The researchers typically control for variables that are known, such as smoking, overweight, etc. Once you control for them, the association between alcohol and cancer becomes even weaker. I used the unadjusted correlations. Had I used the adjusted ones, the relative risks would be even less. The reason why I didn’t use the adjusted correlations is simple: the detailed data that would allow me to derive the hazard and odds ratios aren’t made available, neither in the papers, nod in the supplemental material