The Decline That Isn't Reaching Everyone

In November 2024, a counselor at a Phoenix harm reduction organization — call him Marcus — started keeping a second list. The first list was his caseload: the 28 people with opioid use disorder he was actively supporting. The second list had just one column: the names of people from the first list who had died.

By March 2025, four names were on it. By June, six. Marcus, who has been in recovery himself for eleven years and has worked in harm reduction in Maricopa County for seven, had never lost six clients in eight months. Not even during 2022, the peak year. “The number is supposed to be going down,” he told a colleague over the phone in late April, reading a headline about the national overdose decline. “It is not going down here.”

He was right. And the data, released in full by the CDC on June 17, 2026, confirms it — if you know how to read past the headline.

The Number the Headline Uses Is Real. It’s Also Incomplete.

The CDC’s National Vital Statistics System released provisional data this week showing 69,147 drug overdose deaths in the 12-month period ending January 2026. That’s a 13.2% decline from the same period a year earlier. Overdose deaths have now fallen from their peak of 111,451 in the 12 months ending August 2023 — a reduction of more than 40,000 deaths over roughly two and a half years. The longest sustained drop in decades.

That is real. It is significant. It is not the whole story.

Buried in the demographic breakdowns — in the regional and racial subsections that rarely make the press release headline — is a second story. The 15-month decline that began in August 2023 applied, in the CDC’s own data, primarily to white Americans. For Black Americans, American Indian and Alaska Native people, Hispanic and Latino Americans, and multiracial individuals, deaths continued rising during that same period. A peer-reviewed analysis published in PMC in June 2026 put the numbers plainly: nearly 5,000 more people of color died from overdoses in 2023 than in 2021, while deaths among white Americans dropped by more than 6,000.

That is not a marginal asymmetry. That is the difference between a declining epidemic and a shifting one — one that is moving away from the communities that have had the most access to naloxone, buprenorphine, syringe services, and treatment infrastructure, and toward the communities that have had the least.

Arizona Is Running the Controlled Experiment No One Asked For

The national number — 13.2% down — is an average. Averages obscure geography. And the geography of this epidemic, right now, is actively hostile to any celebratory reading of the data.

Arizona is the clearest counterexample. A paper published in Drug and Alcohol Dependence in May 2026 documented what local clinicians like Marcus already knew: Arizona overdose deaths were up 20% in the first eight months of 2025 compared to the same period in 2024. Deaths involving fentanyl specifically were up 40%. January, February, March, and April of 2025 each set new monthly records for Arizona overdose deaths, going back to 2018.

In Maricopa County — the Phoenix metro, home to roughly 65% of Arizona’s population — fentanyl was involved in 59% of all fatal overdoses in 2024, methamphetamine in 67%, and the county averaged more than three fentanyl deaths per day. The county health department’s overdose dashboard, updated monthly, shows those numbers climbing.

The communities hit hardest within Arizona are not random. Hispanic Arizonans saw a 43% increase in overdose deaths in that 2025 window. Black Arizonans, 25%. The 18-to-24 cohort saw a 35% jump; 25-to-34-year-olds, 29%. These are the populations with the fewest treatment options in the Phoenix metro. They are also the populations Rize Recovery was built to serve.

These are the populations with the fewest treatment options in the Phoenix metro.

Two Theories for the Decline. One Is More Dangerous to Be Wrong About.

The scientific community is genuinely split on why the national decline is happening. The answer matters for policy — and for what happens next.

Theory one: infrastructure. The expansion of naloxone access (three brands now available over the counter, following the FDA’s June 16 approval of Rextovy), the post-DATA-waiver surge in buprenorphine prescribing, the billions in opioid settlement funds flowing to states, the decade of harm reduction infrastructure building — all of this, the argument goes, is finally working. The decline reflects the overdose prevention system functioning at scale for the first time.

Theory two: fentanyl potency. A December 2025 preprint — not yet peer-reviewed — argued that the primary driver of declining deaths is a reduction in fentanyl’s potency and purity in the illicit supply, based on DEA data showing fentanyl potency fell contemporaneously with the death decline. The drug got less deadly, and deaths fell as a result. Infrastructure deserves less credit than it’s getting.

The CDC’s own statement on the June 17 data release called the decline “multifactorial” and declined to assign relative weights to the competing explanations.

Here is why the stakes on this question are high: if Theory One is correct, and infrastructure is the driver, then the current federal rollbacks to harm reduction funding are not merely counterproductive — they are actively killing people. The communities where deaths are still rising are disproportionately the communities where harm reduction infrastructure has always been thinnest. Syringe service programs, naloxone distribution, community-based buprenorphine — these are exactly what’s being defunded. If infrastructure works, cutting it works in the other direction.

