90% CI of ratio of SDs for NTI drugs

[satish123]

Dear all,

As we know for NTI drugs guidelines http://www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidances/ucm201283.pdf has suggested that use both (1)scaled and unscaled ABE, in addition to that (2)calculation of 90% CI of ratio of within subject standard deviation of test to reference product.

here my problem is (2). i don't know how to calculate (2). but i tried it with 'COVPRMS' of 'PROC MIXED' from normal ln transformed data (not from ilat, dlat). i calculated within subject standard deviation for test and reference using 'estimates' in 'COVPRMS'. now i just want to know am i doing it in the correct way or not?

Thanks and Regards,

Satish.