[kelnos]

Yes, why not? You don't re-test every single one, though: you spot-check a statistically significant percentage of them. Or maybe you do check all of them, but only for a one month period every year (a month that changes every year, and isn't known to the testing lab, so they can't game the system).

Another option is to send "control samples" to the testing lab, something you know already is infected with something they should be testing for. Do this enough times, and you'll know if they're accurately reporting the bad samples.

This type of thing is the only way for anyone in any kind of organization to verify that their outsourcing is effective and they're getting the result they want.