Bioequivalence and Bioavailability Forum • Bioequivalence of Sample Size (Over Power)

bjkim97
☆

Korea / Seoul,
2011-02-08 18:54
(5194 d 03:49 ago)

Posting: # 6596
Views: 6,677

Bioequivalence of Sample Size (Over Power) [Power / Sample Size]

Dear All

Hi! Every One. I have a question

Bioequivalence Studies to determine the number of subjects.

Expected Ratio and ISCV have to decide the number of subject familiar

But one kind of question we have got a lot of subject by setting the number of Power is more than 95%, 99%

Number of subjects to make decision about what I wonder if there are problems

If you have any comment on this point I want to hear your opinions.

^^;;

Byung-Ju Kim.
Bioequivalence Scientist
Tel : +82 2 317 2081 / +82 10 3955 1601

Helmut
★★★

Vienna, Austria,
2011-02-08 21:30
(5194 d 01:13 ago)

@ bjkim97
Posting: # 6598
Views: 5,888

Don’t Over Power

Post reply

Dear Byung-Ju,

seems that you have a high budget. That’s nice and decreases your changes of failure (producer’s risk β = 1 – power).
But: Most guidelines suggest a power of 80–90%; you may run into problems with the Ethics Committee. It’s the job of the EC to judge the risk of subjects (drug effects, common risks due to repetitive venipuncture, AEs, etc.) and the advantage for the health system (cheaper drugs). It’s not the job of the EC to decrease the risk of failure of a pharmaceutical company, which is commonly 10–20%. You can give it a try with 95% power (if you expect a high drop-out rate), but I don’t think that any EC will accept 99% power (more than doubling the sample size from 80% power; example below for T/R 0.95):

CV% power│ 15 20 25 30 ─────┼─────────────── 80 │ 12 20 28 40 90 │ 16 26 38 52 95 │ 18 32 46 66 99 │ 26 44 68 94

There’s another pitfall. Did you ever come across the term ‘forced bioequivalence’? The sponsor suspects that the test will deviate a lot from the reference but will not present a sample size estimation based on that. Instead a sample size estimation with high power for a better (though not realistic) value is presented. Example:

∆ 11% (T/R 89%), n=44, 79.9% power (realistic; not presented)
∆ 5% (T/R 95%), n=44, 99.1% power (presented)

Regulators might not like that. Quote:

We are interested in public health –
not in the profit of the pharmaceutical industry.

—
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

Dr_Dan
★★

Germany,
2011-02-09 10:31
(5193 d 12:12 ago)

@ Helmut
Posting: # 6599
Views: 5,811

Don’t Over Power

Post reply

Dear Byung-Ju, dear Helmut

the advantage of overpowering a study (seen from the company perspective) is that you get very narrow 90% confidence intervals. So if you have a formulation which is borderline bioequivalent (for example if your point estimator is round about 85% or 120%) you can get results within the acceptance range if you include much more subjects than needed according to the intra subject variability of the drug. There is a certain risk that the regulatory authority could refuse a marketing authorisation because only by overpowering the study became acceptable.
Kind regards
Dan

—
Kind regards and have a nice day
Dr_Dan

Helmut
★★★

Vienna, Austria,
2011-02-09 12:10
(5193 d 10:33 ago)

@ Dr_Dan
Posting: # 6602
Views: 5,824

Expected deviation

Post reply

Dear Dan!

I agree with your points made about PEs ‘far away’ from unity. But if you expect a large deviation, IMHO the study should be powered for that – and not concealed by ‘overpowering’ for an unrealistic PE. The ‘old’ EU guideline was pretty specific (3.1 Design c):

The number of subjects required is determined by

the error variance associated with the primary characteristic to be studied as estimated from a pilot experiment, from previous studies or from published data,
the significance level desired,
the expected deviation from the reference product compatible with bioequivalence (∆) and
the required power.

Unfortunately the current guideline (4.1.3 Subjects / Number of subjects) reduced this section to :blahblah:

The number of subjects to be included in the study should be based on an appropriate sample size calculation.

❝ There is a certain risk that the regulatory authority could refuse a marketing authorisation because only by overpowering the study became acceptable.

I wouldn’t expect that. BE is solely defined by inclusion of the CI within the acceptance range. I’m only concerned whether regulators might not accept the protocol – that’s where the term ‘forced BE’ has its place.

See also FDA’s guidance, Appendix C (though speaking about PBE and IBE, ∆ ≤5% is suggested):

Sample sizes for average BE should be obtained using published formulas. Sample sizes for population and individual BE should be based on simulated data. The simulations should be conducted using a default situation allowing the two formulations to vary as much as 5% in average BA with equal variances and certain magnitude of subject-by-formulation interaction. The study should have 80 or 90% power to conclude BE between these two formulations. Sample size also depends on the magnitude of variability and the design of the study. Variance estimates to determine the number of subjects for a specific drug can be obtained from the biomedical literature and/or pilot studies.

—
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

bjkim97
☆

Korea / Seoul,
2011-02-09 18:28
(5193 d 04:15 ago)

@ Helmut
Posting: # 6606
Views: 5,778

Don't Over Power

Post reply

❝ seems that you have a high budget. That’s nice and decrease your changes of failure (producer’s risk β = 1 – power).

- Than twice the cost of failure to invest in one successfully doing is considered a profit.

❝ But: Most guidelines suggest a power of 80–90%; you may run into problems with the Ethics Committee. It’s the job of the EC to judge the risk of subjects (drug effects, common risks due to repetitive venipuncture, AEs, etc.) and the advantage for the health system (cheaper drugs). It’s not the job of the EC to decrease the risk of failure of a pharmaceutical company, which is commonly 10–20%. You can give it a try with 95% power (if you expect a high drop-out rate), but I don’t think that any EC will accept 99% power (more than doubling the sample size from 80% power; example below for T/R 0.95):

❝ CV%

❝ power│ 15 20 25 30

❝ ─────┼───────────────

❝ 80 │ 12 20 28 40

❝ 90 │ 16 26 38 52

❝ 95 │ 18 32 46 66

❝ 99 │ 26 44 68 94

❝

❝ There’s another pitfall. Did you ever come across the term ‘forced bioequivalence’? The sponsor suspects that the test will deviate a lot from the reference but will not present a sample size estimation based on that. Instead a sample size estimation with high power for a better (though not realistic) value is presented. Example:

❝ ∆ 11% (T/R 89%), n=44, 79.9% power (realistic; not presented)

❝ ∆ 5% (T/R 95%), n=44, 99.1% power (presented)

All of the comments I receive a case of empathy.
But one kind of different opinions, some more than 80% or 90% of the test to make sure once you reach a positive result as doing the rights of the subjects that can guarantee the Doors
In South Korea did not bring positive results if the re-test.
So far no clear rules for re-test has not been established.
Do you know of other countries to review the re-test regulations?

❝ Regulators might not like that. Quote: We are interested in public health – not in the profit of the pharmaceutical industry.

The public health is my first priority will be to work

Thank you for giving me feedback

BJ