Yep, you're exhibiting precisely that kind of bias. I've known a variety of addicts and there's absolutely no statement one could make about them with 99% certainty that it will apply to a given individual.

I don't mean that one should be naive about things. You might correctly observe that a high percentage of Oxycontin addicts exhibit behaviour X, and therefor you might want to keep a wary eye out for that behaviour when dealing with an individual Oxycontin addict. Of course that way lies confirmation bias, which one must also guard against. But it's still better than saying that because most Oxy addicts exhibit trait X, this individual addict will necessarily exhibit trait X. That's worse than confirmation bias becuause it's pre-confirmation bias. You're treating statistics about a group as evidence about an individual -- and that's wrong.