Log in
Register|
AngusMcLean ★★ USA, 2010-10-27 05:23 (5296 d 09:37 ago) Posting: # 6084 Views: 6,288 |
|
|
Tuesday October 26, 2010: Concept of power of bioequivalence study troubles me. The BE test of the pharmacokinetic parameters (Cmax and AUC) should be powered to be 0.8. Now, if after completing a BE data analysis, assume the parameter passes since it falls within the confidence intervals of 0.8-1.25. If at that point the power is calculated and the value is 0.5 does it mean "you got lucky" but "do not try it again". Another point is if the parameter fails the BE test by being outside the confidence intervals for test v reference then you calculate the power. Will the value be low say ~0.4. or can the power of the test on this parameter failing BE still be 0.8 or above? Please can someone me through this? Angus Edit: Category changed. [Helmut] |
|
d_labes ★★★ Berlin, Germany, 2010-10-27 13:00 (5296 d 02:00 ago) @ AngusMcLean Posting: # 6087 Views: 5,405 |
|
|
Dear Angus, have a look at this post and search for others. We had this theme already discussed often. Have also a look at Helmut's famous lectures / presentation especially the latest: "Sample Size Challenges in BE Studies and the Myth of Power" — Regards, Detlew |
|
yjlee168 ★★★   Kaohsiung, Taiwan, 2010-10-27 14:06 (5296 d 00:55 ago) @ d_labes Posting: # 6089 Views: 5,381 |
|
|
Dear D. Labes & Helmut, Great presentations & refs. Just cannot miss this one. Many thanks.  ❝ latest: ❝ "Sample Size Challenges in BE Studies and the Myth of Power" Edit: THX for the compliments! I had 50 minutes for the presentation today - therefore I started with a polite "Please fasten seat belts!"  [Helmut]— All the best, -- Yung-jin Lee bear v2.9.2:- created by Hsin-ya Lee & Yung-jin Lee Kaohsiung, Taiwan https://www.pkpd168.com/bear Download link (updated) -> here |
|
ElMaestro ★★★ Denmark, 2010-10-27 13:41 (5296 d 01:19 ago) @ AngusMcLean Posting: # 6088 Views: 5,974 |
|
|
Ahoy Angus, ❝ The BE test of the pharmacokinetic parameters (Cmax and AUC) should be powered to be 0.8. I completely agree with d_labes and will add that power is not a formalised requirement. Consider it from a regulator's view; it is all about risks and benefit when she or he gives his nod to a dossier or shreds it. Given a dossier, the assessor has a study report with the actual results available. Power at this point will not affect neither the risk or the benefit, only the result of the trial counts here. Power is your chance of showing equivalence with a given number of subjects provided your assumptions (around CV and T/R) are met. Therefore, power is not your headache and it is not the regulator's headache either. Power is a headache that belongs to the guy in the Armani suit who allocates money to your project. Such people often are not too well versed with science but they know how make a Windsor tie knot and other very useful stuff. Give such guy a graph of power versus N comes and present it as "chance of success as function of N" or even "chance of approval as function of cost" (in some cases give him an aspirin as well). — Pass or fail! ElMaestro |
Ing. Helmut Schütz