In looking for examples of 'roof' /rʊf/, I came across a couple of discussions that also mentioned 'root' /rʊt/, so I suspect all those words may be in the same group with 'room' /rʊm/, at least for AE.
But I'm not sure about the theory that it might be characteristic of the US Midwest -- some participants in this thread seemed to suggest the opposite, if I followed the symbols correctly (U is evidently used to represent /ʊ/).
https://www.waywordradio.org/discussion/topic...
Using those other words, I came to some discussions about variation within both the US and the UK, with links to other sources.
http://dialectblog.com/2012/06/27/goose-room-... cites
the Harvard Dialect Survey
http://www4.uwm.edu/FLL/linguistics/dialect/s...
and John Wells citing Daniel Jones's EPD
http://phonetic-blog.blogspot.com/2010/09/roo...
One contributor mentioned a Wiki page on "Shortening of /uː/ to /ʊ/," which might be this one:
https://en.wikipedia.org/wiki/Phonological_hi...
In searching for it I also came across, e.g.,
https://english.stackexchange.com/questions/3...
https://www.antimoon.com/forum/t6657.htm includes this post:
As luck would have it, this US dialect survey had a question on words like that: http://cfprod01.imt.uwm.edu/Dept/FLL/linguist... . And you can look here: http://cfprod01.imt.uwm.edu/Dept/FLL/linguist... to see the percentages by state.
I haven't chased down all those secondary links, but my preliminary impression is that it does vary, but not always or only according to geographic region, and there's evidently also some room for disagreement over which pronunciation was 'older.'