69,973 Deaths. The Number That Proves Harm Reduction Works—and the Policy Quietly Undoing It.

---

title: “69,973 Deaths. The Number That Proves Harm Reduction Works—and the Policy Quietly Undoing It.” slug: samhsa-test-strip-ban-overdose-decline-reversal-2026 subtitle: “A 13.9% drop in overdose deaths is the best news in years. Washington’s response: defund the tools that helped get us here.” post_type: article primary_category: harm-reduction lens: [harm-reduction, policy] tags: [fentanyl, xylazine, medetomidine, samhsa, fentanyl-test-strips, naloxone, harm-reduction, overdose-prevention, federal-policy] meta_description: “The CDC’s May 2026 provisional data shows 69,973 overdose deaths in 2025—a 13.9% decline. The same spring, SAMHSA pulled federal funding for fentanyl test strips. Here’s what’s at stake.” read_time_minutes: 9 schema_type: NewsArticle sources:

• url: https://www.cdc.gov/nchs/products/databriefs/db549.htm title: “Drug Overdose Deaths in the United States, 2023–2024” publisher: CDC NCHS source_tier: A published_date: 2026-05-13
• url: https://www.statnews.com/2026/04/27/trump-administration-samhsa-clear-shift-from-harm-reduction/ title: “Trump administration signals clear shift from harm reduction” publisher: STAT News source_tier: B published_date: 2026-04-27
• url: https://news.ashp.org/news/ashp-news/2026/04/30/ashp-opposes-new-restrictions-on-fentanyl-test-strips-for-harm-reduction title: “ASHP Opposes New Restrictions on Fentanyl Test Strips for Harm Reduction” publisher: ASHP source_tier: B published_date: 2026-04-30
• url: https://drugpolicy.org/resource/federal-cuts-threaten-overdose-prevention/ title: “Federal Cuts Threaten Overdose Prevention” publisher: Drug Policy Alliance source_tier: B published_date: 2026-06-01
• url: https://www.fiercehealthcare.com/regulatory/hhs-cuts-2b-addiction-and-mental-health-grants-reverses-course-restore-funding title: “In 24 hours, HHS cuts, then restores $2B in addiction and mental health grants” publisher: Fierce Healthcare source_tier: B published_date: 2026-01-14 ai_generated: true

---

The provisional numbers came out on May 13, quietly, in the form of a CDC data brief that most national outlets covered with a paragraph and moved on. Sixty-nine thousand, nine hundred and seventy-three drug overdose deaths, projected for the twelve months ending in December 2025. That represents a 13.9 percent decline from the prior year—the largest single-year drop in more than two decades of counting overdoses. In a public health emergency that has killed over a million Americans since 1999, a number that trends downward feels, briefly, like permission to breathe.

Then you read what happened in April.

On April 24, 2026, Chris Carroll—acting director of the Substance Abuse and Mental Health Services Administration for fifteen consecutive months without a confirmed replacement named by the administration—signed updated guidance formally ending federal funding for the public distribution of fentanyl test strips, xylazine test strips, and medetomidine test strips. Not for clinical or institutional use: law enforcement screening evidence, clinical labs testing patients, public health officials conducting supply surveillance still had access. For everyone else—the harm reduction worker handing a strip to someone at a syringe exchange, the community health outreach coordinator slipping a packet to a client before they go home to use alone—the federal funding was cut off.

This is not a peripheral policy change. It is a decision about the relationship between the federal government and the reality of drug use in America. And it is happening at the worst possible moment, for reasons that deserve scrutiny, while a number that should give us hope is being used as cover.

