roody
●    

2005-10-17 07:50
(6738 d 10:03 ago)

Posting: # 40
Views: 11,915
 

 Four way crossover study [Design Issues]

Please give me the details on designing of four-way crossover BA/BE study. I want to conduct in fed and fasted conditions in approximately 16 volunteers with only one test and reference formulation. What are the important characterisitics of such design specially randomising in four periods?
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2005-10-17 14:45
(6738 d 03:08 ago)

@ roody
Posting: # 41
Views: 11,188
 

 Williams' Design

Hi Roddy!

You should apply a 4-period 4-sequence Williams' Design.

The four sequences are:
+-----+---+---+---+---+
| S/P | 1 | 2 | 3 | 4 |
+-----+---+---+---+---+
|  1  | D | C | A | B |
|  2  | A | D | B | C |
|  3  | B | A | C | D |
|  4  | C | B | D | A |
+-----+---+---+---+---+

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

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

2007-12-15 11:09
(5949 d 05:43 ago)

@ Helmut
Posting: # 1377
Views: 10,143
 

 Williams' Design

Dear Mr Helmut,
Can you give the rationale for applying "Williams' Design" in 4 way cross over study. Does any guidance recommends.

with regards

Ram mohan

--
Edit: Full quote removed. Please see this post! [HS]
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2007-12-15 14:21
(5949 d 02:31 ago)

@ drrammohanj
Posting: # 1380
Views: 10,374
 

 Williams' Design

Dear Ram!

❝ Can you give the rationale for applying "Williams' Design" in 4 way cross

❝ over study.


See this post for an example of a 6×3 (6 sequence, 3 treatment) Williams' design (including some references), and this one as well..

Williams' designs are balanced for eventual 1st-order carry-over effects.
Each treatment is preceeded once by each of the 3 other treatments or is given as the 1st treatment of the study.
In a set of classical Latin square design (ABCD|BCDA|CDAB|DABC), each treatment (when not the 1st treatment) is always preceeded by the same treatment (B by A, C by B, and D by C), which can bias the estimates in presence of differential 1st-order carry-over.

❝ Does any guidance recommends.


To my knowledge only Brazil ANVISA's, but please use it unconditionally - otherwise you may run into statistical troubles.

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

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

2007-12-18 18:36
(5945 d 22:16 ago)

@ Helmut
Posting: # 1393
Views: 10,273
 

 Williams' Design

Dear Helmut,
we have a study with William's design now with only 8 subjects (as every guidelines say BE study should be done with minimum of 12 subjects, IS IT ETHICAL to do study with 8 subjects) with 3 test formulations (T1, T2, T3) and one reference. I don't know why sponsor want to do study with so few subjects ? pre-pivotal, to know which of his test fomulation is having good bioavailability in comparision to reference. Is it possible to do statistical analysis with pk samples of just 4 subjects as in this situation (2T1&2R, 2T2&2R, 2T3&2R).

with regards
Ram Mohan
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2007-12-18 19:07
(5945 d 21:45 ago)

@ drrammohanj
Posting: # 1395
Views: 10,116
 

 Pilot study in 8 subjects?!

Dear Ram!

❝ we have a study with William's design now with only 8 subjects (as every guidelines say BE study should be done with minimum of 12 subjects, IS IT ETHICAL to do study with 8 subjects) with 3 test formulations (T1, T2, T3) and one reference.


IMHO simply not. The search feature of forum will give you some ideas why small sample sizes for pilot studies are not giving us reliable estimates of both the PE and CVintra. You may go with the minimum sample size of 12 if you repeat the 4 sequences 3 times; but I would recommend this only if you already know that you are dealing with a drug of low variability (let's say 10% or less).

❝ I don't know why sponsor want to do study with so few subjects ?


An unfortunate combination of costiveness and lack of knowledge? :-D
If you are working in a CRO, you will have to swallow the bitter pill and start some education...

❝ pre-pivotal, to know which of his test fomulation is having good bioavailability in comparision to reference.


The smaller the sample size the less reliable are the estimates… :-(

❝ Is it possible to do statistical analysis with pk samples of just 4 subjects as in this situation (2T1&2R, 2T2&2R, 2T3&2R).


If you really adopt this design, you will have 8 subjects (not 4). You have 4 sequences, each of administered to 2 subjects. Another reason not to perform such a small sized study are potential drop outs. Since you have 4 periods, it is much more likely that you will have to deal with a heavily inbalanced evaluation…

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,636 registered users;
111 visitors (0 registered, 111 guests [including 10 identified bots]).
Forum time: 16:53 CET (Europe/Vienna)

With four parameters I can fit an elephant,
and with five I can make him wiggle his trunk.    John von Neumann

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