‘Safe Drug Sites’ Haven’t Failed. Federal Policy Has

A recent Washington Post column by Charles Fain Lehman and Kevin Sabet claims that current data prove safe drug sites do not save lives. But that is not what the data say.  They are right to warn against over-promising; supervised consumption sites are not a complete answer to overdose deaths. Harm reduction should not become an article of faith immune from evidence. Reversing an overdose in an injection site does not necessarily change population-level overdose mortality. A 2025 systematic review found that large provincial studies generally did not show significant reductions in overdose mortality, while some smaller urban studies suggested possible protective effects. These findings should be treated as a subject for rigorous evaluation, not as a battleground for slogans.

But the Washington Post op-ed goes beyond caution. It treats uncertainty as proof of failure. That is the same narrative shortcut that has distorted opioid policy for decades: take a partial truth, flatten the complexity, and turn it into a moral conclusion. The opioid crisis has repeatedly been misread through simple stories—that prescribing caused it, enforcement will fix it, treatment access alone will solve it, harm reduction enables it. Each story contains fragments of truth, but none is adequate.

The key question is not simply whether these sites, in their current limited form, have reduced citywide or statewide overdose deaths. The better question is: What would we expect a tiny number of legally embattled, politically constrained sites to accomplish in a vast and lethal fentanyl market? If a city has only one or two sites, if they are not located where overdose risk is greatest, if their hours are limited, if people fear police surveillance or public stigma, and if federal law threatens their existence, it is hardly surprising that population-level effects are difficult to detect.

That does not prove the model fails; it may indicate that we have never allowed the model to be implemented at the scale, location, and integration necessary to test it fairly.

The federal law at issue should not be treated as a neutral fact of nature. Yes, 21 U.S.C. § 856 prohibits maintaining “drug-involved premises,” and in United States v. Safehouse, the Third Circuit Court of Appeals held that such sites violate that statute. But legality and wisdom are distinct. A law written in the era of “crack-house” panic should not be presumed fit for a fentanyl-era public health emergency. The legal debate often centers on whether supervised consumption sites violate federal law; the more urgent question is whether federal law, as currently written, undermines public health.

If the law prevents communities from piloting, studying, and improving these sites, then the law itself becomes part of the failed overdose response. We cannot demand definitive evidence while criminalizing the conditions needed to generate it. That is a policy trap, not a scientific standard.

The Lehman and Sabet opinion also leans too heavily on the idea that better answers already exist, especially methadone and buprenorphine. While it is true that these medications save lives,  availability is not access, and access is not acceptance.

CDC data show the gap: in 2022, only 25.1% of U.S. adults who needed opioid use disorder treatment received medications. Furthermore, SAMHSA’s pooled 2022–2024 data similarly show that only a minority of adults with past-year opioid use disorder received these medications. We cannot assume that people at the highest risk will seek, accept, remain on, or even see themselves as candidates for opioid use disorder medications.

This is the “if we build it, they will come” fallacy. Many people with severe opioid use disorder do not trust treatment systems, fear stigma or withdrawal, lack transportation or insurance, have had prior negative experiences, or are simply not ready to stop using. A public health strategy that depends solely on people presenting for treatment will inevitably miss those at greatest risk.

That is precisely where harm reduction matters. It creates contact before readiness. It says: We will try to keep you alive until you are ready for treatment. That is not surrender; it is clinical realism.

Early New York City data should be interpreted carefully but not dismissed. During the first two months of operation, the city’s overdose prevention centers recorded thousands of supervised uses, more than 100 staff interventions, and zero onsite deaths. While this does not prove a citywide mortality effect, it shows that the centers reached people at high risk. Furthermore, claims that such sites inevitably damage neighborhoods deserve scrutiny. A JAMA Network Open study found no significant increases in crimes or emergency calls in neighborhoods where New York’s two centers were located. 

The right conclusion is not that supervised consumption sites are the singular answer. Real prevention requires changing the conditions that make harmful substance use more likely: trauma, poverty, housing insecurity, untreated mental illness, community disinvestment, criminalization, and lack of opportunity. Short-term survival strategies—naloxone, syringe services, outreach, drug checking, methadone, buprenorphine, and supervised consumption—should be linked to housing, employment, treatment, peer support, and long-term recovery pathways.

We should evaluate supervised consumption sites honestly. We should not inflate overdose reversals into mortality statistics, nor should we pretend that the federal law blocking them is working. 

The overdose crisis will not be solved by one intervention, nor by prohibiting imperfect ones because they fail to do everything. Harm reduction keeps people alive today; treatment helps those ready to accept it. Preventing upstream causes of substance use reduces the number who become vulnerable in the first place. A serious response needs all three. Anything less is another incomplete story.