I see books all the time that go several decades between check-outs, according to the piece of paper in the back. Like nofloyd points, out, the 30% figure should be adjusted to account for books that have been checked out but are not yet in the computer system.
If we can work out a good fit to the frequency at which books get borrowed (some Zipf process?), based on records we do have, then, given the total number of books in a collection, we should be able to estimate the total numbers.
Another caveat is that if, say, I need to check a chapter of some new or old book at an inconvenient hour and can do so on-line, I am not going to make a trip to the library to find a physical copy. That skews the frequency way down, unless you account for it.