If people REALLY cared, we could get a large sample of coins, count up the numbers for every year and compare to the ratios that each year was produced in.
eg if there were 100 million coins produced in 1980 and 100 million coins produced in 2000, you would expect to find the same numbers of 1980 and 2000 coins in a large sample of coins - although losses of the older coins would make things not equal, and you could see how big the difference was.
I have considered doing that, and think it would be a very good way to work out the rate they disappear from circulation, but would take a lot of work.