Countries — what worked, what didn't, what transferred, and what was political theatre

Reading frame

Three things to hold in mind reading this page.

One: country evidence is observational, not RCT. No country runs the controlled experiment where one half of the population gets the new policy and the other half does not. Every case study below is a difference-in-differences or interrupted-time-series exercise, with the limitations those methods carry. The strongest claims survive multiple analytic frames; the weaker ones depend on the specific analyst.

Two: country lessons transfer poorly without context. Portugal's 2001 decriminalisation worked partly because it was paired with sustained treatment funding for a decade. France's 1995 buprenorphine scale-up worked partly because France already had a strong primary-care infrastructure to push it through. Iceland's Planet Youth worked partly because Iceland already had unusually high social cohesion. Anyone proposing to copy the model elsewhere without copying the substrate is selling a slogan. The Treatments page makes this point at length about Portugal specifically.

Three: the willpower frame travels everywhere. Whatever the local intervention, the moralising frame is the local opposition. The country studies that follow are partly studies of substantive interventions and partly studies of how each country managed the moralising blocker. Singapore manages it by leaning all the way into the moralism. Portugal manages it by replacing one moralism with another. France manages it by routing around it through GP practice. Each is interesting.

Portugal — decriminalisation + funded treatment (the headline case)

The full version is on the Treatments page. The short version: in 2001, Portugal decriminalised possession of all illicit drugs for personal use (not legalised — possession is an administrative offence now, routed to the Comissão para a Dissuasão da Toxicodependência rather than to court). The reform was paired with major expansion of treatment infrastructure: needle exchange, OST, supervised consumption, housing-first programmes. Funded through the 2000s.

The Hughes & Stevens 2010 evaluation in the British Journal of Criminology found reductions in problematic use, drug-related harms, criminal-justice overcrowding, and HIV transmission, with no major increase in overall use. The 2012 austerity dissolution of the IDT into SICAD, and the subsequent chronic underfunding of harm-reduction services, produced a measurable post-2018 increase in overdose deaths — a 45 per cent jump in 2021 specifically.

What transferred: the architecture (decriminalisation + CDT + funded treatment + harm reduction). What didn't: the sustained political consensus that paid for the architecture for a decade. The lesson: decriminalisation without funded treatment is just legalisation by the back door, and the consequences fall on the most vulnerable users.

France — primary-care buprenorphine, 1995

The single largest reduction in opioid overdose deaths ever recorded in a developed-country national population. In 1995, France made buprenorphine available on prescription by any GP, with no specialist gatekeeping requirement. Within a decade, opioid overdose deaths fell by approximately 79 per cent ([Auriacombe et al., American Journal on Addictions, 2004](https://pubmed.ncbi.nlm.nih.gov/15204673/); subsequently confirmed in cohort follow-ups).

The mechanism is the structural-availability move applied to pharmacotherapy. The drug existed elsewhere; France was the first major country to push it through primary care rather than restricting it to specialist addiction services. Within a year, more French opioid users were on buprenorphine than methadone — because the GP was a primary-care visit away and the addiction service was a specialist-referral and waiting-list problem.

What transferred: the principle that pharmacotherapy access is rate-limited by where prescribing is allowed, not by the existence of the drug. Most countries have not followed. US prescribing was restricted by the DATA-Waiver requirement for buprenorphine until 2023 — and overdose deaths rose for two decades under that restriction. The lesson: structural access to the molecule, at the primary-care level, is the largest single policy lever in opioid mortality.

Switzerland — heroin-assisted treatment

The Swiss approach started in 1994 with the Federal Office of Public Health authorising a multi-site pilot of supervised diacetylmorphine (pharmaceutical heroin) prescription to methadone-refractory long-term users. The pilot expanded into a national programme after Swiss voters approved it in 1997 and again, more decisively, in 2008.

The mechanism: for the subset of opioid users for whom methadone does not retain them in treatment, supervised heroin at fixed doses, in a clinic setting, with social-work support, produces measurably better outcomes — lower illicit drug use, lower illegal income, higher employment, lower mortality. Six RCTs subsequently replicated the Swiss finding in the Netherlands, Germany, Spain, the UK (RIOTT), and Canada (NAOMI, SALOME). The EMCDDA Insights monograph is the canonical review.

