Overdose Deaths Are Plunging. Washington Just Defunded the Reason Why.

At McLean Hospital outside Boston, a research team led by Dr. Hilary Connery has spent years asking a question that almost nobody else in addiction medicine asks directly: before your last overdose, did you want to die? In their surveys of adults hospitalized after nonfatal opioid overdoses, more than half said yes, at least a little, and roughly one in five said yes, completely. Not confusion. Not an accident they wandered into. A decision they were sitting with, that a dose of fentanyl-laced heroin happened to interrupt — sometimes by killing them, sometimes not.

That single fact should complicate everything you think you know about the best public health news in a decade. Overdose deaths in the United States fell 27% in 2024 after killing more than 105,000 people a year through 2023, and early data from 2025 shows the decline continuing. Brown University epidemiologist — a researcher who studies death and disease patterns across whole populations — Brandon Marshall called it “unprecedented to see overdose deaths come down so dramatically, so quickly.”

The system that helped cause this decline is being defunded in the same year a growing body of research says a meaningful share of the deaths inside that count were never overdoses at all — they were suicides wearing an overdose’s paperwork.

Both of those sentences are true at once, and neither cancels the other out. This is a piece about what happens when a real win and a real blind spot collide in the same public health system, at the same time, and almost nobody is naming both out loud.

The Best Year Nobody Fully Understands

Sarah Wakeman, an addiction medicine physician at Mass General Brigham, put the stakes plainly: “Every single overdose death is really a failure, because no one should die from an opioid-related overdose when we have an antidote.” That antidote is naloxone, the medication that reverses opioid overdose in minutes and is now distributed by the millions of doses through community harm reduction programs — the field’s term for keeping people who use drugs alive and safer during active use, rather than waiting for abstinence first.

A new Commonwealth Fund analysis by researcher Evan Gumas is careful about why the numbers are falling, and its caution is itself the finding: there is no single cause. Gumas points to wider naloxone distribution, expanded access to buprenorphine and methadone — the medications that treat opioid use disorder and are often shortened to MOUD — better treatment access in and around prisons, a possible decline in fentanyl’s potency, and, bluntly, the fact that so many of the people most vulnerable to overdose already died during the crisis’s worst years, shrinking the population left at risk. Virginia saw a 40% drop after a statewide naloxone saturation push. Maryland doubled its naloxone distribution and saw a 25% decline. New York increased medication access in its prisons sevenfold between 2022 and 2025.

None of this is uniform, and Gumas doesn’t pretend it is. Every U.S. state’s overdose deaths fell except South Dakota, which saw three more. Scotland’s 2025 preliminary data suggests its own decline may be reversing, a reminder that whatever combination of policy and circumstance produced this drop is neither fully understood nor guaranteed to hold. And separately, a Medicaid funding fight now moving through Congress could strip treatment access from roughly 156,000 people who currently get opioid use disorder care through it — a projection, not yet a fact on the ground, but one epidemiologists are already treating as a real threat to the trend, not a hypothetical one.

The Money Pulled Out From Under It

Here is the part that should stop you cold: the same year this decline became undeniable is the year the federal government gutted the agency most responsible for the tools that produced it. Since January, the Substance Abuse and Mental Health Services Administration has lost more than half its staff, had $1.7 billion in state block grants canceled, and seen roughly $350 million in addiction and overdose-prevention funding eliminated. Block grants are lump sums of federal money states use to run their own addiction and mental health programs on the ground — the funding stream, not a slogan. STAT News reported that at SAMHSA’s Center for Mental Health Services alone, more than half its roughly 130 employees are gone, including nearly everyone who ran youth mental health programs.

A federal judge in Rhode Island has since blocked the administration’s broader plan to fold SAMHSA into a new mega-agency — worth naming, because it means the reorganization itself is not yet reality. But an injunction against a restructuring plan does not restore the staff who already left or the block grants already canceled. Those cuts are in effect right now, this week, while the decline they’re being cut into is still happening. Marshall told STAT the funding decisions “risk and compromise much of the progress we have made.” You don’t get to celebrate a hard-won decline and hollow out the agency that helped produce it in the same news cycle without one of those things being dishonest.

Those cuts are in effect right now, this week, while the decline they’re being cut into is still happening.

We Have Done This Before

This is not the first time the U.S. pulled funding from an intervention that was working. In 1988, at the height of the AIDS epidemic among people who inject drugs, Congress — led by Senator Jesse Helms — banned federal money for syringe exchange programs, even as the CDC and the Institute of Medicine amassed evidence that clean syringes prevented HIV transmission without increasing drug use. The ban held 21 years, through both parties, while tens of thousands who injected drugs died of AIDS. It was lifted in 2010, then partly reinstated within two years. The pattern was never about the evidence. It was about which population Congress decided was worth protecting. Today’s SAMHSA cuts are that same instinct in new packaging: proof of concept treated as a reason to stop, not invest.

