of e.g. 0.87 in the latest study, well then you might use that for the calculation and your sample size could go astronomical. After all, the PE from your latest study is a maximum likelihood estimate. Some companies will plug in a PE which is included in the latest observed CI but which is perhaps closer to unity than the latest PE. For example, if the CI was 0.72-1.01 then the PE was around 0.853, and it can be argued that the true PE might be 0.95 - at least there is not significance against that assumption. But plugging in 0.95 for sample size with 80% power or something like that although tempting is a risk of considerable magnitude; the apparent power of 80% is nice to report to management but is not likely to reflect the chance of approval.
Let's hear your numbers. What were your CIs, and what was the sample size?
—
Pass or fail!
ElMaestro

Complete thread:

UA Flag
Activity
 Admin contact
23,655 posts in 4,993 threads, 1,571 registered users;
140 visitors (0 registered, 140 guests [including 19 identified bots]).
Forum time: 15:30 CEST (Europe/Vienna)

Competence, like truth, beauty and contact lenses,
is in the eye of the beholder.    Laurence J. Peter

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5