Swiss outcomes by 2010: opioid overdose deaths down 50 per cent from peak, HIV transmission among IDUs down 60 per cent, drug-related crime down 70 per cent in the populations enrolled in HAT or methadone. Heroin-related crime in Zurich's notorious Platzspitz "needle park" essentially ended within five years of the policy shift.

What transferred: the principle that prescription heroin for the methadone-refractory subgroup is operationally feasible, cost-effective relative to incarceration, and produces measurable mortality and crime benefits. What didn't: the political consensus to do it at all. Most countries find the moralising blocker insurmountable on prescription heroin specifically. Canada has the largest non-European programme; the US has none.

Australia — plain packaging

Australia 2012 was the first country in the world to mandate plain packaging on tobacco products. The cigarette pack lost all branding except the product name in a standard font; the rest of the surface became a warning image and standardised earth-tone packaging colour (selected as the colour least appealing to focus-group consumers).

The Chipty review for the Australian Treasury, 2016 found measurable additional reductions in adult smoking prevalence attributable to plain packaging, on top of the contributions of price increases and other interventions. The effect is debated in detail (the tobacco industry funded multiple counter-studies) but the policy is now standard in the UK (2016), France, Ireland, Norway, New Zealand, and others. WHO recommends it.

The mechanism is structural at the level of cue exposure. The pack itself is the largest marketing surface tobacco has, particularly post-display-ban when the pack is the only visible product feature. Removing the branding removes a load-bearing piece of the addictive cue architecture (see Pathway § Wanting — cue-induced craving via incentive-sensitization). The intervention does not change what the consumer knows about tobacco; it changes the visual environment in which the purchase decision is made.

What transferred: the principle that packaging is part of the addiction architecture, not just marketing. Now standard policy across high-income tobacco-control countries. The lesson: structural interventions on cue exposure work. They generalise to other industries (UK HFSS supermarket placement rules use the same mechanism).

UK — smoking ban, sugar tax, minimum unit pricing in Scotland

The UK is the rare developed-country example of three structural-intervention success stories running in parallel within the last twenty years. The Structural page covers the smoking story. Quickly here:

• Smoking ban 2007. Workplaces, pubs, restaurants. Combined with display ban (2012/15) and the duty escalator, took adult smoking from 22 per cent to under 13 per cent in fifteen years.
• Soft Drinks Industry Levy 2018. Producer-side reformulation tax. The industry-response claim is solid (~30 per cent sugar reduction per product via reformulation). The downstream consumer-health claim is contested — Snowdon at the IEA points out the figure is per-product not per-consumer, and the much-cited JAMA Pediatrics 2024 girls-only obesity effect is a classic multiple-comparison flag. The structural mechanism (industry reformulated under regulatory pressure) is defensible; the population-health effect is not yet earned. Full treatment on the Structural page.
• Scottish MUP 2018. 50p per unit floor. Reduced sales in heaviest-drinking deciles. The headline "reduced alcohol-attributable deaths" claim is heavily confounded by COVID — MUP started May 2018, COVID arrived March 2020, and Scottish alcohol-specific deaths actually rose in absolute terms across the post-MUP period (peaking at 1,245 in 2021, highest since 2008, per NRS). The Public Health Scotland evaluation reports a 13.4 per cent reduction against a synthetic-control counterfactual, not an observed reduction. The price-floor mechanic worked at the level it was designed to; the population-mortality effect is plausible but smaller than the headline and inferentially fragile. Full treatment on the Structural page.

What unites the three: all are upstream regulatory interventions on the industry or the environment, not downstream educational interventions on the consumer. All produced measurable population-level reductions in the dependent variable. None depended on cultural change or moral persuasion to operate. The UK is, in this narrow sense, a structural-intervention success story across three substances.

Iceland — Planet Youth (cross-reference)

Full treatment on the Structural page. The single largest sustained reduction in adolescent substance use ever recorded in a national population. Past-30-day adolescent drunkenness from 42 per cent (1998) to 5 per cent (2016). No classroom drug-education content — entirely structural and parental-engagement based. International replication is mixed but the Iceland-specific evidence is uncontested.

