In my experience, campaigns fold not because of mechanical issues but just because they run out of steam.
...I have a vague recollection that WotC did some kind of market research that backed that up.
When they were developing 3.0 they discovered that people 'rarely played beyond 10th level.' Or something like that. They didn't try to find out /why/, per se. But they also determined that players found the game had a 'sweet spot' at mid levels when it worked best.
I suppose it's a chicken-and-egg thing. After 25 years of playing a game that got pretty wonky after name level, people were found not to care much for high level play. So WotC mostly just playtested 3.0 at 1-10, and it ended up not working so well after 10th.
My experience with AD&D and 3e bears that out, too. In AD&D it was hard to get a party to 2nd level and maintain interest from the group - but, if you got them to 3rd or maybe 5th, they were hooked, and might go to 9th or even low teens. Of course, that was heavily-modified AD&D, because, really, who didn't modify it.

3e, no campaign I was in went beyond 14th. The group didn't break up, they'd just drop the campaign and start another with new characters when it wasn't fun anymore. In 4e, I'm in a campaign that's at 18th and still going, and running a casual one that's hit 14th. It still works at those levels in a way prior eds didn't.
It'll be interesting to see how 5e does, if it can extend the sweet spot through high levels like 4e did - if it runs into problem after 11th (when Mearls has said it 'changes') or just has a different 'feel' but is still playable.