Something else that seems to be overlooked here is that the cohort were folks who already had heart disease. Since smoking is heavily correlated with poor cardi

Something else that seems to be overlooked here is that the cohort were folks who already had heart disease. Since smoking is heavily correlated with poor cardiac outcomes, I imagine a significant swath of that 6,837 cohort were likely to be smokers. Even if they aren't smokers, taking an already sick test group and trying to draw general conclusions about certain inputs is madness. Epidemiology can be a great way to form hypotheses, but this entire discipline of research has been misapplied for the better part of a century now.