It's worth following the link to the BMJ and reading the full article to see just how much of the data is "modelled", by which they mean, "assumed or made up".
But this is the same thing Wingpig linked to this morning. The trouble is these schemes don't save lives. At best they delay death. These people will all die some time. And that's the problem with them. Just like most health interventions they offer jam tomorrow. So when they say the health benefits outweigh the risks, they forget to mention that the health benefits (in terms of death delayed) accrue in some probabilistic future whereas the risks are faced today. When you present it like that it makes sense to stay off the bike and avoid today's risk even if it means foregoing some dubiously beneficial extra years at the end of your life.
That seems to me to be real public health challenge - for almost every area such as physical activity, smoking, drinking, diet - to get people to change their behaviour on a distant promise.