Cannabis and public health: a global experiment without control

Every few weeks, new findings on the effects of legalizing recreational cannabis use are published. Thus, the review of Hall and Lynskey ¹ – or any review for that matter – can only provide a preliminary summary of the collected evidence to date.

Looking into public health effects of le­galization, two seemingly easy indicators may be prevalence and patterns of canna­bis use, as both are potentially linked to health and social problems in the presence or absence of legalization ² . However, the main source for both indicators are surveys, with their severe limitations, as today's surveys are neither based on representative sampling frames nor on high response rates. In addition, in the case of cannabis, we are dealing with a (formerly) illicit and stigmatized substance, making comparisons over time even more challenging. More reliable measures such as wastewater analyses are needed here, but these measures cannot assess patterns of use or individual behaviors.

There are good indications that cannabis‐related hospital (emergency rooms, psychiatric wards) admissions have increased in legalizing US states, possibly driven by an increase in frequent use. In addition to further monitoring these trends in harms, Hall and Lynskey ¹ suggest tracking treatment demand to allow for a short‐term evaluation of legalization's effects. However, as the authors acknowledge, there are many confounding determinants involved in such evaluations, including access to and availability of treatment, coercion, potency of used products, use of synthetic cannabinoids, help‐seeking behavior, stigmatization and public perception of cannabis use and associated problems. For instance, despite the liberalization of cannabis in Canada, with increases in use over the past years, there have been decreases in treatment rates, partly because liberalization seems to have led to higher thresholds for treatment seeking, as use per se is no longer considered problematic ³ . As long as these confounding determinants cannot be disentangled, treatment demand data should probably not be used as an indicator to evaluate the public health effects of legalization.

We also disagree with the statement that it is too early to evaluate the consequences of legalization on the legal system. We do not see why other domains can be evaluated now but not this one, one which has been put forth as the main argument by opponents of prohibition, and could quite easily be measured.

However, there are some general limitations in any evaluation at this point, especially since circumstances associated with legalization seem to be changing rapidly. In other words, legalization is not a clearly defined phenomenon, because it takes myr­iad forms on a spectrum ranging from tight control to open markets, even within a coun­try (such as in the US or Canada, where states or provinces decide on implementation).

Looking at the evidence gathered in the US so far, it becomes apparent that most evaluations will fail to identify causal determinants. For instance, if a potential rise in traffic fatalities in legalizing states is found, it may be attributed to “legalization” per se, when the underlying reason may in fact be a greater impairment of the drivers due to an increased use of high‐potency products – a phenomenon only accelerated by legal­ization.

Thus, identifying causal agents and processes poses methodological challenges which may not be overcome easily in analyzing natural, large‐scale experiments. To improve understanding of the effects of le­galization, we strongly advocate for small‐scale, controlled experiments, such as those proposed in the city of Berlin, Germany ⁴ . There, the effects of legal access to cannabis are to be studied in a restricted sample of registered users, while users without legal access serve as controls. Such experiments – limited both spatially and temporally – will allow researchers to examine how increased availability impacts on consumption patterns and related risks in greater detail, and thereby provide an evidence base for formulating large‐scale regulation models.

In any experiment, pre‐defined outcomes (e.g., changes in cannabis‐related arrests) may be in the limelight, but unin­tended consequences should not be ignored. One prime example is the dramatic increase of tetrahydrocannabinol (THC) exposure in North America, driven by new products and modes of administration, which facilitate the intake of higher doses of THC compared to, for instance, smoking cannabis in a joint.

Specifically, oil cartridges can contain several hundreds of doses of THC, and regular users may use more of those cartridges a day. As such products are more widely available in jurisdictions with legal access to cannabis, and as THC has been linked to cognitive impairment, use disorder severity, and psychotic symptoms, the catalyzing effects of legal cannabis markets with regard to THC exposure should be thoroughly evaluated and compared to illegal markets. The main barrier here, however, would be to obtain reliable and comparable estimates for the control group (with no access to legal cannabis). Again, these methodological limitations reiterate the need for small‐scale, more tentatively conducted ex­periments.

Active experimentation is in line with Campbell's vision of an experimenting society to solve complex problems ⁵ . Part of this vision is a more active role for policy formulation, but also some clear empirical principles for evaluation. If experiments such as cannabis legalization fall short according to pre‐determined criteria, societies should be able to adapt and change directions.

Largely irrespective of the evidence collected so far, current politics seem to be final and one‐directional: once a widening of the cannabis market via more liberal medical marijuana policies or via legalization is sanctioned, market forces seem to be the sole drivers of the future course, and mainly fueled by the desire to increase revenues and shareholder values.

This development also extends to low‐ and middle‐income countries ⁶ . Thailand provides a prime example. In this society with less than 1% prevalence of cannabis use, medical marijuana has been introduced, and the government announced future legalization of recreational canna­bis use based on unrealistic claims of mas­sive income benefits for households grow­ing and selling cannabis to industry ⁶ .

As a result of these market utopias, rational exploration of alternative governance models – which are more public health oriented – stand little chance. Canada provides a good example here: what started as “legalization with strict control” has evolved into quick increases in availability and looser controls, led entirely by market forces. With a legalized market now in place, the illegal market is still thriving, and there is no sign that it will cease to⁷.

For example, in Canada's major magazine Maclean's, an article published one year after legalization contrasted the purchase of legal cannabis (taxed, more expensive, of lower quality) in a li­censed store at considerable distance with “placing a delivery order from my friendly, local unlicensed shop; they take credit card payment at the door, I can redeem loyalty points, it's less expensive, and the weed? Well, it's dank” ⁸ . The existence of such options was confirmed by federal police, who warned about continued illegal sales for just such reasons: home delivery, options for credit card payment, and nation‐wide shipping, in addition to often significantly lower prices ⁷ . Yet there is no push by government to enforce business practices for the legal options.

In this situation, all that seems to be left for public health is to document the consequences of these developments. In this respect, contributions like the review of Hall and Lynskey ¹ are important, but they also show the difficulties in arriving at any firm conclusions.