There's a strawman-tilting screed up over at substance.com from my current favorite anti-drug-war-warrior Maia Szalavitz. She's trying to assert that Trying to Scare Teens Away From Drugs Doesn’t Work.
In this she cites a few outcome studies of interventions that last over relatively short periods of time and address relatively small populations. I think the most truthful thing in her article is probably contained in this quote:
Another study, which used more reliable state data from the CDC’s Youth Risk Behavior Survey, concluded that “When accounting for a preexisting downward trend in meth use, effects [of the Montana Meth Project] on meth use are statistically indistinguishable from zero.”
This points out the difficulty in determining broad, population based outcomes from either personal introspection (where a lot of the suspicion about anti-drug messaging comes from, let's face it) or rather limited interventions. Our public policy goals are broad- we want to affect entire national populations...or at least state populations. In my view, we need to examine when broad national popular behavior shifted, if it did, if we want to understand how to affect it in the future.
The following originally appeared 21 July 2008.
If you are a reader of my posts on drug abuse science you will have noticed that it rarely takes long for a commenter or three to opine some version of "The (US) War on Drugs is a complete and utter failure". Similarly, while Big Eddie mostly comments on the liberty aspects (rather than the effectiveness) of the WoD himself, a commenter to his posts will usually weigh in, commenting to a similar effect.
Now I'm open to all the arguments about personal liberty trade offs, economic costs, sentencing disparities, violations of other sovereign nations and the like. Nevertheless, I'm most interested in the fundamental question of whether the War on Drugs worked. That is, to reduce drug use in the US. For those who believe it has not worked, I have a few figures I would like explained to me.
I'm following up a story I started in a prior post by putting up the long term trends for cocaine use in the US. These data are from the 2006 Volume II monograph which focuses on the 18 yr old and older populations. As you will recall my hypothesis was / is that the Len Bias fatality had a dramatic effect on cocaine use. I still think this is the case and that this explains much of the timing of a reduction in cocaine prevalence observed consistently from the 18 yr old to 45+ age groups. However Len Bias's death was not an exclusive effect and must be considered in the context of changes in other drug use patterns. That context is something I want to delve into just a little bit.
As always, I depend on the data from the Monitoring the Future survey (www.monitoringthefuture.org) and I am pulling the figures from the 2006 Volume I monograph which focuses on the 8th, 10th and 12th grade populations in contrast to the older age cohorts outlined in the first graph.
First up are the annual prevalence rates for powder cocaine, which I provide for reference to the previous graph for the older age ranges. I apologize for the blurry figures but my imaging skills are not up to any better- luckily, these reports are freely available on the MtF website. (I also encourage you to get the reports yourself because there are slight changes in the questions asked in some cases- if you see a discontinuity in the longitudinal data this is probably why.) The longest term trends are available for 12th graders, additional grades were added into the survey in the early 1990's. Prevalence of cocaine was reasonably steady in the 1979-1986 interval and it is stunningly apparent that cocaine became less popular with 12th graders after 1986 . It is also clear that it took about 5 additional years for prevalence to drop to the most recent nadir. So it wasn't all about Len Bias (he died of cocaine-related cardiac complications on June 19, 1986).
So, if it isn't all about Len Bias, perhaps we should see similar effects on population prevalence of other illicit drugs?
Marijuana and Amphetamine
It seems reasonable to turn our analysis to two perennial high-prevalence drugs for high school populations; marijuana (duh!) and the amphetamines. (In MtF parlance, the amphetamine class is for tablet or other prescription preparations after 1982.) In this case, the prevalences were at peak in the late 1970s and started to decline in the very early 1980s. Interestingly, there is no evidence of a change in the established trends from 1986-1987 as is observed for powder cocaine; I think this supports the Len Bias hypothesis. Nevertheless we can also see this as additional evidence for something else driving drug use downward.
This brings us to what are illicit drugs for most of these populations but, of course, licit drugs for individuals who have reached the legal age; 21 (alcohol) or 18 (cigarettes; this may be a substantial fraction of 12th graders). In theory, we might use these data to try to dissociate the anti-drug messaging from the drug interdiction / legal penalties side of the equation. Not perfect, but at least a hint.
