>
conflating pre-existing factors (already owning a dog) to interventions (going out and buying a dog to try to fix something).This is a really good point.
We often talk about confusing causation with common causality. (I.e. "active people have lower cardio risk and also get dogs" vs. "dogs lower cardio risk"). That's a good thing to check for, and essential to lots of debates. (For example, early data on charter school performance proved hugely distorted by the fact that educationally-engaged parents are the ones who look into charter schools.)
But we tend to forget the subsequent questions: even if an effect is causative in the wild, does that mean it works as intervention? And even if it is an intervention for some people, have the people it helps already found it? (As an extreme example, hypothyroidism is causative in fatigue and depression. Iodized salt is preventative, but not an intervention for severe deficiency - and doctors already check for treating every depressed person with iodine or thyroid hormones would be a terrible approach.)
Dog ownership is probably both common-cause and causative for cardio health. People who walk/run often are more likely to get a dog since their habits accommodate that, but also a dog is likely to push people towards exercise they wouldn't have gotten. As an intervention, though, it's a lot harder to say. Getting a dog changed my exercise behavior for the better specifically in the marginal cases where I would have almost run/walked, then yielded to inertia or poor scheduling. For people who don't lie near that margin, it's much less obvious what effect a dog might have.