Bioequivalence and Bioavailability Forum

Hi all,

By reading this nice lecture: Page 23- 25, I get more clear idea about the 95% upper confidence bound concept. Many thanks!

Here I have one following question (maybe very naive), I was wondering that why we need rearrange the SABE model
ln(0.8)*sigmaW/sigma0 <= muT-muR <= ln(1.25)*sigmaW/sigma0
to the linear form
(muT-muR)2-(ln(1.25)*sigmaW/sigma0)2<=0
and then calculate the 95% CI from the linear form? What will happen if we use the regular unscaled BE way to calculate 90% CI of mu_T-mu_R and compare with the scaled BE limits?

Thank you very much!

Best regards,
Winnie