The trends for annual prevalence of alcohol were very stable from 1978-1988 whereupon a decline was observed (questions were altered in 1993, making further comparison tricky). The trends for 5-drinks-in-a-row (currently the definition of a "binge") in the past two week interval were very stable from 1978-1983 and thereafter exhibited a slow decline until the early 1990s. Very reminiscent of the above mentioned drugs.
In this case, please note that we've shifted to 30-day prevalence rates (any, daily); obviously this is frustrating for direct comparison but this is what they provide in the monographs. Unfortunately the more recent monographs (it is currently on a reliable annual update schedule with available pdfs, the older ones are not available) seem to only start with the 1986 data in the Tables so one is left with their figures for the earlier part of the trends. With that caveat, we can see that cigarette prevalence in the high school population was reasonably stable during the interval in which the prevalence rates for the illicit-for-all drugs mentioned above were in decline.
So Did the War on Drugs Work or Not?
I do think the jury is still out on this one and the problem of shifting definitions about goals and successes is quite difficult. I feel confident the comments will stray afield a bit and explore some of these issues. However, as I intimated at the outset,
for those of you who insist vociferously that the War on Drugs (considered inclusively with the Just Say No, D.A.R.E, main-stream media reporting, and all that stuff that is frequently rolled into a whole by the legalization crowd) is an abject failure...for those of you who insist vociferously that you cannot tell teenagers anything about the dangers of recreational drugs and expect them to listen to you...
I would like these data explained to me.
Update 7/23/08: Followup post from Scott Morgan at StoptheDrugWar.org
Updating the data/charts referred to in this post: The recent trends for marijuana use in the US haven't experienced any dramatic change from the levels reached in the 1990s after the great rebound. The Annual Use [PDF] and Daily Use [PDF] charts for 8, 10 and 12 graders from Monitoring the Future show us that. The trend for alcohol [PDF] and cigarette [PDF] use in the past 30 days continues to trend downward, with only a slight blip in the early 90s, instead of the lasting directional change for marijuana. As always, go over to the Monitoring the Future site to look up your favorite drugs for yourself.
The linked PDFs combine the prevalence charts with ratings of Risk, Disapproval and Availability. I would draw your attention to the way estimates of Risk for trying a psychotropic drug once or twice correlate with substance use. This, and the aforementioned Len Bias situation, address Maia's thesis in a way.
It isn't that scaring people off drugs doesn't work. It's just that they have to actually be scared.
On a conciliatory note, I agree emphatically with Maia's point here:
Let’s face it: The real problem with teen marijuana use is not a high school kid smoking a joint on a Friday night—it’s a kid who can’t get through the school day without one.
Those MtF data [PDF] show that about 7% of high-school seniors are using marijuana almost* daily.
A typical US public high school classroom will have what, 40 kids in it? 30-50 as a range estimate? So your average kid is in class, easily, with 99 other kids. SEVEN of them are smoking pot every damn day.
And being noddingly familiar with the profile, that "daily" is going to mean all day, every day in a significant subset of those individuals. Compare those daily rates to the annual rates across time. Notice how they trend along with each other?
Remember that when some cannabis denialist type comes along and tries to claim that wider spread availability and incidence doesn't matter because it is only the "addictive personality" that drives compulsive marijuana use and "everybody that wants to smoke marijuana already does anyway".
We may not be able to draw an absolutely direct line between daily use, addiction and a serious drug problem but they sure as hell are correlated.
Legalization, medical marijuana woo and cannabis denialism in general reduce the perception of risk. This drives greater use. Broader availability drives broader use. More use gets you to more daily use. More daily use leads to dependence. Dependence increases the incidence of people with a cannabis problem, no matter how an individual construes that.
Conversely, people being scared** of bad outcomes reduces use, therefore reduces the incidence of frequent use, dependence and, ultimately, the individuals with a cannabis problem.
This is the hypothesis the actual long term epidemiological evidence supports.
The fact that it doesn't tell us how to actually scare** people off a given substance doesn't affect this reality.
*25 of 30 days IIRC.
**so to speak