> This seems like an extraordinary claim, given that safety systems in all other places have no such effects (or at least, not as significant).

There are examples where safety systems cause increased risk.

Football helmets are one. The helmet doesn't protect very well against concussions and the presence of the helmet makes it more likely for players to use their heads as a weapon.

I don't know about this bike helmet example since the actual data isn't available, but there are definitely times when counterintuitively having a safety feature is riskier.