You cannot compare a bullet passing through a heart - an entirely physical process - with diseases from cancer - complex biochemical processes. This is statistically and logically wrong - like comparing apples and oranges, although your example is even more wrong. Apples and oranges are at least fruits, but heart and cancer?
The entire field of ontology is complex in itself. I agree that alcohol CAN (!) increase the chance to get cancer, but - again - Paracelsus plays an imminent role, which is even not considered! Also possible interactions with all the other environmental, psychological etc variables are missing. This diminishes considerably the value of this study.
If I have time and the motivation, I could dig deeper into the design. But it seems to have remarkable flaws and thus makes it in the context of medicine and science a pub, which should be considered with caution!