## Non-parametric N [Power / Sample Size]

Hi Alex,

❝ So we would calculate our N based on a non-parametric test for T-max differences (I have heard some epidemologists saying its ok to calculate with T-Test and increase like 15%)

That’s black magick. Why 15%?
• If (if!) the distribution is normal, the Wilcoxon signed-rank test (for paired differences) and the Mann–Whit­ney U test (for independent samples) have an asymptotical power of $$\small{3/\pi\approx95.5\%}$$.
• If the distribution is not normal (e.g., tmax), power of nonparametric tests are higher than the ones of the corresponding t-tests.

❝ Kind of related, I guess no concern exists about the almost sure increase in Cmax if it doesnt go beyond the therapeutic range inst it?

No idea. In the EMA’s bracketing approach you perform a Non-Superiority test with an upper margin of 1.25. All of this stuff (including BE) is not related to the therapeutic range at all.

❝ Not possible to calculate N for Tmax but to perform certain simulations with some parameters.

Even then it’s not an easy job.

❝ So better to run a pilot without and seein what comes?

Yes.

❝ if you are right Tmax will come sooner (with a higher Cmax).

Likely. But as I wrote before: Under the premise of no saturable pre-systemic metabolism.

❝ Would it support a SmPC change?

No idea. Once you have the result of a pilot, consider a scientific advice. Sweden?

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

22,428 posts in 4,694 threads, 1,598 registered users;
11 visitors (0 registered, 11 guests [including 7 identified bots]).
Forum time: 14:13 CET (Europe/Vienna)

Operational hectic replaces
intellectual calms.    Alexander Huiskes

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