If you interpret the arrows not as logical implications (“If X smokes, then X gets heart disease”) but as conditional probabilities (“The likelihood of X getting heart disease given that X is a smoker is higher than the likelihood of X getting heart disease given that he is not a smoker”),
