Log in
Register|
jagsuraiah ☆ 2009-03-30 15:16 (6292 d 16:22 ago) Posting: # 3415 Views: 6,847 |
|
|
HI every one, Paired equivalence method: sample size-> Paired equivalence->multiplicative 1) multiplicative Cv=0.62 T/R=0.95 POWER=0.80, LOWER=0.80 UPPER=1.25 then N=72 Cv=62 T/R=0.95 POWER=0.80, LOWER=80 UPPER=125 then N=1774 sample size-> Paired equivalence->Additive 2) Additive Cv=0.62 T/R=0.95 POWER=0.80, LOWER=0.80 UPPER=1.25 then N=108 Cv=62 T/R=0.95 POWER=0.80, LOWER=80 UPPER=125 then N=108 Two-sample equivalence method: sample size-> Two-sample equivalence->multiplicative 1) multiplicative Cv=0.62 T/R=0.95 POWER=0.80 LOWER=0.80 UPPER=1.25 N=141 Cv=62 T/R=0.95 POWER=0.80 LOWER=80 UPPER=125 N=3546 sample size-> Two-sample equivalence->Additive 2) Additive Cv=0.62 T/R=0.95 POWER=0.80 LOWER=0.80 UPPER=1.25 N=213 Cv=62 T/R=0.95 POWER=0.80 LOWER=0.80 UPPER=1.25 N=213 which method we have to take for finding sample size in BA/BE. Which method is accepted by regulators. give suggestion to me. thanks in advance. suri -- Edit: Subject line changed; please try to be more specific the next time. Helmut |
|
Helmut ★★★   Vienna, Austria, 2009-03-30 16:08 (6292 d 15:30 ago) @ jagsuraiah Posting: # 3418 Views: 5,635 |
|
|
Dear Suri! It seems that you have (mis)used some piece of software. Which one? Or if a general statistical package was used, which program/macro/routine? Some programs give the number of subjects / sequence, others the total... Just to give you an idea: ❝ Paired equivalence method: ❝ sample size-> Paired equivalence->multiplicative ❝ 1) multiplicative ❝ Cv=0.62 T/R=0.95 POWER=0.80, LOWER=0.80 UPPER=1.25 then N=72 I guess you want to know the total sample size in a 2×2×2 cross-over design with CVintra of 62% (=0.62)? 142 subjects are needed (or 71 / sequence). ❝ which method we have to take for finding sample size in BA/BE. Sample size calculation depends on
❝ Which method is accepted by regulators. Any one – if the method is suitable for the design. — Dif-tor heh smusma 🖖🏼 Довге життя Україна! ![[image]](https://static.bebac.at/pics/Blue_and_yellow_ribbon_UA.png) Helmut Schütz ![[image]](https://static.bebac.at/img/CC by.png) The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
jagsuraiah ☆ 2009-03-31 14:01 (6291 d 17:37 ago) @ Helmut Posting: # 3423 Views: 5,592 |
|
|
thanks for your answer. if it is 2x2 cross over study, intra cv =0.62 T/R=0.95 then which method we have to take for sample size calculation. Paired equivalence method or two sample equivalence with regards suri |
|
MGR ★ India, 2009-04-01 10:24 (6290 d 21:14 ago) @ jagsuraiah Posting: # 3430 Views: 5,662 |
|
|
Dear Suri, I think you have used SAS software (using Analyst) to calculate the sample size. Generally the methods mentioned by you are used by most of the people. And you have to take the multiplicative method for Cmax, AUC parameters in BE stuides. Thank you, — Regards, MGR |
|
LKN ● 2009-04-01 12:44 (6290 d 18:54 ago) @ MGR Posting: # 3433 Views: 5,542 |
|
|
Dear All, If the Cv is 62% isn't it possible to have the wider acceptance range for equivalence as in HVD 0.7-1.33 - with the implications for sample size as well? |
|
MGR ★ India, 2009-04-01 16:46 (6290 d 14:52 ago) @ LKN Posting: # 3440 Views: 5,555 |
|
|
Dear LKN, ❝ If the Cv is 62% isn't it possible to have the wider acceptance range for equivalence as in HVD 0.7-1.33 - with the implications for sample size as well? It is not possible to have the wider acceptance range for equivalence as 0.7 - 1.33 for USFDA submissions. As looking into CV we can go for RSABE according to the new recommendations of the USFDA as the drug is an HVD. If you want the details on the RSABE you can go through the search feature in this forum and you can find a lot of references. Thank you, — Regards, MGR |
|
Helmut ★★★ Vienna, Austria, 2009-04-01 17:06 (6290 d 14:32 ago) @ MGR Posting: # 3441 Views: 5,599 |
|
|
Dear LKN & MGR! ❝ ❝ If the Cv is 62% isn't it possible to have the wider acceptance range for equivalence as in HVD 0.7-1.33 [...] Oops! Just a reminder: Regulators commonly define an acceptable deviation from the reference of 20%. In the old days (no log-transformation of data) this meant an acceptance range of 80%-120%. When the log-transformation came into play in the mid '80ies - in order to keep the symetry around 100% - by convention (!) the lower limit in the untransformed scale was set to 100%-20% = 80% and the upper limit to its reciprocal 1/80% = 125%. In log-scale -0.2231 | +0.2231 (symetrical around zero). There was some debate whether the limits should be set to 100%+20% = 120% and 1/120% = 83.3...% (log-scale -0.1823 | -0.1823), but people advocating this approach lost - maybe just by chance... Therefore there are no limits like 0.7-1.33! In a very old European BE-draft (again from the '80ies) 70%-142.85714% (=1/0.7) were mentioned; currently 75%-133.33% are given as an example and sametimes a narrower range of 90%-111.11% for NTIDs.— 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 |
Ing. Helmut Schütz