How Cannabis Is Legalised Matters More Than Whether It Is

For the governments across Latin America, Asia and Africa now weighing whether to loosen their cannabis laws, a major international review delivers a pointed message: the design of a legalisation framework shapes public health far more than the decision to reform at all.

That is the central finding of a narrative review published in The Lancet Psychiatry by researchers at the University of Bath, who pulled together 25 years of evidence from the Americas, Europe, Africa, Oceania and Asia. Led by Professor Tom Freeman and Dr Rachel Lees Thorne with an international team of collaborators, the study tracked three measures across countries that changed their laws between 2000 and 2025: how common cannabis use became, how often use tipped into addiction, and hospital admissions for psychiatric conditions. The authors stress it is a review of natural experiments — it identifies associations between policy and outcome, not proof of cause.

The reference point is Uruguay. In December 2013 it became the first country in the world to legalise adult-use cannabis. It did so not as a liberal experiment but as a security measure, after President José Mujica argued the region was losing its fight against drug trafficking. The International Narcotics Control Board warned at the time that the law would put Uruguay in breach of the 1961 Single Convention on Narcotic Drugs. More than a decade on, there is finally enough data to test who was right.

The commercial model carries the clearest risks

The strongest evidence in the review comes from North America, where for-profit retail markets are well established and tracking is best. In Canada, past-year cannabis use among people aged 16 and over rose from 22% before legalisation in 2018 to 27% by 2020, settling at 26% by 2023–24. Rates of cannabis use disorder climbed among adults, product potency rose after commercial sales began, and the number of daily or near-daily users in the United States overtook daily drinkers — a shift the authors treat as historically significant.

The psychiatric picture is more specific. In Ontario, emergency department visits for cannabis-induced psychosis rose more than 220% between 2014 and 2021, a trend that had started before legalisation but accelerated as stores multiplied and higher-potency products reached shelves. In Colorado, hospitalisations for patients with both psychosis and cannabis use disorder climbed from 3.4 to 8.5 per 100,000 after non-medical sales began. The review found no consistent evidence that legalisation changes the overall rate of psychotic disorders in a population, but commercial markets were tied to more psychosis-related hospital visits — particularly where psychosis occurred alongside addiction. Canada’s commercialised model is the most closely studied national example outside the United States.

Tightly controlled markets show little change

Where governments kept commercialisation out, the data looks different. Uruguay built a non-commercial system: pharmacy sales with potency caps, member-only social clubs and home cultivation, all overseen by the state regulator, with no advertising and no for-profit retail. Two studies found no significant rise — and in some measures a reduction — in use among young people, measured against a matched control group in Chile.

Other controlled frameworks point the same way. Germany’s interim evaluation, published in September 2025, found no link between its April 2024 legalisation and changes in consumption across three population surveys. A Swiss randomised trial that gave 378 participants either illegal-market access or regulated legal access found no significant difference in addiction scores after six months. In the UK, a period of reduced enforcement between 2004 and 2009 produced no substantial change in use, while hospital admissions for cannabis-induced psychosis fell.

The authors are cautious about these results. European datasets are thin and short — Germany’s covers only months, the Netherlands’ pilot has yet to report, and the Swiss trial ran for half a year. The evidence points to an absence of harm, they note, not a guaranteed absence of harm over time.

Thailand shows how fast it can turn

The starkest case is Thailand, which in 2022 became the first Asian country to legalise non-medical cannabis before moving to roll it back. A rush of dispensaries drove a five-fold increase in patients diagnosed with cannabis-related disorders; inpatient cases of cannabis-induced psychotic disorder rose from 477 in 2017 to 2,713 in 2023. In June 2025, the Ministry of Public Health began requiring a medical certificate for every purchase from a licensed shop — what the authors call an unprecedented test of whether a commercialised market can be reversed.

Medical access carries its own warning. It is now the dominant route to legal cannabis worldwide, yet the review flags poorly regulated programmes: Australia has issued more than 700,000 prescriber approvals across over 250 conditions, many with limited evidence behind them. Legal medical access helps when it rests on solid evidence, the authors conclude, but wider access without that foundation may deliver little benefit while raising harm.

For governments still deciding, the takeaway is about structure rather than a simple yes or no. “In a rapidly changing global cannabis policy landscape it is increasingly important to ask how policy will change, rather than if it will change at all,” said Freeman. The harder problem is generalisability: most of the high-quality evidence still comes from the United States and Canada — the commercial markets — which leaves thinner ground for the Latin American, African and Asian governments weighing reform amid shifting global attitudes toward cannabis, the very readers the review most wants to reach.

References:

^{1. Freeman, Tom P.; Lees Thorne, Rachel; et al., “International cannabis policies and their association with cannabis use, cannabis use disorder, and other psychiatric disorders” (2026). The Lancet Psychiatry, vol. 13. https://doi.org/10.1016/S2215-0366(26)00087-8}