The Overdose Decline Is Real. The People Still Dying Are Increasingly Black, Indigenous, and Killed by Stimulants.

The headline number is real: U.S. drug overdose deaths dropped 24.4 percent between 2023 and 2024. A study published yesterday in the journal Addiction — led by Joseph Friedman, a resident physician at UC San Diego, with co-authors at NYU, UCSF, and UCLA — confirmed what provisional CDC data had suggested for months. Roughly 80,000 Americans died of overdoses in 2024, down from approximately 105,000 the year before. It is the largest single-year decline ever recorded. And it is the first time the crisis has retreated across all four of the epidemiological “waves” researchers use to describe the overdose era.

This is genuinely good news. It should be acknowledged as such.

And then we should talk about the rest of the numbers.

What drove the decline — and what that means

The dominant driver was fentanyl. Deaths involving fentanyl alone dropped from 31,193 in 2023 to 19,673 in 2024 — a 37 percent reduction. Deaths involving fentanyl combined with stimulants fell from 41,583 to 28,062. Fentanyl was the engine of the crisis, and fentanyl appears to be the engine of the improvement.

Why? The research team was careful here, and it’s worth being careful in turn. Friedman told UC San Diego that the causes are “likely complex and multifactorial.” The decline may reflect the cumulative effect of expanded naloxone access, buprenorphine and methadone treatment reaching more people, fentanyl test strip distribution, harm reduction infrastructure that has been built over years. It may also reflect something about the illicit drug supply itself: a December 2025 preprint in MedRxiv argued that decreased fentanyl potency at the street level — a market phenomenon, not a policy one — may be a primary driver of the improvement. Both things can be true simultaneously. Treatment and harm reduction expanded. The supply also changed. Parsing which one did more work is an empirical question that will require years of data to answer.

What the data cannot sustain is the interpretation — which will arrive in press releases within weeks — that the problem is solved and the policies that have been cut or constrained over the past 18 months should stay cut.

The counter-trend: stimulant-only deaths are rising

In 2023, stimulant overdose deaths that did not involve opioids accounted for 17.3 percent of all overdose fatalities in the United States. In 2024, that figure rose to 23.8 percent.

That is not a rounding error. Stimulant-only deaths rose from 18,142 to 18,907 — an increase of roughly 4 percent in absolute numbers — even as total overdose deaths fell by nearly 25 percent. The denominator shrank. The stimulant numerator did not. A corresponding CDC MMWR study published this year, tracking overdose deaths involving stimulants from January 2018 through June 2024, showed that 59 percent of all overdose deaths during that period involved stimulants, with 43.1 percent co-involving opioids and 15.9 percent involving stimulants without any opioid.

This matters for two interconnected reasons. First, there is no FDA-approved medication for cocaine use disorder and no FDA-approved medication for methamphetamine use disorder. The treatment system’s most powerful tools — buprenorphine, methadone, naltrexone — are all developed for opioids. The people now dying from stimulants without any fentanyl in their system are dying in a treatment landscape that has almost nothing pharmacological to offer them. Contingency management — rewarding abstinence with small incentives — remains the only rigorously evidence-based intervention for stimulant use disorders, and it remains inconsistently covered by Medicaid and most commercial insurance.

Second, stimulant deaths disproportionately kill specific populations. Cocaine deaths, per the Friedman study, disproportionately affect Black Americans, who already face systemic barriers to addiction treatment access at every level of the care system. Methamphetamine deaths disproportionately affect American Indian and Alaska Native populations — which brings us to the racial data.

Methamphetamine deaths disproportionately affect American Indian and Alaska Native populations — which brings us to the racial data.

The racial disparities that don’t fit the headline

Even with a 24.4 percent overall decline, Black Americans faced an overdose mortality rate 1.5 times higher than the national average in 2024. They also saw a 29.3 percent rate decrease — meaning the improvement did reach them, and proportionally. But 29 percent of a rate that was already 1.5 times the national baseline still leaves a gap that a national headline cannot paper over.

American Indian and Alaska Native populations face an overdose rate of 50.8 deaths per 100,000 people — more than double the national average of 23.7. This is the highest rate of any racial or ethnic group measured in the study. The geographic concentration of these deaths, the limited treatment infrastructure in many reservation and rural tribal communities, and the specific dominance of methamphetamine in these mortality patterns represent a distinct crisis within the declining aggregate numbers.

Friedman’s own quote deserves reproduction in full here: “We are seeing a historic shift in the overdose crisis. But this is not the end…We need to avoid interpreting declining national numbers as a sign that the crisis has been solved.”

What to watch

Three things the data cannot yet answer:

First, whether the decline holds into 2025 and 2026. Preliminary projections suggest approximately 70,000 deaths for the 12 months ending December 2025 — continued improvement, but slower. The deceleration matters. If the easy gains from naloxone expansion and expanded MAT access have been captured, and the remaining deaths are concentrated in stimulant-using, rural, Indigenous, and marginalized Black communities where treatment infrastructure is thinnest and insurance coverage is most fragile, the next phase of reduction will require something different from the last.

Second, the supply question. A December 2025 MedRxiv preprint makes a case that decreased fentanyl potency — not expanded treatment — was the primary driver of 2024’s improvement. If that’s true, the gains could reverse as quickly as a supply shift. The illicit drug market does not negotiate with public health policy.

Third, the policy headwinds. SAMHSA’s budget has been cut. Harm reduction programs have been defunded. The Medicaid work requirements described elsewhere in today’s digest may reduce coverage for the very populations that accounted for the improvement — people who accessed buprenorphine or naloxone through a Medicaid-funded provider. The gains documented in the June 2 Addiction paper were built on policies. Some of those policies are being dismantled in real time.

The decline is real. Keep it.