IMO, there is just way to many factors to ever come up with a set answer..
Even if you would say that one out of 10 cams with X duration or with a factor y of such and such would wipe a lobe how would you ever know if the cam you installed was the 1st or the 10th cam?????
The biggest problem is with pin pointing what cause the failure... So if you blame 5 cams with x numbers for having failures and all 5 were caused by improper break-in all that data is flawed.....
The only way i could see collecting usable data is to do it all in controlled conditions... my guess would be that in controlled conditions there would be zero failures because you would eliminate all the options for other factors causing the failures......