Medellín — social urbanism (cross-reference)

Full treatment on the Structural page. Library-parks, aerial trams, escalators, child-care centres — physical infrastructure investment in the poorest comunas changed the recruitment economics for armed groups. Homicide rate down ~90 per cent in two decades. Multiple causes, not all attributable to the social-urbanism interventions, but the structural pattern is the most defensible component.

Estonia — OST scale-up after the USSR

Less well-known than Portugal but instructive. Estonia inherited from the Soviet collapse a per-capita IDU population among the highest in Europe and an HIV epidemic among IDUs that peaked in the early 2000s at >50 per cent prevalence in some cities. Beginning around 2005, Estonia scaled up methadone and buprenorphine OST through a combination of public-health funding and EU technical support. By 2015 OST coverage of the diagnosed OUD population was approaching Western European levels and HIV incidence among IDUs had fallen substantially.

The pattern is instructive because Estonia did the structural-pharmacology move (OST scale-up) without the political-cultural move (Estonia did not decriminalise, did not run a Portuguese-style CDT, did not change the legal status of personal-use possession). The drug worked anyway. The lesson: opioid agonist therapy is the rate-limiting intervention; decriminalisation is helpful but not necessary if the OST is delivered. ([EMCDDA Estonia country drug report, 2024](https://www.euda.europa.eu/publications/country-reports/estonia_en).)

Ukraine — OST through the war

The hardest-stress-test case in modern addiction policy. Ukraine has run an OST programme since 2004, expanded substantially after 2014, and has — remarkably — sustained methadone and buprenorphine provision through the full-scale Russian invasion from February 2022 onwards. Reports from the [International Network of People Who Use Drugs](https://inpud.net/ukraine-ost-during-war/) and [WHO Ukraine](https://www.who.int/europe/news/item/15-04-2022-providing-life-saving-opioid-agonist-therapy-during-the-war-in-ukraine) document the operational continuity: take-home doses where clinic access was impossible, mobile pharmacies, cross-border resupply, displaced-patient registration in Western countries.

The achievement is small in absolute terms compared to the wartime mortality elsewhere — perhaps tens of thousands of patients kept in treatment under impossible conditions. The lesson is larger. If OST can be sustained through full-scale war, the operational excuses for not running it in peacetime are mostly excuses. The Ukrainian case is the strongest single rebuttal to the "we cannot afford / cannot organise / cannot maintain" arguments deployed against OST scale-up in lower-resource settings.

Netherlands — coffee shops as separation

The Dutch model, in place since the 1976 Opium Act revision, is often misread as "cannabis legalisation". It isn't. Cannabis is illegal at the national level. What the Opium Act distinguishes is "hard drugs" (heroin, cocaine, meth, MDMA) from "soft drugs" (cannabis, hashish), and the coffee-shop tolerance regime is a prosecutorial discretion arrangement that permits small-quantity cannabis sale and possession at licensed venues without prosecution.

The mechanism is market separation. The original policy theory: by routing cannabis through licensed coffee shops, separate the cannabis user from the hard-drug market they would otherwise have to enter via the same dealer network. Reduce the gateway pressure. The outcome: Dutch cannabis use rates are roughly the European average (lower than UK, US, France in most periods); heroin use among young Dutch adults fell to among the lowest in Europe in the 1990s and has stayed there.

What did not work: the supply side. Coffee shops can sell cannabis legally; they cannot grow or import it legally. The "back-door problem" has produced a parallel illegal supply chain that the Dutch government has spent twenty years trying and failing to regularise. Multiple legal-supply pilots (Maastricht, Tilburg, Breda, ten cities now) are in various stages of evaluation. The policy is incoherent at the supply level and operationally settled at the consumer level — a classic example of an intervention that solved the problem it set out to solve and left an unsolved adjacent problem in place.

What transferred: the market-separation principle. What didn't: the cultural tolerance for the legal incoherence. Most countries cannot run a policy this contradictory at the level of statute, even where the operational outcomes would justify it.

