In this related blog post, you can find links to the CONSORT statement in the Dec 26, 2012 issue of JAMA and a link to my letter to the editor.
Addendum: I should have included this in the video. See the picture below. In the first example, top left, the entire 95% CI favoring "new" therapy lies in the "zone of indifference", that is, the pre-specified margin of superiority, a mirror image of the "pre-specified margin of noninferiority, in this case delta= +/- 0.15. Next down, the majority of the 95% CI of the point estimate favoring "new" therapy lies in the "margin of superiority" - so even though the lower end of the 95% CI crosses "mirror delta", the best guess is that the effect of therapy falls in the zone of indifference. In the lowest example, labeled "Truly Superior", the entire 95% confidence interval falls to the left of "mirror delta" thus reasonable excluding all point estimates in the "zone of indifference" (i.e. +/- delta) and all point estimates favoring the "old" therapy. This would, in my mind, represent "true superiority" in a logical, rational, and symmetrical way that would be very difficult to mount arguments against.