ElMaestro
★★★

Denmark,
2015-03-03 19:02
(3312 d 23:54 ago)

Posting: # 14537
Views: 6,275
 

 Pilot trials - what use are they really? [Study Per­for­mance]

Hi all,

I wonder when and how you experts use pilot trials?
If you have a candidate formulation and you have no clue if you are anywhere close to a match versus the reference -for example when it is a non-trivial formulation- would you then do a pilot trial? And if yes, how would you use the info derived from it??
E.g. your pilot trial on N=12 says the GMR estimate is 92.6 with a CI from 78% to 110%. Would you assume GMR=95% or would you assume 92.6% or something even worse when planning the pivotal?

This paper was just published and discusses such use of pilot trials. As far as I have understood (:-D) the author, who is known to be a fellow of rather dubious qualities, argues that when there is genuine uncertainty about the match (which is to say the true GMR) then a pilot trial is perhaps not of much use. This is because using the GMR for the planning of a pivotal trial isn't necessarily a great idea according to the numbers. Pilots appear to be excellent when the uncertainty is on the variability, but then 2stage approaches may be smarter still.

Do you have any idea how to use a pilot trial to address a potential lack of match :confused::confused::confused:?

Mucho complicado in my opinion and yet the question is so incredibly simple.

Pass or fail!
ElMaestro
Dr_Dan
★★  

Germany,
2015-03-04 10:26
(3312 d 08:30 ago)

@ ElMaestro
Posting: # 14547
Views: 5,062
 

 Pilot trials - what use are they really?

Hi ElMaestro
Honestly, I am no friend of pilot studies and of two stage designs either. IMHO a pilot study makes sense if you
1. use a complete different formulation (the situation you described = proof-of-concept)
2. have an alternative (three period design) or know how to reformulate (IVIVC)
3. can not find any PK data in literature.

❝ E.g. your pilot trial on N=12 says the GMR estimate is 92.6 with a CI from 78% to 110%. Would you assume GMR=95% or would you assume 92.6% or something even worse when planning the pivotal?

No risk, no fun. Do you have an alternative? IMHO a pilot study with only one test formulation does not make any sense and I agree that using the GMR for the planning of a pivotal trial isn't necessarily a great idea. I disagree with your opinion that pilots appear to be excellent when the uncertainty is on the variability, but then 2stage approaches may be smarter still. So if you do not know if variability is 35 or 45% does it matter if you use a replicate design? You can not use replicate in a two-stage.
Kind regards
Dr_Dan

Kind regards and have a nice day
Dr_Dan
jag009
★★★

NJ,
2015-03-04 18:14
(3312 d 00:42 ago)

@ ElMaestro
Posting: # 14550
Views: 4,967
 

 Pilot trials - what use are they really?

Hi Maestro,

Heheh, welcome to the crazy strategies of first-to-file generics. Some companies would make multiple formulations based on patents and run them again the marketed product right after it becomes available in hope that one of the tests will be a promising candidate... Yes I have a friend who ran bioanalytical 8 days after the RLD was available in the market.

❝ E.g. your pilot trial on N=12 says the GMR estimate is 92.6 with a CI from 78% to 110%. Would you assume GMR=95% or would you assume 92.6% or something even worse when planning the pivotal?


I would assume GMR of ~90 and power of 85%-90% based on the CV from the pilot, assuming the CV is far away from 30. I would not go below ~90 but then it's up to senior management who pushes my button :P

John
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,636 registered users;
87 visitors (0 registered, 87 guests [including 10 identified bots]).
Forum time: 18:57 CET (Europe/Vienna)

Nothing shows a lack of mathematical education more
than an overly precise calculation.    Carl Friedrich Gauß

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