Why Are Overdose Deaths Falling So Fast? A New Report Says There's No Single Answer

Something genuinely unusual happened to America's overdose numbers between 2023 and 2024. Deaths dropped from 105,007 to 79,384 — a 27 percent decline in a single year, the kind of shift that normally takes a decade of sustained public health effort, not twelve months. Provisional 2025 data suggest the number kept falling, to somewhere around 70,000.

"It is unprecedented to see overdose deaths come down so dramatically, so quickly," Brandon Marshall, an epidemiologist at Brown University who studies overdose trends, told STAT News. He also added the caveat that matters just as much: "we're nowhere near out of this crisis."

A new Commonwealth Fund report, published September 3 and authored by senior research associate Evan Gumas, set out to answer the obvious follow-up question: what actually caused this? Gumas spent months talking to public health and addiction officials in dozens of U.S. states and several other countries, and the answer he kept getting back was consistent in an almost frustrating way. There wasn't one thing. There were several, layered on top of each other, built up over years — and the report is specific enough about which ones, and where, that it's worth walking through in real detail rather than settling for the one-sentence version.

01The numbers, more precisely than "overdose deaths are falling"

The national 27 percent drop hides a lot of variation worth understanding on its own. Nebraska had the lowest overdose death rate of any state in 2024, at 8 per 100,000 people. West Virginia had the highest, at 45 per 100,000 — and yet West Virginia also posted one of the steepest percentage declines in the country, a 42 percent drop, translating to 556 fewer deaths. In raw numbers, California (down 2,350 deaths, a 23 percent decline), Florida (down 2,052, a 33 percent decline), and New York (down 1,881, a 32 percent decline) saw the largest absolute drops.

Only one state moved in the wrong direction: South Dakota recorded three more overdose deaths in 2024 than in 2023, essentially flat, but the sole exception to an otherwise universal decline across every U.S. state and territory.

The international picture complicates the "something structural changed" narrative in an interesting way. Scotland (-13 percent) and Canada (-9 percent) both saw meaningful declines over the same period, and Puerto Rico dropped 21 percent. But several other countries moved sharply in the opposite direction: Australia rose 9 percent, Brazil 21 percent, Argentina 32 percent, Türkiye 42 percent, and Northern Ireland 53 percent. Whatever's driving the U.S. and a handful of peer countries downward clearly isn't a universal global trend — which is part of why the report spends most of its length on specific, comparable policy choices rather than a single unifying theory.

02Five things researchers say are actually doing the work

Gumas organized what he heard into five recurring categories. None of them alone explains the full picture, but together they form a genuinely detailed account of what changed on the ground.

Naloxone saturation. This is the idea that enough of the opioid overdose reversal drug is available in a community that it's on hand whenever an overdose happens, rather than something that has to be specifically sought out in advance. SAMHSA required states receiving 2022 State Opioid Response funding to build a "naloxone saturation" plan, and states that took this seriously show up repeatedly in the report's findings. Virginia's REVIVE! program, running since 2013, made naloxone largely free to high-risk populations statewide following its 2023 saturation plan. Maryland more than doubled its distribution between 2021 and 2025, now moving over 460,000 doses a year, with growth concentrated in low-barrier settings — vending machines, EMS "leave-behind" programs, outreach teams working directly with people who use drugs. New York built the country's first online portal letting any state resident request free naloxone, distributing 390,000 kits since 2022. Ohio has offered no-cost naloxone since August 2022, including vending machines placed inside prisons and at highway rest stops.

Not every saturation effort is landing the same way, and the report is honest about that. Florida's naloxone distribution nearly doubled between 2018 and 2024, reaching 767,931 doses — but reported overdose reversals stayed roughly flat over the same period, which the report suggests may mean distribution is starting to outpace actual use. And Colorado's naloxone fund reaches all 64 counties at no cost, but the funding isn't permanent, leaving the state's long-term saturation genuinely uncertain.

Broader harm reduction. Fentanyl test strips and similar drug-checking tools let someone see what's actually in a substance before using it. The report notes that direct evidence linking test strips to lower overdose death rates specifically is still mixed, but there's strong evidence they change behavior in ways that plug into other harm reduction strategies — someone who checks a substance and finds unexpected fentanyl is more likely to use less, use with someone else present, or seek other precautions. Ohio legalized fentanyl test strips in 2023 and, by executive order in October 2025, moved to decriminalize testing equipment for emerging substances like xylazine and medetomidine before they even show up as a crisis, rather than reacting after the fact. Maryland's Rapid Analysis of Drugs program collects samples from harm reduction participants and sends them to the National Institute of Standards and Technology for lab analysis, feeding results back into where the state directs resources. Oregon's Save Lives Oregon initiative has distributed more than 770,000 free naloxone doses since 2021, with over 27,000 documented overdose reversals — and deliberately doesn't condition that supply on someone entering treatment.

Expanded coverage for medication treatment. This is where Medicaid does most of the heavy lifting. The report calls Medicaid expansion "the single most influential lever" for getting people access to medications for opioid use disorder — methadone, buprenorphine, and naltrexone — in states where a large share of the population is enrolled. West Virginia, where roughly 30 percent of residents are on Medicaid, and Virginia, at roughly 20 percent, both saw meaningful MOUD access improvements tied directly to that coverage base. Colorado's methadone clinic count grew from 25 to 54 in five years. New York's methadone access expanded from 29 to 43 counties, with the number of programs growing from 98 to 137, plus mobile medication units reaching people who can't easily get to a fixed clinic. Maryland launched a program training paramedics to administer buprenorphine on-site immediately after a nonfatal overdose — treating the moment right after someone survives as the opening to get them started on medication, rather than something to follow up on later once the crisis has passed.

