Finding contact information for certain individuals is more difficult than it might seem initially, though. Especially if someone has left the academic setting per se: a common pattern is for people to leave and go to a business or hospital setting, where real contact information can be guarded heavily. You might see that so-and-so works at such-and-such place (maybe), but not be able to find a working contact address, and might not even see a reasonable way to figure it out.
You're right that using the author-supplied contact info of a potential reviewer is doubly problematic and probably shouldn't be done. However, at some level I can understand why that happens, because the process often devolves into some chain of contacts to get a working email address anyway.