Transitioning to cannabis dependence

The conditional probability of dependence on a given drug is a question that is of substantial interest to users, parents of users, public policy makers and heath care providers. After all, if people simply stopped using a drug once a problem arises then many of the negative effects could be avoided. There is a fair degree of correlation between meeting diagnostic criteria for dependence and someone failing to stop using a drug despite clear and growing negative consequences. (Indeed this is one of the dependence criteria). Therefore, we must consider dependence to be a target of substantial interest.

It can be difficult to estimate the conditional probability of dependence in humans because we mostly have cross-sectional data to work with. And so we must infer conditional probability from dividing the currently dependent population by some denominator. Depending on what one uses for the denominator, this estimate can vary. Obviously you would like some population that uses the substance but what represents a level of “use” that is relevant? One time ever? Use in the past 12 months? Use in the past 30 days?

A new paper by van der Pol and colleagues uses a prospective design to provide additional data on this question.

The authors recruited 600 frequent cannabis users, aged 18-30, and assessed them for cannabis dependence at start, after 18 months and after 36 months using the:

Composite International Diagnostic Interview (CIDI) version 3.0 (Kessler and Ustun, 2004), and required the presence of three or more of seven symptoms within the 12-month period since the previous interview (without requiring the presence of all symptoms at the same time). It should be noted that the CIDI includes a withdrawal symptom, which is not included in the DSV-IV manual.

The study defined “frequent” use as 3 or more times per week for 12 months or more. This is important to remember when trying to assess the conditional probability. It all depends on what you construe as an at-risk population. Here, I’d say these were already rather confirmed cannabis fans.

The authors were interested in the very first incidence of dependence and so therefore excluded subjects who had ever met criteria, this left 269 subjects at intake (retention in the study left N=216 at 18 mo and N=199 at 36 mo). This is another point of interest to me and affects our estimation. Three or more times per week for 12 months or more and 45% of them had never previously met criteria for dependence. There are two ways to look at this. First, the fact that a lot of similarly screened users had already met criteria for dependence suggest that this remaining population was at high risk, merely waiting for the shoe to drop. Conversely it might be the case that these were the resistant individuals. The ones who were in some way buffered from the development of dependence. Can’t really tell from this design….it would be nice to see similar studies with various levels of prior cannabis use.

There were 73 cases of cannabis dependence of the 199 individuals who were followed all the way to 36 months, representing a conditional probability of transitioning to dependence of 36.7% within 3 years.

Now, of course the authors were interested in far more than the mere probability of meeting dependence criteria. They assessed a number of predictor variables to find differences between the individuals that met criteria and those that did not. Significant variables included living alone, mean number of prior cannabis use disorder symptoms, a continual smoking pattern per episode, using [also] during the daytime, using cannabis to “cope”, child abuse incidents, motor and attentional impulsivity and recent negative life events. For this latter, followup analysis identified major financial crisis and separation from someone important as driving events.

As the authors point out in the discussion, the predictors differ from those identified from a more general population. This makes sense if you consider that the range on numerous variables has been seriously restricted by their catchment criteria. The amount of cannabis exposure, for example, did not predict transition to dependence in this study–perhaps because it was well over the “necessary if not sufficient” threshold. This underlines my theme that the denominator matters a lot to our more colloquial estimates of the risks of dependence on cannabis.

Another issue identified in the discussion was the choice to start at 18 years of age for the captured population. Cannabis use frequently starts much earlier than this and many studies of epidemiology suggest that initiation of drug use in the early teens, mid teens, late teens and early twenties confers substantially different lifetime risk of dependence. “The earlier someone starts using, the more likely to become dependent” is the general findings. The authors cite a study showing that the mean age of meeting cannabis dependence criteria for the first time is 18. This is at least consistent with the fact that 65% of their collected sample had previously met criteria for dependence. No study is perfect or gives us the exact answer we are looking for, of course.

A final note on estimating the conditional probability of dependence in the population that uses cannabis 3 or more times per week for over a year. Of the original sample, 331 had already met dependence criteria and were excluded because the interest here was on the first time dependent. If we ignore those 70 people lost to followup during the study, and add the 73 to the 331 then we end up with 76% of those individuals smoking that much cannabis who have already, or will soon, meet dependence criteria.

__
van der Pol P, Liebregts N, de Graaf R, Korf DJ, van den Brink W, van Laar M. Predicting the transition from frequent cannabis use to cannabis dependence: A three-year prospective study. Drug Alcohol Depend. 2013 Jul 22. pii: S0376-8716(13)00228-7. doi: 10.1016/j.drugalcdep.2013.06.009. [Epub ahead of print]. [Publisher, PubMed]