Overdose Deaths Just Had Their Best Year in a Decade. Washington Just Cut the Program That's Doing It.

Elyse Powell has run the North Carolina Harm Reduction Coalition long enough to know what a good year sounds like on the phone: fewer calls where someone is describing a friend on the ground, breath short, asking whether the second dose of naloxone should have worked by now. Her organization reaches more than 5,000 people a year, and by the numbers, 2026 should be one of the best years harm reduction workers like her have had since the crisis began. It is. It’s also the year her own supply chain — funding, test strips, staff — got harder to hold together than at almost any point before it.

Both things are true at once, and the second one is why the first one might not last.

U.S. overdose deaths fell 27% in 2024, according to a Commonwealth Fund report authored by senior research associate Evan Gumas, and preliminary 2025 data point to a further decline. Brown University epidemiologist Brandon Marshall, who has spent two decades tracking this epidemic, put it plainly: “It is unprecedented to see overdose deaths come down so dramatically, so quickly.” The CDC’s own provisional data confirm the shape of the drop, off a base of more than 105,000 annual deaths as recently as 2021 through 2023.

The country just proved harm reduction works at population scale, and is defunding it anyway.

That’s the sentence this story is built around, and it isn’t hyperbole — it’s the plain read of two facts sitting next to each other in the same year. Here is the case for it, and here is what you can still reach for tonight if you’re the one this article is actually about.

The mechanism isn’t a mystery — it has a name, a sample size, and a p-value

Marshall was careful, in the same STAT interview, to note that “the dominant hypothesis is that it’s mostly a function of the drug supply” and that “we need a lot more research rapidly.” He’s right that no single cause explains a national decline this size. But one piece of the mechanism has now been measured directly, not just correlated with a shrinking supply, and it points straight at the interventions currently losing money.

A team led by Marina Plesons and published in the Harm Reduction Journal followed 136 people who use drugs through Miami’s IDEA syringe services program from 2017 to 2025, using a statistical method called a cross-lagged panel model — in plain terms, a way of asking “does A predict less B later, or does B just happen to sit next to A.” The answer here wasn’t ambiguous. Getting naloxone at an SSP visit predicted a measurably lower chance of overdosing before the next visit (a drop of 0.18 to 0.28 on their scale, with a p-value between 0.002 and 0.004 — small enough that this isn’t noise). Across the three years of data, the share of participants reporting a recent overdose fell from 43% to 31% to 28%, while the share who had naloxone on hand when they needed it climbed from 53% to 69% to 79%.

Read that as a machine: the more consistently a syringe program hands someone naloxone, the less likely that same person is to overdose again. It’s not a metaphor for harm reduction working. It’s harm reduction working, with a sample size and a confidence interval attached. The authors are honest about the limits — 136 people at one Miami program is not the nation, the data can’t capture the people who died and never made it to a follow-up visit, and a global pandemic and a shifting drug supply sit uncontrolled in the background. None of that changes what the number says about the mechanism when it’s present.

Mass General Brigham addiction physician Sarah Wakeman framed the stakes the way she sees them in the exam room: “Every single overdose death is really a failure, because no one should die from an opioid-related overdose when we have an antidote.” The antidote she’s talking about is naloxone — the thing that requires it to physically be in someone’s pocket, which requires a program to have handed it to them, which requires that program to be funded.

The people who actually carry that antidote and use it on each other put it more bluntly than any clinician. In interviews published this spring in the journal Addiction, researchers led by Laura Sisson asked 22 people who use drugs in Maryland — all of whom had personally reversed someone else’s overdose with naloxone in the past three months — how they thought about the tradeoffs. A 58-year-old man put the calculation in six words: “I’m concerned about life. Even if it was me I wouldn’t care about the withdrawals.” That’s the whole argument for harm reduction, stated by the person actually holding the syringe, more efficiently than any policy brief has managed.

A 58-year-old man put the calculation in six words: “I’m concerned about life.

The same agency proving this out is losing half its staff

Here’s where the year stops being simple good news. Since the current administration took office, SAMHSA’s workforce has been cut roughly in half — a third of its roughly 900 employees let go by October 2025, another 100 cut during the government shutdown that followed. In January 2026, the administration moved to cancel nearly $2 billion in SAMHSA discretionary grants overnight, then reversed the decision a day later under public pressure. The reversal made headlines. The part that didn’t reverse got less attention: at least $245 million in addiction- and overdose-prevention funding has stayed cut regardless.

