[ajross]

Right, or that there's some other confounding factor that leads smokers to be less likely to contract the disease or seek treatment. This kind of science (teasing data out of uncontrolled sampling) is really hard.