That is usually about the size of it.

 A Big Mac with supersize fries, an apple pie with a 32 oz soda. And the conclusion is the meat caused the heart attack.

There is also the healthy user bias. Is a person that tends to follow food consumption guidelines for their health more or less likely to follow other health benefits lifestyle options like exercise and whole foods over processes foods? Or is it the Big Mac a day person less likely to exercise and eat better? Healthy user bias often creeps into food questionnaire type studies.

For it to be meaningful, a food study needs to be controlled. And it needs to be very long term. That is about impossible both from a cost and human standpoint.

Epidemiological studies can be useful for finding correlations. A step researchers can use to refine hypotheses and design controlled trials where they can tease out the confounding variables.

Correlation is not the same as causation.