New Zealand — smokefree generation and its reversal

The cautionary tale of recent tobacco policy. New Zealand passed legislation in December 2022 introducing a "smokefree generation" — anyone born after January 2009 would be permanently prohibited from purchasing tobacco. The first country to enact such a policy. WHO praised it. UK was preparing similar legislation modelled on it.

The new National-led coalition government repealed the legislation in February 2024 before it had taken effect. Reasoning publicly: lost tax revenue, perceived inequity, libertarian objection. Reasoning operationally: the coalition's revenue forecasts required the tobacco duty stream.

The case is on this page for two reasons. One: it is the cleanest recent example of a structural intervention designed to phase out an entire substance for a population cohort. The mechanism is age-banded prohibition, which is a structural lever the willpower frame does not have an answer for. Two: it is the cleanest recent example of how structural interventions get reversed once the addiction industrial complex — in this case, the tobacco-tax-dependent fiscal architecture — defends its revenue. The Industry page discusses this pattern further.

Singapore — the harsh-penalty counter-case

The case the harm-reduction literature does not like to engage with. Singapore runs the most punitive drug-control regime in the developed world: mandatory death penalty for trafficking quantities above statutory thresholds, mandatory imprisonment for use and possession, civil-rehabilitation orders that can include compulsory residential detention. The regime has been in place essentially unchanged since the 1973 Misuse of Drugs Act.

The outcomes, on Singaporean government data and corroborated by independent assessment: low per-capita illicit drug use compared to other Asian high-income economies, low overdose mortality, no fentanyl crisis, no large IDU HIV cohort. The Singapore Anti-Narcotics Association reports that around 0.5 per cent of the population has any history of drug use — an order of magnitude below the developed-country norm.

The case for taking Singapore seriously: the policy has produced the outcomes it set out to produce. The case against treating it as a model: (a) ethically, the death penalty for trafficking is widely regarded as a violation of human-rights norms; (b) operationally, Singapore is a 5-million-person city-state with controlled borders, a high-trust authoritarian administration, and a starting cultural position that supports the regime — none of which travels to most other contexts; (c) the regime suppresses reported drug use but the underlying prevalence is harder to measure precisely because of the penalty regime; (d) the regime addresses use, not the broader addiction picture — alcohol and obesity outcomes in Singapore are unremarkable.

The honest read: harsh penalties can produce low reported drug use under specific institutional conditions and at moral costs that most societies decline to pay. Singapore is a working counter-example to the claim that prohibition cannot work, and a working illustration of what prohibition costs when it does.

Cross-country patterns

Five patterns emerge across the case studies.

One: structural always beats educational. Every country that produced large reductions in addictive behaviour did so through structural means — price, availability, environmental design, primary-care access to pharmacotherapy. The countries that ran large educational campaigns without structural intervention did not produce the same outcomes (DARE, the US "Just Say No" era, change-for-life UK). This is the Structural page thesis confirmed at the country-comparison level.

Two: pharmacotherapy access at the primary-care level is the largest single opioid-mortality lever. France 1995 demonstrated this in the cleanest available form. Estonia and Ukraine confirmed it under stress. The US under DATA-Waiver demonstrated the inverse. The lesson generalises beyond opioids: any pharmacotherapy whose access is gated to specialist services reaches a small fraction of the population that would benefit. Push it through primary care and coverage expands by an order of magnitude.

Three: decriminalisation without funded treatment is brittle. Portugal demonstrated this in both directions — funded for a decade, results held; cut from 2012, results degraded. The lesson is not that decriminalisation does not work but that the political coalition required to fund the treatment side is the harder maintenance problem.

Four: structural reversals happen. New Zealand's smokefree generation, repealed before implementation. The Portuguese SICAD austerity cuts. The Swiss HAT programme had to be reconfirmed twice at national referendum. The Industry page discusses why this happens: structural interventions create losers (revenue, jobs, ideological positions), and the losers organise.

Five: the counter-cases are real and ethically expensive. Singapore works on its own terms and costs what those terms cost. The intellectual honesty of the harm-reduction literature requires engaging with this rather than pretending the prohibition position has no empirical record. The fair-investigation frame of this microsite requires the same.

The next page is the political-economy argument that ties the failures of structural intervention back to a specific set of actors: the industry that profits from not solving the problem.