Therein lies the problem it is nearly impossible to to make All or many things equal and why it is not that simple in determining a link between smoking and a given type of disease. There will always be differences between the groups (Called Confounders). Besides cigarette smoking smokers % wise generally live less healthy lives compared to nonsmokers. Smokers as a group are more likely % wise not to be physically active compared to the population as a whole (How many Marathon runners do you know smoke?), Smokers are also more likely not to "Eat well" and a higher percentage of poor people smoke compared to those that are well off. Those things by themselves will lead to a higher % of health problems regardless if the person smokes or not.
An example that jumps right out at me is Stomach Cancer which is relatively rare in the west except among people who abuse Alcohol. People who abuse Alcohol smoke at a higher % rate than the general population. So if you take a group of Stomach Cancer patients you are going to have a higher rate of Alcoholics in the group compared to a control group of non stomach cancer suffers and since you have a higher rate of Alcoholics you are going to have a higher rate of smokers. So if you do a study on smoking and stomach cancer and you use controls from the general population yes you are going to find a statistical link. However it is meaniless because of the confounder of Alcohol abuse.
Unfortunately the anti-smoking Nazis often use these confounder differencences between the control & trial group to get the results they want. I heard of a study (I believe done by Stan "The sham" Glantz) on second hand smoke where they used heart patients from a poor area as the trial group while using people from a well off neighborhood as the control which is total BS (Poor people regardless if the smoke or not have a higher rate of heart problems compared to people who live in well off areas).
group of a 100, say bladder cancer patients, and you notice that 60 or 70 of these patient report either currently smoking or past smoking, you have a statistical link.
Well you are never going to find those kind of numbers in most diseases they stick on smokers. It's more like out of a 100 people with the diseases they might find 28 are smokers where in the population of controls they find 25 of of a 100 and usually that small difference is because of a confounder or just by chance.