I have a strong suspicion that this is a spurious conclusion, for the same reason that the "moderate drinking improves health" correlation was debunked: http://theconversation.com/maybe-moderate-drinking-isnt-so-g...

The study doesn't appear to control for differences in wealth. I would suspect that the population of people who bike to work skews wealthier (higher rents close to work centers, less likely to have to go from job to job).

I'm curious how many of these types of longitudinal studies are at their core just confirming the correlation between wealth and health.