They seemingly tend to crop out gender, socio-economic standings, and ethnicity all of which can alter the data. Although they're not used I believe due to scope and variability. It is also noted there are external influences which could in theory "fudge" the data. At least as far as I see it. There also seems to be a tie to smoking and alcohol which perhaps links the two risky behaviors together. An abstaining of one, leading to an abstaining in the other. Which leads in part to what others are saying about other health issues not being tracked or noted.
But ultimately I would say addiction, as a disease, is quite abstract. And it's very social in nature. And that many cultures spread it through coming-of-age rituals. Which makes it difficult to villainize. But also difficult to "pin down" when an activity reaches an unhealthy level. Also people lie. So all of this could lead to fudged data as well.