The Deaths Inside the Deaths

Now the second thread, the one almost nobody has connected to the first. Writing in Psychology Today this month, Traci Sweet, a co-founder of Holon Health, noticed that Recovery Month and Suicide Prevention Awareness Month both fall in September for a reason nobody says out loud: they are largely tracking the same people. Injury epidemiologist Ian Rockett and psychiatrist Eric Caine argue in a July 2026 commentary in the peer-reviewed journal Injury Prevention that as many as 30% of the overdose deaths coroners rule “accidental” or “undetermined” may actually be suicides — deaths that get filed under the overdose crisis instead of the suicide crisis, undercounting both.

“Most drug overdoses are the result of motivated behaviors and are not ‘accidents’ in the strict sense,” Rockett writes. Caine puts the stakes of getting the count wrong even more directly: “Misclassification and undercounting break this chain. Without an accurate measure of burden, it is impossible to gauge how big the problem is.” Misclassification, here, means what it sounds like — a coroner marking a death “accidental” when the evidence, if anyone looked closely, could just as easily point to intent. Rockett and Caine’s proposed fix is a measure called self-injury mortality: instead of forcing an impossible guess about what was in someone’s mind in their final minutes, it counts registered suicides together with most drug self-intoxication deaths, and studies the behavior instead of arguing about the label.

The scale of the overlap Sweet cites is not a rounding error. The U.S. suicide rate climbed 37.5% between 2000 and 2024, and drug poisoning deaths more than tripled over the same period. People with a substance use disorder — a diagnosable pattern of drug or alcohol use causing real harm to health or life, not a moral failing — face roughly 11 times the odds of dying by suicide as people without one. An estimated 44.6 million Americans age 12 and older had a substance use disorder in the past year. Only 16% got any treatment for it.

What the Body Already Knew

This is where you come in, if you’re reading this in recovery or in the middle of active use. Ambivalence — not certainty about wanting to die, but some real, coexisting pull toward surviving even in the moment you didn’t fully want to — is not a rare or shameful footnote. It is, per Connery’s team, closer to the median experience among people who’ve overdosed and lived: 58.5% reported at least some desire to die before their most recent overdose, and about 21% described a definite wish to die, figures drawn from Connery’s ongoing work with people hospitalized for opioid use disorder. If that describes you, or described you once, it is not a symptom you mismanaged. It’s information your treatment team should be trained to ask about directly, and mostly isn’t.

The risk compounds at the exact moments the system treats as “handled.” Binswanger and colleagues found, in a landmark study of Washington State prison releases, that former inmates faced overdose death risk 129 times higher than the general population in the first two weeks after release — the same two weeks the Commonwealth Fund report credits improved prison-adjacent treatment with helping bend the national curve downward. Both things are true: prison-based medication access is genuinely saving lives, and the acute period right after release remains one of the most psychologically dangerous windows in addiction, suicide risk included, that the current system knows how to name but rarely knows how to screen for.

Necessary, Not Sufficient

Naloxone reverses an overdose. It does not ask why someone used enough to need reversing, and it was never built to. That’s not a flaw in naloxone — it’s a description of what one tool can and can’t do. The harm reduction infrastructure this country spent a decade building, and is now defunding mid-decline, is genuinely responsible for saving lives at scale. But “alive” and “screened for suicide risk” are not the same outcome, and a treatment system that measures success purely in overdose-reversal numbers will keep missing the roughly one in five patients who, per Connery’s data, walked in the door having already decided they wanted to die. Integrating routine suicide-risk screening into every point of contact with opioid and substance use disorder care — detox, methadone clinics, SAMHSA-funded outpatient programs, prison release planning — is not a nice-to-have layered on top of harm reduction. Rockett and Caine’s research says it’s the difference between an accurate map of the crisis and a comforting one.

That’s not a flaw in naloxone — it’s a description of what one tool can and can’t do.

Before the argument lands, here’s what hasn’t disappeared, whatever else has been cut: naloxone remains free and federally funded in every state, right now. The 988 Suicide & Crisis Lifeline is still answering calls and texts, 24 hours a day, and is not going anywhere this SAMHSA reorganization has touched. Medications for opioid use disorder are still covered by Medicaid today, even as that coverage is under threat. If you’re trying to find any of it near you, Rize’s Find Help tool is built for exactly that search, and it isn’t behind a paywall or a waitlist.

What Rockett is really arguing is that we have been counting the wrong thing for decades, and calling it precision. A death ruled “accidental” closes a file. A death ruled “suicide” opens a different kind of scrutiny — of the systems, the missed appointments, the person’s own repeatedly stated wish to die that a chart may have recorded and nobody flagged. The overdose numbers are getting better, and that is real, and it deserves the word Marshall used for it: unprecedented. But somewhere inside that falling number are people who were never accidentally dying. They were telling someone, sometimes in a survey, sometimes in a chart note nobody cross-referenced, and the count we built to celebrate their survival isn’t built to hear it.