Granted, although I might counter that given so broad an apparent increase in the scope of the indication, 2017-2021 (? I can't immediately find a date for Ozempic specifically) and 2021- might best be taken as at least partially distinct.

I think that's an argument less well advanced by me than by someone who could design such a study, though, not least because they'll know far better than I whether the proposition makes any real sense.