Over the years, in various jobs, I've used a mathematical tool called "regression analysis." This tool is at the root of almost all of these epidemiological studies where data is gathered from hundreds of people and then the main variable (the number of breast cancer cases per 100 people) is correlated, using this tool, against all the input variables, one of them being, in this case, smoked food.
There are multiple problems with these studies, and none of them are minor. First of all, you can never know everything about everyone, and your study may forget to ask questions about things which might have influenced the outcome. Second, people don't answer completely, truthfully, and sometimes just plain forget stuff. Third, the regression tool is almost impossible to use without introducing your own biases. To use the tool, you have to create a model based on how you assume things work. You then run the math, and the model gives you answers as to which things have the highest correlation to the outcome. The problem is that you often get very weak or almost no correlation to anything, something which renders your research useless. At that point, it is very tempting to go back to the regression model and tweak it until correlations start to appear. While this might at first seem like a reasonable thing to do, it is actually completely and totally wrong because it means you no longer have a hypothetical model, and you are simply twiddling "knobs" in the regression model until correlations appear.
It is well known in this field that you can tweak a regression model until you get seeming correlation with data which is produced with a random number generator!!
So, as already pointed out, studies like this should be taken with a large block of salt, if not entirely ignored.
I'm getting old enough to have seen most of these sorts of studies completely contradicted at a later time, including:
Eat margarine, not butter; oops, we now think margarine is bad, so eat butter instead.
Eggs increase serum cholesterol, so don't eat eggs; oops, we now think serum cholesterol is elevated by the body chemistry, and the cholesterol in the food you eat contributes almost nothing.
I could fill up several more pages with these things.
In the end, I always come back to that great scene in the old Woody Allen movie, "Sleeper:" If you've never seen this, watch the first thirty seconds of this clip. You'll love it.