If Theory Two is correct, and fentanyl potency drove the decline, then the good news comes with a warning: drug supply potency is not under policy control. It fluctuates with trafficking patterns, Chinese precursor availability, cartel economics. A supply that became less dangerous could become more dangerous again. The infrastructure you don’t build when you have the chance is the infrastructure you won’t have when you need it.

In neither scenario is defunding harm reduction a good idea.

What the Racial Gap Tells You About Who the Infrastructure Was Built For

Maia Szalavitz, writing for Filter in April 2026, put the racial disparity question as directly as it needs to be put: “The decline in overdose deaths is real. It is also a white decline.” The communities where deaths are rising, she noted, are communities that have faced decades of barriers to the exact interventions that are credited with producing the national decline: naloxone access, buprenorphine prescribing, syringe services, low-threshold treatment.

The data on this is not subtle. A study published in the American Journal of Preventive Medicine tracked overdose mortality by race from 1999 to 2022 and found that mortality had increased 249.3% among Black Americans over that period — a trajectory that has not reversed. American Indian and Alaska Native mortality rose 166.3%. Hispanic and Latino mortality, 171.8%. White mortality rose substantially too — but from a lower base and with access to treatment infrastructure that was built, historically, with white patients in mind.

The buprenorphine prescribing gap is perhaps the starkest example. Research going back more than a decade has documented that Black patients with opioid use disorder are significantly less likely to receive buprenorphine than white patients with identical clinical presentations. The DATA 2000 waiver elimination in 2022 removed the formal barrier to prescribing, but it didn’t remove the prescribing deserts — the primary care networks and emergency departments in predominantly Black neighborhoods where no one prescribes it. Regulatory equity is not the same as access equity.

Marcus, the Phoenix harm reduction counselor, doesn’t need the research to explain this to him. “The clients I’m losing are not the clients with Cigna,” he said. “They’re the clients on AHCCCS or nothing, living in zip codes where the nearest clinic that prescribes bupe is a forty-five-minute bus ride away and closes at 4 p.m.”

The Arizona Story Is Both Specific and a Warning

Arizona’s counter-trend isn’t random, and it isn’t simply bad luck. It reflects a specific set of converging pressures: border proximity (Arizona is a primary entry point for fentanyl-laced pills manufactured in Sinaloa), a service geography that leaves rural communities in Pima, Yuma, and Mohave counties severely underserved, and a demographic composition that includes some of the exact populations experiencing the sharpest national increases — young Hispanic men, in particular.

The state’s response is mixed. The SOR IV grant — $34.8 million in year one of a three-year federal award — is funding naloxone distribution through AHCCCS and Sonoran Prevention Works. Arizona’s share of the One Arizona opioid settlement, ultimately worth up to $1.215 billion over 18 years, is beginning to flow into treatment infrastructure. These are real resources.

The SOR IV grant — $34.8 million in year one of a three-year federal award — is funding naloxone distribution through AHCCCS and Sonoran Prevention Works.

But the settlement funds move on a bureaucratic timeline. The SOR IV year-one period is half over. And the deaths are happening now — in months, not years. The organizations doing the fastest-response work — mobile harm reduction vans, low-barrier bupe clinics, peer recovery specialists showing up at emergency departments — are precisely the organizations whose funding is most precarious.

“We’re building the plane while people are falling out of the sky,” said one harm reduction coordinator at a Phoenix needle exchange, who asked not to be named because their organization’s federal funding status is under review. “The money that’s supposed to help is coming. The people dying can’t wait for it.”

The Number Will Change. The Question Is Who Benefits.

The 13.2% national decline is not a finish line. It is a data point in a trend that has no guarantee of continuing — and that, by the available evidence, has not reached the communities that need it most.

The fentanyl supply could become more potent again. The federal harm reduction infrastructure — built over a decade and now being dismantled — may or may not be rebuilt. The opioid settlement funds will flow, eventually, to some purposes and not others, depending on which state and county governments make which decisions.

What the June 17 data actually shows is that the overdose response is working — partially, unevenly, for some people and not others. That partial success is being reported as general good news. It is not.

In Maricopa County, Marcus added a seventh name to his list in May. He is not celebrating the national numbers. He is watching the dashboard on the county health department website, the one that updates monthly with new deaths, and he is waiting for it to start moving in the right direction.

It hasn’t yet.

---

Coverage of the overdose crisis, racial disparities in treatment access, and Arizona-specific trends is tracked in the Rize Newsroom opioids section and the The Crisis, By the Numbers category. For real-time facility search by insurance and location, visit /find-help.