The decline didn’t come from nowhere

The 2025 overdose data is the product of more than a decade of contested, incremental, legally fraught expansion of harm reduction infrastructure. The mechanisms behind the decline are not mysterious—they are documented in peer-reviewed literature and contested only by those with political reasons to contest them. Naloxone expanded: by 2024, 47 states had standing orders allowing pharmacies to dispense it without a prescription, a change that happened at the state level because federal pharmacies wouldn’t move fast enough. Medication for opioid use disorder—buprenorphine and methadone—reached patients through telehealth access that the public health emergency unlocked during COVID, extending coverage to rural communities that previously required hours-long drives to the nearest opioid treatment program. Fentanyl test strips, classified as drug paraphernalia under most state laws until 2021, began moving through syringe exchange programs and community distribution at meaningful scale.

The evidence that these interventions saved lives is not a matter of reasonable debate in public health. Multiple studies have found that naloxone distribution is associated with reduced overdose mortality. MOUD reduces all-cause mortality in people with opioid use disorder, including from overdose, by somewhere between 50 and 80 percent depending on the formulation and population. Fentanyl test strip distribution is associated with behavior change: people who receive them report checking their supply, modifying their dose, using with others present, and, in a meaningful percentage of cases, not using at all. The strips don’t create drug use. They give people information that changes how they use it.

The 13.9 percent decline is real, and the people who built and sustained the harm reduction infrastructure that produced it deserve to say so. What is also true is that this decline is unevenly distributed in ways that reveal the persistent failures of that same infrastructure. Researchers studying what they call the “fourth wave” of the overdose crisis have documented a disturbing pattern: deaths are declining faster in white, urban communities; they continue to rise among Black, Native American, and Alaska Native populations, and in rural areas where harm reduction programs never reached full saturation. The national aggregate is genuine. It is also a ceiling that obscures the floor for the most underserved populations—and those populations are the ones most directly dependent on the federal funding that April 24 just cut.

What is also true is that this decline is unevenly distributed in ways that reveal the persistent failures of that same infrastructure.

What the guidance actually says, and what it doesn’t

Carroll’s updated guidance—framed as a clarification of allowable uses under SAMHSA-funded programs—prohibits federal block grant and discretionary funding from being used to purchase fentanyl, xylazine, or medetomidine test strips “for distribution to individuals for personal use.” The carve-out for institutional and professional users remains: law enforcement, clinical labs, and public health officials can still use federal funds for test strips in their professional capacity. The explicit prohibition is on what harm reduction programs actually do, which is put strips in the hands of people who use drugs before they use.

The guidance came alongside a set of other restrictions that, taken together, document a coherent philosophy: no federal funds for sterile water or saline for injection hygiene; no federal funds for “drug paraphernalia viewed as facilitating use”; no funding for overdose hotlines that enable “remote communication with staff.” These are not edge cases. They are the core toolkit of programs that operate on the evidence-based principle that keeping people alive is prerequisite to getting them into treatment. You cannot offer someone recovery services if they are dead.

SAMHSA’s response from the medical community was swift. The American Society of Addiction Medicine, under president Stephen Taylor, committed publicly to “engaging with federal partners to ensure policies reflect evidence-based practices”—language carefully calibrated to preserve access and influence. The American Society of Health-System Pharmacists was less diplomatic. In a statement published April 30, ASHP said it “strongly opposes” the restrictions, citing the established evidence base and calling on SAMHSA to reverse course. Pharmacists understand test strips in a way policy staff apparently don’t: they are a drug-checking technology, not a drug-enabling technology. The difference matters clinically and it matters morally.

The drug supply got more dangerous on the same schedule

Medetomidine was not on most harm reduction workers’ radar three years ago. Today it is appearing in the fentanyl supply in an increasing number of markets—an alpha-2 adrenergic agonist, originally a veterinary sedative, that potentiates opioid effects and extends their duration. Like xylazine, it complicates standard naloxone reversal because it acts through a non-opioid pathway: naloxone will reverse the opioid component of an overdose, but the sedation from medetomidine persists. The reversal agent for medetomidine is atipamezole, which is not yet widely available in harm reduction settings and has not been approved for human use in the US. There is, at present, no consumer-facing solution to medetomidine in the supply except awareness—knowing it’s there, knowing how to respond differently, calling 911 immediately rather than waiting to see if naloxone is sufficient.

