Dynamic Drug Policy

An editorial in the new issue of Addiction questions the static drug policy models that dominate current policy debates:

Drug researchers have long understood that there can be long-term waves of greater and lesser drug use and that upswings can involve epidemic-like spread. These and other dynamics discussed below imply that policy ought to vary over the course of a drug use cycle, but drug policy debates have not yet internalized this perspective.

Because drug problems vary in these complex ways, it seems plausible that drug policy should vary over time as well; yet it is rare to hear someone couch their drug policy recommendations in these terms. This is striking and more than a little troubling. It suggests that the mental models guiding policy discussions implicitly superimpose a static framework on an intrinsically dynamic phenomenon, akin to popular nostrums for get-rich-quick investing that never vary even as economic conditions change over the business cycle.

It is not clear why policy is not discussed more often in dynamic terms. Perhaps disciplinary boundaries and stove-piped bureaucracies create single-issue advocacy. Perhaps both the health and criminal justice perspectives favour individual-level analyses. Whatever the reasons for their absence to date, dynamic perspectives on drug policy are, in fact, possible.

What are the policy implications? The author discusses the stage-specific limitations and strengths of several approaches:

Preventing an initiation in the early stages of an epidemic is tremendously valuable, because it short-circuits a chain reaction that would have involved many people. (In technical terms, the reproductive rate at that point would have been large.) However, primary prevention cannot be timed to react to a burgeoning epidemic because of intrinsic lags. For example, the median age of cocaine initiation in the US is 21 years, but students in school-based prevention programmes are younger, often only 13 years old. Therefore, if school-based prevention interventions were to have any hope of affecting cocaine initiation dramatically, the ideal time to have run them would have been in the early 1970s, 8 years before the peak in initiation. However, no one knew in 1970 that there was a cocaine epidemic brewing. Conversely, the vast majority of cocaine consumed from 1985 to 2005 was consumed by people who were already older than 13 years in 1985. For instance, over 85% of people the Treatment Episode Data Set (TEDS) records as receiving treatment for cocaine between 1992 and 2003 were born before 1973. Consequently, prevention programmes initiated around the time the cocaine epidemic became salient could not possibly have had a dramatic effect on use over the next generation, regardless of how effective they were.

Treatment also has limited ability to stave off a burgeoning epidemic, because early in the epidemic most users do not have a treatable medical condition. Precise estimates are not available because population-level estimates of treatment need exist only for recent years, but need for treatment is correlated with average duration of use. In 2003, 39% of respondents reporting past-year cocaine use to the US Household Survey had been using for 10 or more years. In 1979, the peak year for cocaine initiation, that proportion was just 3%. For drugs that are not injected, the role of harm reduction strategies is similarly limited when most users are not experiencing significant harms with their use.

Enforcement’s effectiveness at suppressing drug use declines markedly as the size of a drug market grows. However, enforcement has unique ability to focus its effects in both space and time. If a crack house opened next door, neither funding school-based prevention nor additional treatment slots would bring rapid relief. Parking a patrol car in front of the crack house would at least displace the activity. Similarly, assume treatment was five times more cost effective than incarceration at reducing drug use. Incarceration could still be twice as cost effective at reducing drug use this year—because incarceration’s effects on drug use are concentrated in the present whereas treatment’s effects may be spread over a decade or more. Hence, these models suggest that supply control programmes may have a unique capacity to disrupt the contagious spread of a new drug, but limited ability to eradicate established markets. (Enforcement may also be able to displace established markets into less destructive forms, such as forcing visible street dealing to convert to discreet meetings arranged by cell phone.)

Harm reduction offers particular advantages later in the epidemic cycle, when use has stabilized at high endemic levels. For injectable drugs in countries with low violence and few street markets, harm reduction may focus on syringe exchange programmes, supervised injection rooms and training ambulance crews to treat overdose. For drugs that are not injected and which are supplied through violent street markets, harm reduction may focus instead on using enforcement to target the minority of dealers who cause the greatest social harm. In either case the premise is that, with or without the harm reduction, the flow of new people into problem drug use will be modest, so reducing harmfulness of drug use has few drawbacks. That may not be a safe premise early in an epidemic, when there are feedbacks that can amplify small shocks to the system into dramatic effects on its trajectory.

I’m not sure I agree with the author’s assumptions about the strengths and limitations of he various approaches, but he makes a compelling argument:

  • that the static models currently being debated and advanced all have their place in policy;
  • they are all inadequate by themselves;
  • and, that there isn’t even a correct formula because the needs, opportunities, crises, etc. are constantly changing.