I think I see the point you are making now. If HIIT to a greater degree prompts less fit participants to drop out than in the other groups, it might skew the results. Not really sure in what direction though. If it had been a study on the effect of HIIT vs less intense training on cardiovascular fitness, this kind of dropout would bias the results in the direction of HIIT. I have no idea of what effect it would have when the variably of interest is mortality.