That article does a terribly job of reporting the actual study. You can find the full thing, plus a file of supplementary data, on this page:
http://jech.bmj.com/content/early/recent(at least, I can: it may be that I'm silently getting past a paywall by virtue of being at work. If so I won't distribute copies, as that would be illegal, but I'm happy to look up the answers to questions.)
At a quick read, they've got an interesting, apparently solid study that does show the effects summarised. They report data by IQ sextile (so the highest sextile is those with IQs of 130+) and it does seem to show that the likelihood of using illegal drugs goes up with IQ (even after adjusting for social class, income etc.) e.g. is higher in those with 130+ than in the next group down etc., not just that those at 107+ are more at risk than those at 93- (or whatever it was). This is interesting and potentially worrying. However, looking at the tertile data, the difference between the risk in the lowest tertile and the middle one is typically much bigger than the difference in risk between the middle and the highest tertile. IOW, assuming one were not terribly worried about the risks for an average-IQ child, one shouldn't be all that much more worried about the risks for a 130+ child.
It would, of course, be interesting to know what happens further out on "our" tail, but I don't think they have that data (you can imagine the difficulties).
