that in itself could be healthy user bias (if a healthier subset was taking up the vaccine).

did they control for that?

Its a cohort study, so you can only control for confounders. The 2nd paragraph of the discussion addresses the healthy-vaccinee effect you're referring to.