23narchy in the UK

Ben Goldacre, Saturday 7 November 2009, The Guardian

Obviously it’s pleasing to see, in the storm of commentary over Professor Nutt’s sacking, that everyone outside of politics now recognises the importance of scientific evidence in devising laws. But a strange reasoning twitch has appeared, in the arguments of politicians and right wing commentators. Science can tell us about the molecules, they say, about their effect on the body, and the risks. But policy is a separate domain: a matter for judgement calls on social and ethical issues. Only politicians, they say, can determine the correct way to send out a clear message to the public. It is not a matter for science.

Interestingly this is wrong. Alongside research into the risks of drugs, lots of research has also been done examining the deterrent impact of different laws, classifications, and levels of enforcement. Since every piece of research has its own imperfections (and nobody has yet conducted a randomised controlled trial on drugs policy) you can make your own mind up about whether you find this research compelling.

One strategy is to compare different countries. A World Health Organisation study from 2008, published in the academic journal PLOS Medicine, compared drug use and enforcement regimes around the globe. It was clear: “globally, drug use is not distributed evenly and is not simply related to drug policy, since countries with stringent user-level illegal drug policies did not have lower levels of use than countries with liberal ones.”

Alternatively you can compare drug use between states within one country, if they have very different enforcement regimes, as happened when some parts of the US liberalised their laws a few decades ago. In 1976 Stuart and colleagues found that cannabis use in Ann Arbor, Michigan, wasn’t affected by reductions in cannabis penalties, when compared with three neighbouring communities which kept penalties the same. In 1981 Saveland & Bray looked at national drug use surveys from 1972 to 1977 and found that cannabis use was higher in the ‘decriminalised’ states, both before and after the changes in law, and when they looked at rates of change, although cannabis use was increasing everywhere, the most rapid increase was actually in the states with the most severe penalties. In the same year Johnson and colleagues used survey data on high school use and found decriminalisation had no effect on attitudes or beliefs about drugs. These studies are old, of course, but only because the liberalisations in the law which they rely on for data happened a long time ago.

Another line of evidence comes from “before and after” studies, when laws are changed. Cannabis use in the UK dropped, of course, after cannabis was moved from class B to class C. Prohibition of alcohol in the USA from 1920 to 1933 is the most famous example: here, alcohol use fell dramatically when prohibition began, and the price of alcohol rose to 318% of its previous level. But by 1929, this initial impact had begun to wear off, and rapidly: alcohol consumption had risen to 70% of pre-prohibition levels, was still rising when prohibition was repealed, and the price had fallen to only 171% of pre-prohibition levels. Notably, this reversion to old patterns of use occurred despite escalating expenditure on enforcement, which rose by 600% over the same period. There are many more examples.

This is not an unresearchable question. It is clear that there are many other factors at play in all of these studies, and if they are not sufficiently rigorous for the government, or a brief informal dip into the literature is not enough (it shouldn’t be) then they should commission more formal research: because it is a basic tenet of evidence based policy that if you discover a gap, you flag it up, and commission more work to fill it.