Drug Policy Criminalization versus Treatment

What the evidence says

The clearest and most heavily replicated line of evidence in this set concerns arrests, not overdose or health outcomes: across a systematic review of drug policy reforms, decriminalization and harm-reduction policies consistently decrease drug-related arrests, particularly among people who use drugs, and this decrease occurs without a corresponding increase in violent or property crime in most of the studied contexts, including Oregon, Washington, and California's needle-exchange legalization. Cannabis decriminalization and legalization specifically decrease arrests for cannabis possession, and cannabis legalization in Canada shows no measured increase in violent or property crime. At the same time, racial disparities in who gets arrested persist after these reforms, and police discretionary practice can maintain de facto criminalization even where the law has formally changed — meaning the arrest reduction is not automatic and depends on enforcement behavior, not statute alone.

A second line concerns criminalization's effect on health-seeking behavior, especially around pregnancy. Drug testing of pregnant patients by health care providers decreases engagement in prenatal care, and punitive prenatal substance-use policies decrease prenatal care utilization both among people who use substances and at the population level. This effect has a measured gradient: women in counties with medium levels of pregnancy criminalization are 6% more likely to initiate prenatal care late or not at all, and women in the highest-criminalization counties are 11% more likely, compared with women in counties with no arrests. A separate source on reproductive healthcare broadly reports the same direction — women avoiding prenatal and reproductive services out of fear of legal consequences, judgment, or custody loss — though without a comparable quantified rate.

A third line concerns what happens to people once they are inside the criminal legal system. Most people who use heroin in the U.S. (70%) encounter that system, and opioid use disorder prevalence in surveyed New Jersey jails averages 23% (ranging 8–49% across facilities). But contact does not translate into treatment: fewer than 5% of people referred to community treatment by the criminal legal system receive medications for opioid use disorder, and overdose risk in the weeks after release from incarceration runs up to 100 times higher than in the general population. Separately, a policy change that removed a criminal-legal lever — New Jersey's elimination of cash bail — decreased average jail length of stay from 7.4 days in 2014 to 3.7 days in 2017, roughly a 50% reduction, though this measures time in custody, not a drug-specific health outcome.

Restricting harm-reduction access shows the same directional pattern as criminalization more broadly: laws limiting access to sterile injection equipment increase HIV-related risk behaviors among people who inject drugs, and criminalization combined with inadequate treatment services and lack of safe drug supply increases serious harms — withdrawal, drug poisoning, and violence exposure — among street-involved women. The inverse also appears: trust-based, non-punitive outreach increases care access and increases these women's capacity to navigate health and substance-use services independently.

On drug use itself, the evidence is thinner and more narrowly scoped. A meta-regression built on decriminalization events in 12 countries and 4 U.S. states projects that decriminalization produces a short-term increase in cannabis use — about 1 percentage point in past-12-month use and about 0.5 percentage points in past-30-day use — though the source flags that part of this increase may reflect people becoming more willing to report use once it is no longer labeled a crime, rather than a genuine behavior change. No claim in this set quantifies a population-level change in overdose rates as a direct result of decriminalization; the only overdose figure here is the post-incarceration risk spike, which is about criminal-legal exposure, not decriminalization. Public belief runs somewhat against the arrest evidence: in a British Columbia survey, only 33% of adults supported decriminalization, and 55% disagreed that it would reduce overdose rates — a belief the arrest-and-crime literature above does not directly test either way, since none of the reviewed studies measure overdose outcomes from decriminalization.

Where the evidence splits

Decriminalization's effect on arrests is not uniform across jurisdictions. In Oregon and Washington, and across a 36-study systematic review, decriminalization decreases drug-related arrests without a consistent increase in violent crime. But in Tijuana, Mexico, the 2009 decriminalization reform had no significant impact on arrests for drug possession, and police enforcement against drug users increased instead. The sources describe drug-policy-reform effects overall as "mixed and context-dependent," and separately note that police can maintain discretionary enforcement practices despite formal decriminalization — but they do not tie that discretion mechanism specifically to the Tijuana case, so the reason for that jurisdiction's divergence from Oregon and Washington is not explained in the sources on this page.

Thresholds and dose-response

Prenatal care delay increases with the level of county-level pregnancy criminalization: women in counties with medium criminalization rates are 6% more likely to initiate prenatal care late or not at all than women in counties with no arrests (adjusted RR 1.06), and women in the highest-criminalization counties are 11% more likely (adjusted RR 1.11).

Open empirical questions

• No claim in this set quantifies the effect of decriminalization on population-level overdose rates; the only overdose figure here concerns risk after release from incarceration, a different exposure entirely.
• Why enforcement outcomes diverged between Oregon/Washington and Tijuana is not explained beyond noting the jurisdictions differ; whether police discretion, resourcing, or another factor caused the Tijuana result is not measured.
• Whether the cannabis-use increase from decriminalization reflects genuine behavior change or increased willingness to self-report is not resolved by the available data.
• Whether decriminalization's contribution to Portugal's reported outcomes is separable from the concurrent treatment-service and reintegration expansion is not measured.