They said they controlled for those confounding variables. I bet the raw data used in this study also says smokers and drinkers are more often sick than the others.