The numbers first, from the National Center for Health Statistics provisional counts, twelve months ending December, all reported as complete:
| Year | Opioid-involved overdose deaths |
|---|---|
| 2022 | 82,851 |
| 2023 | 80,719 |
| 2024 | 55,007 |
| 2025 (provisional) | 45,008 |
That is a fall of about a third from the peak in two years, and roughly 46% by the provisional 2025 figure. In a field where every trend for two decades went the wrong way, this is genuinely remarkable.
The candidate explanations
- Naloxone saturation. Over-the-counter availability and mass distribution mean more overdoses are reversed. This would show up as fewer deaths without fewer overdoses.
- A change in the illicit supply. Composition, potency and adulterant mix shift constantly. A less lethal mix produces fewer deaths with no behavioral change at all.
- Treatment expansion. Wider access to buprenorphine and methadone, including telehealth initiation.
- A depleted susceptible population. The grimmest hypothesis: that the cohort at highest risk has been substantially reduced by the deaths of the preceding years. If this is a large part of the answer, the decline is not a policy success.
- Behavioral adaptation among people who use drugs — test strips, not using alone, smaller doses.
What it is probably not
A delayed effect of prescription reduction. That had largely happened by 2019, and deaths rose for three years afterward. A four-to-seven-year lag with a large intervening rise is not a plausible causal shape.
Why an agency should care about the uncertainty
Because the attribution determines what you keep funding. If it is naloxone, defunding naloxone because deaths are down would be self-defeating in the most direct way possible. If it is supply composition, none of the credit belongs to any program and the trend can reverse without warning. If it is treatment access, that is the thing to protect.
This is also a reason to insist on a comparison group in any evaluation. A program running from 2023 to 2025 that reports falling overdose deaths in its population has reported the national trend. See pilot design.
More: after the 2016 guideline · what population trends can tell you.
Sources
Every figure on this page is traceable to the source listed here.
- National Center for Health Statistics. VSRR Provisional Drug Overdose Death Counts (dataset xkb8-kh2a), 12-month-ending counts for the United States. Data accessed 20 August 2026. View source.
- Tanz LJ, Stewart A, Gladden RM, Ko JY, Owens L, O’Donnell J. Detection of illegally manufactured fentanyls and carfentanil in drug overdose deaths — United States, 2021–2024. MMWR Morb Mortal Wkly Rep. 2024;73(48):1099–1105. PMID 39636782. View source.