A question about Heroin addiction

[bcubed]

I have seen the addiction rates for herion quoted (in particular, by Francis Moraes in book and 'net) as 15-20%. The point in stating this, I suspect, is to show that the public conception that anyone who uses H is a junkie is a misconception.

However, I thought about this relative to alcohol (where the addiction rate is 4-5%). No drinker STARTS OUT addicted: they might drink for years before this, and after getting addicted, they dont necessarially stay that way indefinitely: they could sober up, drink without addicition ("chip" you could say) or die, ending addiction. Thus, the % of drinkers who will ever EXPERIENCE addiction is quite different than the % who are addicted at any point in time).

Obviously, I took this thought and considered H. If the 15-20% is accurate, how many % of starting H users ultimately become addicted at some point? This might be more useful to somone considering H. It would also be useful to know the median length of addiction. (A substance that addicts 60% of users, but typically only for a month of their time, might be preferable to one that addicts just 40%, but for a peroid of five years).

Any ideas, or links, on these concepts?