Log in
Register|
Anu ★ India, 2013-08-26 09:15 (4318 d 05:02 ago) Posting: # 11341 Views: 8,105 |
|
|
Hi all, Greetings! Kindly help me in knowing how can we change the values lying between confidence Interval of (-20% to 20%) e.g (90% lower limit (-1.07) & 90% upper limit (-0.30)) into the values lying between the CI (80-125)%. Thanks & Regards Anu |
|
Dr_Dan ★★ Germany, 2013-08-26 13:05 (4318 d 01:11 ago) @ Anu Posting: # 11343 Views: 6,774 |
|
|
Dear Anu Could you please repeat your question in other words? I do not get the sense. Kind regards Dan — Kind regards and have a nice day Dr_Dan |
|
Anu ★ India, 2013-08-27 08:26 (4317 d 05:50 ago) @ Dr_Dan Posting: # 11352 Views: 6,704 |
|
|
Hi Dr_ Dan, ❝ Could you please repeat your question in other words? I do not get the sense. Sorry if the language was not clear. I just need to know in FDA the accepted confidence intervals ranges are (80%-125%) & (-20% to +20%). I have a data for a reference product whose 90% CI is (-1.07% to -0.30%) on the CI scale of (-20% to +20%). I want to transform it into the CI scale of (80%-125%). Can I do this? Thanks & Regards Anu |
|
Helmut ★★★   Vienna, Austria, 2013-08-27 17:07 (4316 d 21:10 ago) @ Anu Posting: # 11357 Views: 6,624 |
|
|
Hi Anu, ❝ I just need to know in FDA the accepted confidence intervals ranges are (80%-125%) & (-20% to +20%). The acceptance range is 80.00–125.00% (90% CI in percent rounded to two decimals). ❝ I have a data for a reference product whose 90% CI is (-1.07% to -0.30%) on the CI scale of (-20% to +20%). This doesn’t makes sense to me. Do you mean data for a test product (in % of the reference)? ❝ I want to transform it into the CI scale of (80%-125%). ❝ Can I do this? Duno. I don’t believe in these numbers – where do they come from? In % of the reference the CI would be 98.93–99.70%. Assuming a 2×2 cross-over that would translate even for the smallest acceptable sample size of 12 into a CV of 0.52% [sic]. Simply impossible. — 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 |
|
Anu ★ India, 2013-08-28 11:08 (4316 d 03:08 ago) @ Helmut Posting: # 11365 Views: 6,542 |
|
|
Hi Helmut, ❝ This doesn’t makes sense to me. Do you mean data for a test product (in % of the reference)? Actually the 90% CI is for the test-reference ❝ Duno. I don’t believe in these numbers – where do they come from? I don't have any raw data, just got this confidence interval (on -20% to +20%) scale from my boss. And its a two way cross over study with 21 subjects. On the basis of which my objective is to calculate the sample size. Which I will be doing with R software and the FARTSSIE Excel sheet. So, it came in my mind why not to transform this CI into (80.00 to 125.00)%. It should be possible? Does there exist any such transformation relation between these two different CI scales i.e (80.00 to 125.00)% and (-20.00 to +20.00)? Thanks & Regards Anu |
|
jag009 ★★★ NJ, 2013-08-26 17:48 (4317 d 20:29 ago) @ Anu Posting: # 11348 Views: 6,671 |
|
|
Hi, ??? You mean from PK data or just the confidence interval values? John |
|
Anu ★ India, 2013-08-27 08:28 (4317 d 05:48 ago) @ jag009 Posting: # 11353 Views: 6,631 |
|
|
Hi John, Need to transform just the confidence interval values. Thanks & Regards Anu |
|
ElMaestro ★★★ Denmark, 2013-08-27 08:47 (4317 d 05:29 ago) @ Anu Posting: # 11354 Views: 6,751 |
|
|
Hi Anu, ❝ Need to transform just the confidence interval values. If you have CI results on the log-scale then you just need to exponentiate them. Which is what your protocol also says, right? Judging the numbers you have it does, however, sound like something went seriously pear-shaped with the BE study, the formulation or the data evaluation. Please paste the raw untransformed data and let's have a look. — Pass or fail! ElMaestro |
|
Anu ★ India, 2013-08-28 11:12 (4316 d 03:05 ago) @ ElMaestro Posting: # 11366 Views: 6,629 |
|
|
Hi Elmaestro, ❝ If you have CI results on the log-scale then you just need to exponentiate them. Which is what your protocol also says, right? Right. ❝ Judging the numbers you have it does, however, sound like something went seriously pear-shaped with the BE study, the formulation or the data evaluation. Please paste the raw untransformed data and let's have a look. Right now I don't have the raw data, will definitely upload once I'll have it. Thanks & Regards Anu |
Ing. Helmut Schütz