Biden could easily win that district by that much and still lose the state. Donnelly received about 6% more than Clinton in rural areas, so if Trump performs like he did in 2016 in those areas, but tanks in IN-5, then Biden probably ends up with about 44-45% of the vote statewide which is not impossible.
In fact, Biden at 45% statewide in Indiana seems eminently plausible on a Biden+10 victory nationwide, which is where the national polls are at roughly now as well. That would be perfectly in line with the UNS.
And, yes, obviously, once again, the poll is of a district that is not representative of the national swing. Biden gaining 25 points net on Clinton in this district seems like a lot to me, but on the other hand the indications of the very largest swings have generally been in the suburbs of Southern cities (Dallas, Houston, Atlanta), and, while Indianapolis is in the North, it's definitely the most "Southern" Northern metropolis. This district could trend very strongly Democratic while most of the rest of the state trends Republican for an overall Indiana swing that is roughly in line with, or possibly even less than, the national swing.