Reaching people leaving incarceration. People recently released from prison face a dramatically elevated overdose risk, concentrated specifically in the two to four weeks right after release — tolerance drops during incarceration while access to the same drug supply often returns quickly. In the U.S., substance use disorders are roughly 12 times more common among people entering prison than in the general population, and opioids account for more than half of post-release overdose deaths. Every jurisdiction Gumas spoke with flagged this window as critical. New York now provides services across all 58 jails and 41 state prisons, reaching close to 20,000 people, with medication for opioid use disorder use increasing sevenfold between 2022 and 2025. Maryland's 2024 Medicaid waiver extends coverage to incarcerated people 90 days before their release date, not after — a pilot currently running in two facilities ahead of a planned statewide rollout. Ohio provides naloxone kits to everyone leaving custody and, in 2024, began piloting long-acting injectable buprenorphine at release.

Better data, faster. The last category is less visible but arguably foundational to everything else working. Rhode Island's Overdose Spike Alert System triggers an automatic public health response once nonfatal overdoses cross a threshold of 55 cases in a week, letting resources move to a specific area while a spike is still happening rather than after. Ohio's RecoveryOhio Early Warning System uses emergency medical reports, naloxone administration data, and law enforcement information to predict which zip codes face elevated overdose risk up to 30 days in advance. An Oregon-Idaho data-sharing agreement caught a measurable decline in the drug supply's fentanyl potency between 2023 and 2024 — a shift that likely wouldn't have been detected without two states specifically comparing notes across a border. In Missouri, a spike in deaths among Black men in North St. Louis led the state to direct resources specifically to Black-led community organizations running drop-in centers in the area; the report notes that reported deaths there have since fallen sharply.

03The parts of this story that don't fit neatly into a policy win

It would be a disservice to present this as a straightforward story about good policy choices paying off, and the report itself resists that framing. Sarah Wakeman, an addiction medicine physician at Mass General Brigham who wasn't involved in the research, offered a more difficult explanation to STAT alongside the policy factors: "so many people died during the years of really accelerating death rates that actually, there's not as large of a population at risk of dying." Years of catastrophic losses may have shrunk the pool of people most vulnerable to a fatal overdose, which is a genuinely uncomfortable thing to hold alongside the good news.

The drug supply itself is also shifting in ways nobody fully controls. Fentanyl's potency has been declining, and while state surveillance systems can track that shift, they can't explain why it's happening or guarantee it continues. Marshall, who reviewed the Commonwealth Fund report before publication, told STAT that "the dominant hypothesis is that it's mostly a function of the drug supply" when it comes to explaining why some states and regions vary so much from others — a more skeptical read on how much credit specific state policies deserve versus forces largely outside any state's control.

And the trend isn't guaranteed to hold. Provisional 2025 data already show upticks in several states, and in Scotland specifically, after 2024's decline. The report is blunt that "mortality declines in the U.S. and internationally are a step in the right direction, but continued efforts to reduce preventable deaths due to drug use are far from over."

The international comparison cuts both ways here too. Argentina's 32 percent increase and Brazil's 21 percent increase in overdose deaths both coincide with broader social service and harm reduction disruptions tied to economic austerity in those countries — a reminder that these numbers move in the other direction just as fast when the underlying support systems erode. Canada offers a nearer-term version of the same warning: despite what the report describes as overwhelming evidence of their effectiveness and cost-effectiveness, Alberta and Ontario ended provincial funding for supervised consumption sites as of June 2026.

04Where this connects directly to U.S. policy right now

This is the part of the report worth sitting with longest if you've been following the federal policy shifts already covered on this site. Gumas's analysis doesn't just describe what's working — it explicitly names the two biggest current threats to continuing it, and both are already in motion.

The first is SAMHSA's April 2026 guidance withdrawing federal grant funding for fentanyl test strips, syringes, and related harm reduction supplies — the same policy shift covered in depth elsewhere on this site. The report cites that guidance directly as a risk to the progress documented in its own findings, since drug-checking tools and broader harm reduction access are among the five categories it identifies as contributing to the decline in the first place.

The second is Medicaid. The report calls Medicaid expansion the single most influential factor behind improved access to medication treatment, and separately cites a KFF analysis warning that federal Medicaid funding reductions would significantly and adversely affect access to both MOUD and naloxone — the same two tools the report spends most of its length documenting as effective. Put simply: the financing mechanism the report credits most for the last two years of progress is the same one facing the most direct funding threat going forward.

None of this means the $77 million in new SAMHSA grants announced in late August, also covered on this site, is irrelevant — prevention, treatment, and crisis response funding all matter. But measured against a Medicaid system that the report's own author calls the single most influential lever in this entire story, $77 million is a modest counterweight to a much larger structural risk.

05What this actually means if you're trying to make sense of it

If you're evaluating a treatment program right now, or trying to understand how serious the overdose crisis still is, the honest takeaway from this report is neither "it's solved" nor "nothing has changed." Roughly 70,000 Americans are still projected to have died of an overdose in 2025 — a number that would have been treated as an unprecedented national catastrophe a decade ago, even after two straight years of sharp decline. The specific tools behind that decline are well documented and mostly not mysterious: naloxone that's actually accessible in the moment it's needed, medication treatment that people can get and afford, support that starts before someone leaves incarceration rather than after, and data systems fast enough to catch a local spike before it becomes a trend. Programs and states that have invested seriously in those specific things are the ones showing up throughout this report with real, measurable results attached to their names.