"failing" can be taken several ways ~ if device testing isn't performed on a regular basis, the effectiveness of the check valves can only be guessed at. For a PVB, that doesn't trouble me very much, since the spring-loaded vacuum breaker has not yet shown any tendency to stick.
I suspect the process of gaining agency approvals is ample demonstration of the relief valve performance of an RPZ. Our testing is all about check valves. For a DCVA, I consider it essential, since the device is sealed, and contains nothing else but check valves. The air openings on PVBs and RPZs make the check valves in those devices less critical to the task of protection.