Bioequivalence and Bioavailability Forum • “Group effect” in BE

Helmut
★★★

Vienna, Austria,
2017-05-27 16:07
(2907 d 20:10 ago)

Posting: # 17422
Views: 9,985

“Group effect” in BE [Surveys]

Dear all,

I created an open survey to gather information about the current (and past) state of affairs. Ten questions, less than five minutes to answer. Feel free to join.
Sneak preview (what to expect) below:

Multigroup studies

Design

How do you design a multigroup study?

Staggered
(i.e., immediately after one group has finished a period, the next group starts theirs)
Stacked
(i.e., complete all periods of one group before the next group starts)
Both
(case by case, e.g., staggered for single dose studies and stacked for multiple dose studies)

Pooling of data

Procedures (1/2)

Do you routinely pool data?

Yes
I don’t take multigroup nature of study into account.
Yes
But I adjust the statistical model accordingly.
Case by case
I follow a decision scheme.

Pooling of data

Procedures (2/2)

Do you justify in the protocol why you consider pooling of data a valid approach?

Always
Never
Case by case
Please give an example.

What is your justification of not using group-terms in the model?

Based on the FDA’s
- The clinical study takes place at one site,
- all study subjects have been recruited from the same enrollment pool,
- all of the subjects have similar demographics, and
- all enrolled subjects are randomly assigned to treatment groups at study outset.
My interpretation of the EMA’s BE guideline
It is a source of variation that cannot be reasonably assumed to have an effect on the response variable.
Other
Please specify.

Models

Procedures

Do you assess models recommended by the FDA?
Group, Sequence, Treatment, Subject (nested within Group × Sequence), Period (nested within Group), Group-by-Sequence Interaction, Group-by-Treatment Interaction.

Group, Sequence, Treatment, Subject (nested within Group × Sequence), Period (nested within Group), Group-by-Sequence Interaction.

Sequence, Treatment, Period, Subject (nested within Sequence).

Yes
No
Don’t know

Do you follow the decision scheme recommended by the FDA?
If the G×T term in model 1 is ≥0.1, assess BE by model 2 of pooled data.

If the G×T term in model 1 is <0.1, assess BE by model 3 of the groups.

Yes
No
I pool data and use model 2 without a pre-test.
No
I pool data and use model 3 (as if it is a single group study).
Don’t know

Models

Observations (please skip if you didn’t assess model 1)

How often did you observe a significant Group-by-Treatment interaction?

In much less than 10% of studies
In about 10% of studies
In much more than 10% of studies
Don’t know
Other
If possible, give the number of studies with a significant G×T term and the total number of studies.
Example: 7 / 65.

Applicability

Regulatory acceptance (protocol)

Was your approach accepted?

Yes
Justification for not assessing the G×T. Pooling of data and evaluation by model 3 planned.
Yes
Following the decision scheme planned; dependent on the G×T by model 1:
p ≥0.1: Pooled data by model 2.
p <0.1: Data of the largest group by model 3.
Yes
Pooled data by model 2 (without the pre-test) planned.
Yes
Pooled data by model 3 planned but nothing specifically stated.
No
(please give the reason for the rejection of your approach and – if possible – the agency and year).

Applicability

Regulatory acceptance (report)

Was your approach accepted?

Yes
Justification for not assessing the G×T given in the SAP.
Evaluation of pooled data by model 3.
Yes
The decision scheme observed.
Evaluation of pooled data by model 2 or model 3 of the largest group.
Yes
Evaluation of pooled data by model 2 (without a pre-test) performed.
Yes
I didn’t take the multigroup nature of study into account.
Evaluation of pooled data by model 3.
No
(please give the reason for the rejection of your evaluation and – if possible – the agency and year).

Problems

Deficiency letters, rejected studies, etc.

In which country / jurisdiction / with which agency did you face the most problems?

USA
MENA States
Russian Federation
Eurasian Economic Union (except Russia)
European Economic Area (EU; Iceland, Liechtenstein. Norway)
Brazil
ASEAN States
Other
(please give the reason for the rejection of your evaluation and – if possible – the agency and year).
Alternatively: Agency and year where you faced no problems.

—
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

Ben
★

2017-05-31 22:02
(2903 d 14:15 ago)

@ Helmut
Posting: # 17437
Views: 6,937

“Group effect” in BE

Post reply

Dear Helmut,

thank you for this post. This sounds very interesting and I am actually excited to have some kind of insights of the results.

Will you present this at some point in time?

❝ Do you assess models recommended by the FDA?

❝ ...

❝ Do you follow the decision scheme recommended by the FDA?

❝ ...

Maybe I completely missed it, but I couldn't find those recommendations in the guidances. Is there a reference, would you be so kind and share it if possible?

Thank you,
Ben.

Helmut
★★★

Vienna, Austria,
2017-05-31 22:38
(2903 d 13:38 ago)

@ Ben
Posting: # 17438
Views: 7,010

“Group effect” in BE

Post reply

Hi Ben,

❝ […] I am actually excited to have some kind of insights of the results.

❝ Will you present this at some point in time?

I will write a paper on it. What I have so far over there. The survey started last Saturday; six respondents as of today. ;-)

❝ ❝ Do you assess models recommended by the FDA?

❝ ❝ ...

❝ ❝ Do you follow the decision scheme recommended by the FDA?

❝ ❝ ...

❝

❝ Maybe I completely missed it, but I couldn't find those recommendations in the guidances.

Stop searching. Nothing specific about multigroup studies in the FDA’s 2001 biostats guidance.

❝ Is there a reference, would you be so kind and share it if possible?

See f.i. here or there. You may find some deficiency letters by the FDA with the same wording on the net. this:
"Group-by-treatment interaction" FDA filetype:pdf

Model 1, testing of the G×T term at the 0.1 level and dropping if n.s. (⇒ Model 2) is described by Bolton and Bon.*

Bolton S, Bon C. Pharmaceutical Statistics. Practical and Clinical Applications. New York: informa healthcare; Fifth edition 2009. p. 629.

—
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

Ben ★ 2017-06-05 19:25 (2898 d 16:52 ago) @ Helmut Posting: # 17443 Views: 6,616	“Group effect” in BE Post reply
	Thank you for the info! Best regards, Ben.