Bad luck [Study As­sess­ment]

posted by Helmut Homepage – Vienna, Austria, 2018-08-01 13:37  – Posting: # 19122
Views: 1,861

Hi Mikalai,

» […] Unfortunately, results one of analytes, a parent drug, fell sightly outside of the 80-125% range, below 80% for Cmax.

Bad luck. If you planned the study for assumed CV, T/R-ratio, and a dropout-rate for a desired power π the chance that you fail to demonstrate (if all assumptions turn out to be correct) is β = 1 – π. In other words, if you aimed at 80% power, 15 of studies of products which are BE will fail by pure chance. That’s part of the game.

» […] and post-hoc power was less than 80% for this analyte.

Although post-hoc power is irrelevant, if you planned for 80% and fail to show BE it will be <80%. As expected.

» Are there any ways we can defend bioequivalence without conducting another BE trail?

I don’t know how your agency deals with that. For the EMA and the FDA chances are close to nil.

» All other analytes fell within 80-125%.

Nice, but you have to demonstrate BE for all.

» Specifically, what should we put in the protocol …

You can’t change the protocol once the study is done, right?

» … or in a supporting letter to justify our lower border of CI around 79% for Cmax?

I don’t see how you could do that.
Example: Study planned for CV 25%, T/R-ratio 0.95, 80% power ⇒ n 28. All as expected, except a worse T/R 0.88  ⇒ 90% 78.66–98.45%. If you play around with α you will see that the patients’ risk will be 7.97% instead of ≤5%. I don’t think that any agency will accept that.

Cheers,
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

Complete thread:

Activity
 Admin contact
20,134 posts in 4,245 threads, 1,385 registered users;
online 17 (0 registered, 17 guests [including 5 identified bots]).
Forum time (Europe/Vienna): 01:40 CET

A Camel is a Horse designed by committee.    Anonymous

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