Bioequivalence and Bioavailability Forum • PAR not yet

PAR not yet [Two-Stage / GS Designs]

posted by ElMaestro – Denmark, 2011-10-27 15:25 (5350 d 16:52 ago) – Posting: # 7554
Views: 18,185

Hi tOT,

❝ “Method C” as proposed in this publication is not acceptable, as type I error is not controlled on the 5% level and alternative methods securing type I error control are available.

❝

❝ They go on to suggest method B because "The authors demonstrate by simulations that the power of this method is only slightly lower than for “Method C”, resulting in comparable mean total number of subjects."

❝

❝ Any comments?

Difficult.
It's their call, really. To distinguish between the methods in practice I think one needs:

To decide which T/R is realistic. 0.95 is a guess but do you have any idea really apart from dissolution?
You proposed number of subjects in stage 1 (n1).
Futility vs. power at the given T/R and n1 for the entire range of CVs.

All in all that's just not straightforward. Especially not if the true T/R differs from the expectation. That blows the entire approach. Wouldn't it be wonderful to use some kinda meffud that takes te observed T/R from stage 1 into account *cough*cough*?

Sounds to me like you should use method B if you wish to avoid discussion and if in vivo T/R really is 0.95.

—
Pass or fail!
ElMaestro

Complete thread:

Two-Stage Sequential Designs Helmut 2010-05-20 16:53
- Two-Stage Sequential Designs yjlee168 2010-05-20 21:28
  - Two-Stage Sequential Designs Helmut 2010-05-21 01:06
    - EMA: Two-Stage Sequential Designs d_labes 2010-06-15 16:02
      - EMA: Two-Stage Sequential Designs Helmut 2010-06-15 16:23
        
        EMA: Two-Stage Sequential Designs The Outlaw Torn 2011-10-20 10:30
        
        EMA: Two-Stage Sequential Designs Helmut 2011-10-21 15:01
        
        PAR? ElMaestro 2011-10-21 15:23
        
        PAR not yet Helmut 2011-10-21 16:02
        
        PAR not yet The Outlaw Torn 2011-10-25 08:57
        
        PAR not yet The Outlaw Torn 2011-10-27 11:07
        
        Potvin C not acceptable! d_labes 2011-10-27 13:16
        PAR not yetElMaestro 2011-10-27 13:25
        
        PAR not yet The Outlaw Torn 2011-10-31 07:45
        
        No (?) comment :-( Helmut 2011-10-31 15:04