That awareness comes from test strips. A person who can check their supply for xylazine before using knows to call for help faster if naloxone doesn’t work. A harm reduction program that can distribute medetomidine test strips can tell clients what new threat has arrived in their specific ZIP code’s supply. This is exactly the moment when the federal government decided to stop funding it.

Nitazenes compound the problem further. This class of synthetic opioids—initially developed in the 1950s and abandoned for clinical use but now appearing in illicit supplies globally—are, in some variants, significantly more potent than fentanyl. DEA intelligence has tracked nitazene contamination in fentanyl-marked supplies in multiple US cities. The detection of nitazenes requires strip technology that is still being developed and distributed. Federal funding was one of the mechanisms for getting that technology to scale.

The SAMHSA guidance does not engage with any of this. It categorizes test strip distribution as a problem to be prohibited rather than a capacity to be expanded, and it does so at the precise moment when the supply-side threats that test strips detect are multiplying faster than the tools to address them.

”Life-long medication”—the tell in the text

There is a phrase buried in Carroll’s broader set of guidance letters that deserves more attention than it has received. In addressing medication-assisted treatment for opioid use disorder, the letters warned programs against framing methadone and buprenorphine as appropriate for “life-long medication use.” This is not a minor administrative note. It is a statement of clinical philosophy—and the philosophy is empirically wrong in a specific, peer-reviewed, falsifiable way.

It is a statement of clinical philosophy—and the philosophy is empirically wrong in a specific, peer-reviewed, falsifiable way.

The American Society of Addiction Medicine’s clinical practice guidelines, updated in 2024, are unambiguous: opioid use disorder should be treated as a chronic condition requiring long-term, potentially indefinite management, in the same way that hypertension or type 2 diabetes requires ongoing pharmacotherapy. The evidence for this position is not contested among addiction medicine specialists. Patients who remain on MOUD have substantially lower overdose mortality, lower criminal legal involvement, and better functional outcomes across time horizons from one year to a decade than patients who discontinue. The abstinence-only model—the idea that medication is a bridge to get off rather than a treatment in itself—has been tested. It produces relapse and death at predictable rates.

When the acting director of SAMHSA puts the phrase “life-long medication use” in a policy letter as a thing to avoid, the signal cascades through the system. Treatment providers worried about federal funding compliance may discourage long-term prescribing. Patients being told by their clinic that they should be tapering—even though they’re stable—may lose their medication and relapse. The deaths that follow are diffuse and delayed, attributable in the data to “overdose” rather than to policy. They won’t show up in any future SAMHSA guidance as a consequence of this one.

69,973 is the floor, not the ceiling

The CDC number is worth sitting with for a moment before moving past it. Sixty-nine thousand, nine hundred and seventy-three is the projected floor—the worst-case projection for 2025, based on provisional data that will be revised as delayed death certificates are processed. The actual number for 2025 may be lower. The trend is real. The infrastructure that produced it should be celebrated, sustained, and expanded.

What is not guaranteed is that it holds. States can backfill some of the federal funding for harm reduction from their own budgets—and some will. Most won’t at the scale needed, and the states where the unmet need is greatest—rural, low-tax, politically hostile to harm reduction framing—are the states least likely to move. The private philanthropic sector can absorb some of the gap. It cannot absorb all of it, and it cannot be required to serve any particular geography or population.

The lesson of the last twenty years is that overdose death is a policy outcome, not a natural disaster. It responds to decisions—about funding, about what counts as medicine, about whether the lives of people who use drugs are worth protecting before they get clean. The 13.9 percent drop is the return on a set of decisions that accumulated across two administrations and fifty state legislatures and thousands of county health departments and nonprofit harm reduction programs and emergency department nurses who started stocking naloxone next to the trauma bay. April 24 is a decision in the other direction.

The next mortality report will tell us how big the reversal is. We don’t have to wait to know which way we’re headed.