assumed (!) T/R ratio [Power / Sample Size]

posted by Helmut Homepage – Vienna, Austria, 2017-10-11 21:18 (2473 d 02:06 ago) – Posting: # 17888
Views: 4,296

Hi CECIF,

❝ […] it is possible to expect a priory a T/R ratio different to the unity (in a extent 0.8 to 1.20).


Sure. Note that in BE we are using a multiplicative model (or an additive on log-transformed data). Hence, a maximum accepted Δ of 20% translates into a BE-range of 0.80–1.25 in the raw domain (not 0.80–1.20).

❝ However it is not clear to me, what criteria or in which cases can I assume an apriori ratio as big as 0.8 or 1.20, if knowingly I am trying to demonstrate bioequivalence and then I should be expecting a T/R ratio of 1.00.


If you would expect a T/R-ratio exactly at one of the BE-boundaries, power would equal the Type I Error – which is the nominal level of the test or lower.
Anything else within the BE-range is fine, although T/R-ratios deviating a lot from unity could require extremely large sample sizes.

❝ In a particular case I happen to have a test product with content of the active compound that is above 110% of the content of the reference product.


I would try to get a batch of the reference which is closer to the test. F.i., according to the EMA’s BE-GL contents should not differ more than 5%. Only if you could prove (!) that it was impossible to find a better matching batch, a content correction is acceptable if stated in the protocol.

❝ Could this product be evaluated for bioequivalence, by designing a study with an apriori T/R ratio of 1.10?


In principle, yes. But think twice before going there. I suggest to get the package PowerTOST for the statistical system [image] (open source). Then you can perform all needed estimations yourself. Example:

library(PowerTOST)
sampleN.TOST(CV=0.2, theta0=1.10, targetpower=0.8, design="2x2")

Gives

+++++++++++ Equivalence test - TOST +++++++++++
            Sample size estimation
-----------------------------------------------
Study design:  2x2 crossover
log-transformed data (multiplicative model)

alpha = 0.05, target power = 0.8
BE margins = 0.8 ... 1.25
True ratio = 1.1,  CV = 0.2

Sample size (total)
 n     power
32   0.810068


Even if you will perform a content-correction don’t assume a T/R-ratio of 1.
  1. No analytical method is perfectly accurate and precise. Ask your analysts and you will be surprised. A method with an inaccuracy if 2% is already excellent.
  2. The method was validated for the test. You can only hope that it “works” equally good for the reference. Asking the originator for the CoA of the batch is futile. ;-)

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

Complete thread:

UA Flag
Activity
 Admin contact
23,112 posts in 4,858 threads, 1,644 registered users;
89 visitors (0 registered, 89 guests [including 17 identified bots]).
Forum time: 23:25 CEST (Europe/Vienna)

Skill is a function of chance.
It’s an intuitive best-use of chance situations.    Philip K. Dick

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