May I relate an example of what you're all talking about?
I had a student who got a little slow on his demonstrated (simulated) asymmetric approach to land. Since it was a competency check I allowed him to do so. It's not every day someone tries to kill me, so I watched with great interest.
At about 200 feet he realised he was not on profile, and slow as well. So he added significant power to the live engine. Regrettably, and not unexpectedly, he added only a small amount of rudder, and that after the power increase.
What followed was a textbook approach to Vmc, but he did not recognise it, the aircraft continuing to yaw and roll for long enough for him to know how serious the situation was. (No flaming please - yes I did immediately take over, and no we did not get anywhere near the actual Vmc of the day, although the student's lack of rudder input made it look like we did.)
This student had followed the standard course of training, which included several approaches to a situation resembling Vmca. He was quite current, and primed for the test. He had excellent ground knowledge of Vmca, having discussed the theory & effects in class. Upon encountering unexpected yaw and roll on an asymmetric approach, he froze.
I suspect that he is not alone, and that the relatively few Vmca fatalities have more to do with Lycoming reliability than good training.
(Closing note - in this particular individual's case, his asym approaches are now invariably on or above profile and 5kts above normal approach speed. So he at least has learned something quite valuable.)