Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Biotech / Medical : Theravance - THRX -- Ignore unavailable to you. Want to Upgrade?

To: Biomaven who wrote (40)	8/28/2006 12:59:45 PM
From: bio_kruncher	Read Replies (2) \| Respond to of 74

I think Dew was think of the case where you can find noninferiority and inferiority as it is not possible to find superiority and not find noninferiority. I think that in this case no p-value adjustment was needed nor taken but the issue is with how they described test as one-sided vs. two-sided. The customary way to describe statistical tests for registration trials is to have two-sided test at 0.05. However, we are really doing one sided tests at 0.025 because we can only claim effectiveness if new treatment is better than control. So they did a one-sided test(at 0.025) and got a pvalue of 0.053. This may make a better press release than a p-value from a two sided test which would have been 0.106, but is equivalent. In these studies, noninferiority was needed to claim effectiveness. If the test of superiority in MRSA was conditional on first achieving noninferiority overall then no alpha adjustment would be needed. The superiority is really a marketing issue. They could have presented a claim on the label and therefore promoted it more easily.

To: Biomaven who wrote (40)	8/28/2006 2:13:49 PM
From: DewDiligence_on_SI	Respond to of 74

I don’t know of cases where this happened but it’s a theoretical possibility as explained in my reply to bio_kruncher. The FDA’s definition of non-inferiority varies greatly from trial to trial, especially in the medical-device arena where trials seeking a finding of non-inferiority are very common. The superior-but-not-non-inferior anomaly is bound to happen in due course if it hasn’t already. To sum up: the “two bites at the apple” logic does not apply to dual endpoints of superiority and non-inferiority if the statistical-analysis plan adheres to the Hochberg correction. However, the “two bites” logic does apply (at least in theory) if the SAP does not employ Hochberg. Regards, Dew