Bioequivalence and Bioavailability Forum • Problematic design

Problematic design [Design Issues]

posted by Helmut – Vienna, Austria, 2013-12-31 16:02 (4562 d 10:09 ago) – Posting: # 12120
Views: 22,471

Hi randomizers,

why do you want to perform such a design? Which is your “target” regulation? Do you want to market both T1 and T2 or are the “candidate formulations” and you will drop one of them? In the latter case I would suggest pilot studies and performing the pivotal with the “survivor” in a more conventional design.
If you want to submit the study to the FDA, keep RTR (“non-recommended in vivo study”) in mind!

To get a balanced design is rather difficult. The conditions are:

Each treatment administered once per subject.
Each treatment administered at the same frequency in each period.
Each treatment administered at the same frequency before other treatment.
Each potential comparison across periods at least once.

#1–#3 are recommended and #4 is desirable. None of your previous layouts come even close… The “best” I was able to come up with (acc. to Jones/Kenward 2003, othogonal latin squares in Table 4.3 rearranged) is:

R R T1 T2 R R T2 T1 R T1 R T2 R T2 R T1 R T1 T2 R R T2 T1 R T1 R R T2 T2 R R T1 T1 R T2 R T2 R T1 R T1 T2 R R T2 T1 R R

Condition #1 is fulfilled (ignoring the replication), so are #2 and #4. Condition #3 is not fulfilled.* If you know a better one, please let me know. IMHO, full balance could only be achieved if all 4! = 24 permutations are used. Since you don’t want to end up with zero degrees of freedom, that would mean 48 subjects. In their 2001 guidance the FDA argued against a 4-period 4-sequence fully replicated design (instead of the simple RTRT|TRTR). I’m afraid statistics is more demanding than clicking a button in SAS.
Which model/code do you intend to use?

If you are not interested in comparing the two tests, my design is balanced for the pairwise comparisons T1/R ∧ T2/R.

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

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

Complete thread:

Two Tests with Same Reference Noha Abdeljawad 2012-11-27 08:18 [Design Issues]
- Crossover designs with more than 2 treatments d_labes 2012-11-27 09:51
  - Crossover designs with more than 2 treatments Noha Abdeljawad 2012-11-29 07:43
- Sounds wrong to me ElMaestro 2012-11-27 12:10
  - Sounds wrong to me Dr_Dan 2012-11-27 14:32
  - Sounds wrong to me Noha Abdeljawad 2012-11-29 07:52
- Two Tests with Same Reference jag009 2012-11-28 21:26
  - FDA & multiplicity Helmut 2012-11-30 14:04
    - FDA & multiplicity jag009 2012-11-30 15:16
      - FDA & multiplicity Helmut 2012-11-30 16:12
        
        FDA & multiplicity jag009 2012-11-30 22:47
        
        FDA & multiplicity Helmut 2012-12-01 02:00
        
        FDA & multiplicity jag009 2012-12-03 21:53
    - Multiplicity: Mixed feelings ElMaestro 2012-11-30 15:24
      - Multiplicity: Mixed feelings Helmut 2012-11-30 15:52
        
        Multiplicity: Mixed feelings martin 2012-12-03 09:24
      - Multiplicity: Mixed feelings kumarnaidu 2013-12-12 11:03
        
        More infos, please Helmut 2013-12-12 15:27
  - Two Tests with Same Reference Noha Abdeljawad 2012-12-03 08:29
    - Two Tests with Same Reference jag009 2012-12-03 15:35
      - X-way d_labes 2012-12-04 09:25
        
        design kvgreddy06 2013-12-19 15:32
        
        design kvgreddy06 2013-12-30 16:48
        
        design intuitivepharma 2013-12-31 06:44
        
        design kumarnaidu 2013-12-31 08:24
        
        Problematic designHelmut 2013-12-31 15:02
        
        Problematic design kvgreddy06 2014-01-03 14:46
        Problematic design kumarnaidu 2014-12-24 12:56
        
        Problematic design Helmut 2014-12-24 16:10
        
        Problematic design kumarnaidu 2014-12-26 05:38
        
        Two fully replicated designs? Helmut 2014-12-27 15:52