Then, in April 2026, new guidance ended federal funding support for fentanyl test strips outright — reversing prior guidance that had explicitly allowed programs to buy them. If you’ve never used one, a fentanyl test strip is the harm-reduction equivalent of checking an expiration date: a five-dollar way to find out, before you use, whether what you’re holding contains something that will kill you. Taking federal money out from under that isn’t a budget line item. It’s removing the exact tool that turned into a measured, published reduction in overdoses in the Miami study above.

It also lands at the worst possible moment on the ground, because the opioid supply itself just got harder to test for. Medetomidine — a veterinary sedative 100 to 200 times more potent than xylazine — has taken over Philadelphia’s fentanyl supply with startling speed: from 29% of dope samples in May 2024 to 90% by March 2026, while xylazine fell from 97% to 28% over the same stretch. It’s since turned up in North Carolina, Ohio, and Illinois. Naloxone doesn’t touch it — we cover what that means for the people testing their own supply separately today. Nabarun Dasgupta, who directs the UNC Street Drug Lab and testified before Congress about it this spring, described what emergency doctors are now seeing: “Medetomidine withdrawal is unlike anything most emergency physicians have seen before… abrupt stopping can trigger a crisis that hits fast and hard,” severe enough that most hospitalized cases need an ICU bed. Test strips for it exist — at roughly $200 a box of 100, compared with $35 for fentanyl strips — right as federal support for buying any test strips at all is disappearing.

Powell, who has watched three or four of these adulterants cycle through in as many years, doesn’t sound like someone who thinks scheduling the next chemical will fix this: “It’s getting really mucky and messy, and things are turning over really quickly,” she said. “By the time everyone gets their arms around something, it will be some other thing.” Her sharper point was about the policy reflex itself: “There’s just this sort of knee jerk response to criminalize it… every time they schedule one of these new things, it’s just replaced with something else.”

We have run this experiment before, and we already know how it ends

If that pattern sounds familiar, it should. In 1988, Congress — led by Senator Jesse Helms — banned federal funding for needle and syringe exchange programs in the middle of the HIV/AIDS epidemic, on the theory that funding clean syringes would look like endorsing drug use. The ban held for over two decades. President Obama lifted it in 2009; Congress reinstated it in the FY2012 omnibus spending bill after Republicans retook the House, on the same argument, one generation later. Each time, the public-health case for the intervention wasn’t the open question — HIV transmission rates among people who inject drugs are not seriously disputed by anyone in the field. The open question was always whether Washington could tolerate looking like it was making drug use safer in order to make it survivable. It took the syringe-exchange fight more than twenty years, and an uncounted number of preventable HIV infections, to land on “yes, mostly.” We are now re-litigating that same argument with a different acronym — SSP instead of NSEP, test strips instead of syringes, medetomidine instead of xylazine — and betting the multi-year decline in overdose deaths on the outcome.

The open question was always whether Washington could tolerate looking like it was making drug use safer in order to make it survivable.

What providers, case managers, and the rest of us do this week

If you work in this field, the actionable version of this story isn’t “advocate for more funding,” although that’s not wrong. It’s narrower: audit whether your program’s naloxone and test-strip supply is funded through a line item that’s currently vulnerable, and if it is, diversify now — state opioid settlement dollars, philanthropic harm-reduction grants (the Comer Family Foundation and Elton John Foundation US branch are both currently accepting applications), and local health-department budgets are proving more durable than federal discretionary grants this year. Don’t wait for the next SAMHSA guidance memo to find out your test strips aren’t reimbursable anymore.

And if you’re the person this whole apparatus exists for — if you’re using, or you’re not sure you’d call it that but you know what withdrawal feels like at 4 a.m. — here is the plain version, no lecture: naloxone still works on everything except medetomidine, it is still free at most syringe programs and many pharmacies, and it is still federally legal for you to carry it and use it on someone else without repercussion. That has not changed this year, whatever else has. Test before you use if you can get a strip. Never use alone if you can help it. None of that is a moral position. It’s just what the data above says keeps people alive long enough to have a next year at all.

Powell will keep taking those calls either way — the ones about a friend on the ground, the ones asking if the second dose should have worked by now. The number of them is falling for the first time in years. Whether it keeps falling isn’t really a medical question anymore. It’s a budget one, and the deadline for answering it is this Congress, not some future one that gets to inherit a smaller crisis because this one did the work.