The Decline Is Real. The Reason Should Unsettle Everyone.

The number released by the CDC on May 13, 2026 is genuinely good: 69,973 drug overdose deaths for the 12 months ending December 2025, a 13.9 percent decline from the prior period. Thirty-two thousand fewer dead than the 2023 peak of roughly 107,000. If you’ve been watching this crisis for any length of time, you remember what it looked like when the number went in the other direction, year after year, for a decade and a half. A 13.9 percent decline is not nothing. It is, in fact, the first sustained multi-year improvement in living memory.

The harder question — the one that determines whether we’re watching a reversal or a reprieve — is why it’s happening. A new analysis published in The Lancet Regional Health – Americas (PIIS2667-193X(25)00236-4) by researchers from the University of Maryland, the University of Chicago, and Stanford offers answers that should make anyone paying attention significantly more uncomfortable with the good news.

The treatment narrative is not the right one

The public health reflex, when overdose deaths decline, is to credit the interventions: more naloxone, more MOUD, expanded harm reduction, better treatment access. Those interventions matter. They save individual lives. They are also, according to the Lancet analysis, not the primary explanation for what happened in 2024 and 2025.

The researchers modeled two distinct mechanisms. The first: fentanyl market saturation. Fentanyl has now so thoroughly replaced heroin in the illicit opioid supply that the “saturation” effect — the mass introduction of a far more dangerous drug to a population previously using a different dangerous drug — is largely complete. The acute danger spike that comes when a drug first infiltrates a market, killing people who don’t know their tolerance has changed, is over in the fentanyl story. What remains is still devastating, but it’s a different phase. There are fewer new introductions.

The second mechanism is more significant and harder to explain: decreased drug use overall. The Lancet team estimated that if the population exposed to opioid overdose risk had remained constant at prior-year levels, there would have been roughly 109,783 additional deaths. The decline isn’t just fewer people dying per use — it’s fewer people using. Why the at-risk population contracted is not definitively established. Demographic aging of the cohort most exposed in the late 2010s. Some proportion entering treatment and sustaining recovery. Some proportion having already died. The modeling cannot fully disentangle these.

What the researchers can say is what didn’t drive the decline: there is no evidence that the decrease in mortality is primarily attributable to treatment system scale-up or harm reduction expansion.

What the Chinese government did that Congress didn’t

The Lancet analysis points to a third factor that complicates every domestic policy conversation about the opioid crisis: supply disruption. Chinese regulatory actions cracking down on the chemical precursors used to synthesize fentanyl, implemented in stages between 2019 and 2023, appear to have substantially reduced the purity and availability of illicitly manufactured fentanyl in the North American supply chain. The international regulatory action had a measurable domestic effect.

This is politically inconvenient in every direction. The public health argument that treatment and harm reduction are working is undermined by data suggesting the primary driver was a foreign government’s pharmaceutical chemical controls. The “tough on supply” argument that drug supply restrictions are the solution is simultaneously validated (supply disruption did reduce deaths) and complicated (the disruption came from Chinese regulators, not DEA scheduling or border interdiction). Neither camp gets a clean narrative.

What the data does establish is that the decline is externally dependent. Fentanyl precursor supply is not stable. The chemical suppliers that were constrained are adapting. Emerging synthetic opioids — nitazenes, metonitazene, and novel analogs — are appearing in the drug supply as fentanyl-adjacent compounds. The CDC provisional data shows fentanyl-involved deaths fell only 2 percent — far less than the overall 13.9 percent — suggesting that the synthetic opioid floor is nearly immovable. Heroin deaths fell 33 percent, but heroin is now a rounding error in the overdose landscape: the market has already moved on.

Emerging synthetic opioids — nitazenes, metonitazene, and novel analogs — are appearing in the drug supply as fentanyl-adjacent compounds.

Why this matters for infrastructure decisions made right now

The Lancet findings don’t say harm reduction doesn’t work. They say the overdose decline can’t be straightforwardly attributed to the harm reduction and treatment infrastructure we’ve been building. That distinction matters enormously when the federal government is in the middle of dismantling that infrastructure.

SAMHSA’s April 24 guidance banning federal funding for fentanyl test strips, clean syringes, and sterile water for injection came after the overdose death numbers had already started improving. The administration’s argument — that the restrictions are necessary because harm reduction “facilitates drug use” — lands differently when the data shows the decline isn’t primarily harm-reduction-driven. But the counterargument is equally clear: if the decline is driven by supply dynamics and population changes we don’t fully control, then the harm reduction infrastructure is precisely what will determine whether the trend continues when those supply conditions shift.

Supply-driven declines can reverse. The fentanyl precursor market is not permanently suppressed. The populations at risk are not permanently smaller. If the drug supply changes again — a new analog gaining market share, precursor controls loosening, a new drug entering — the harm reduction infrastructure is what stands between that shift and a renewed spike in deaths. You don’t find out that the safety net works by watching it work in good conditions. You find out when conditions change.

The 69,973 figure represents genuine progress and an open question. The question is whether the interventions we’re currently funding are the ones that will matter when the supply dynamics shift — and whether we’ll have them in place to find out.