We agree with Macleod & Hickman that the relationship between cannabis use and psychosis has been given undue prominence in the recent UK policy debate about the reclassification of cannabis from Class B to Class C in 2002 and back to Class B in 2008 [1]. From a public health perspective, psychosis is a relatively rare consequence of cannabis use (assuming that it is causal), with a risk of approximately one in 50 among regular cannabis users compared with approximately one in 100 among non-users. Psychosis is also an outcome that is difficult to prevent: very large numbers of potential cannabis users would need to be deterred from using in order to prevent one case of schizophrenia [2]. Further, the evidence suggests that simply reducing or increasing the severity of criminal penalties for cannabis use has little, if any, effect on rates of cannabis use [3].

We disagree, however, that it would be premature to act on the available evidence because of the possibility that the association between cannabis use and psychosis is explained by uncontrolled residual confounding. A causal relationship between cannabis use and psychotic disorders may not be proven ‘beyond reasonable doubt’, but in the absence of any specification of plausible uncontrolled confounders there are good reasons for believing that cannabis is much more likely than not to be a contributory cause of these disorders [4–7]. The epidemiological evidence and the biological plausibility of the relationship are strong enough to warrant giving advice to young people about this possible risk, along with information on other potential adverse effects of cannabis. Although the risk of psychosis may be modest from a public health viewpoint, the potential effects of a psychotic illness on a young person's life chances are so substantial that it would be socially irresponsible not to do so [8].

We also agree that there are more prevalent adverse health effects of cannabis use that warrant greater efforts to discourage its use by young people. We would, however, give much greater priority than do MacLeod & Hickman to cannabis dependence. It affects approximately one in 10 life-time users and one in six who initiate use during adolescence [9], and rates of treatment seeking for problematic cannabis use have increased in many developed countries over the past few decades. This increase is not solely an artefact of increased diversion of arrested cannabis users into treatment, because the same trend has been observed in the Netherlands, where cannabis use has been decriminalized de facto since 1976 [10]. Most critically, it is among cannabis-dependent individuals that many of the adverse physical (e.g. respiratory) and psychosocial outcomes attributed to cannabis use (e.g. educational underachievement, use of other illicit drugs, and depression) are found most commonly [11].

There are good arguments for treating cannabis use as a public health issue rather than one for the criminal justice system [12]; but there are also major policy impediments to this happening that have probably played an important, if less obvious, role in the recent UK debate about cannabis reclassification. These include: international drug control treaties specifying that signatory countries must criminalize cannabis use, and public opposition to the liberalization of penalties for cannabis use [13], despite evidence that removing criminal penalties for cannabis use has minimal, if any, effects on rates of use [3].

Finally, we agree with MacLeod & Hickman that the role of evidence in cannabis policy debates in many countries is constrained by a powerful implicit media simplification of the central policy issue; namely, that cannabis use is either harmless, in which case its use should be decriminalized, or it is harmful, in which case cannabis use should continue to criminalized [14]. This simplification is encouraged by advocates of cannabis law reform arguing that cannabis use should be decriminalized because it causes little, if any, harm to users. As long as this simplistic policy framing exists, any credible evidence that cannabis use causes harm will be given great weight by opponents of more liberal cannabis policies.

Declarations of interest