The authors of one such article replied in letters in the Lancet that they saw no cut-off of the advantage of training beyond the 50 mins per day reported in their study. Their data supported the advantage of up to 120 mins per day of strenuous exercise on mortality rate, ie the more you do the better it gets, though the curve starts to flatten out. (They didn't have the data to say further than that.)
This does not contradict the possibility that a few rare cases of death around intense exercise will occur, possibly due to a prior enlarged heart, for instance. The same authors thought it a bad policy to put the general population off exercising because of a few such cases. The risks of not exercising are much bigger.
People need to get the idea that what is good for the general population may be bad for a couple of rare people but that should not be seen as giving people a get-out. If adopted by everyone, swimming regularly would be good for our general risks of death, even if a few we couldn't identify in advance sank because they were born with a brick in their head. Naturally, it would be good if we could put more work into identiying those few people at risk! Meanwhile, the obesity epidemic is costing us in health services about the same as smoking and each such individual is losing several years of life.
Edited: 03/12/2012 at 07:58