Accounting for confounders is hard. Otherwise randomised controlled studies wouldn't be needed, and we'd not have taken this long to walk back the consensus that red meat causes cancer.