If factors A and B are correlated on non-randomized sample, that only implies that And B belong to the common web of causes of effects, not that A causes B (since there can be a factor C that causes A and B). E.g. it could be that those who run more are precisely those who are have better health or whose biological age is genetically lower than their chronological age etc.
Another common feature of such junk science is that they always blame the victim for whatever problem they focusing on. And as with smokers and those of higher weight than what is officially declared to be "normal", that means they are setting up a pretext to tax you extra for your "sin".
It's the same old scam that medieval churches ran for centuries with selling of indulgences, except that language was modernized for the church of scientism (i.e. pseudo-science).