Last June, multiple mainstream media outlets reported that the number of drug overdose deaths in the United States would skyrocket in early 2025, based on data from the Centers for Disease Control and Prevention (CDC).
A new study from Northwestern University finds that the surge is not real and that the apparent spike is an artifact of statistical modeling and not a true reversal of the overdose trend. In fact, the study found that overdose deaths have continued to decline since peaking in August 2023, marking the longest sustained decline in more than 40 years.
Lead author Lorie Ann Post said the findings directly address speculation that the CDC may have misreported data due to political pressure or interference, or intentionally falsely concealed public health data.
Many people think that the CDC’s drug overdose data is manipulated, but that’s actually not the case. We can trust them because they are scientists trying to do their best work even in difficult circumstances. There was no clear incentive for any administration to inflate these numbers. This wasn’t politics. ”
Lori Ann Post, Director, Buehler Center for Health Policy and Economics, Northwestern University Feinberg School of Medicine
This study was announced today (April 8th). American Journal of Public Health.
what happened?
To create the June 2025 report, CDC scientists used data from previous years (2022 and 2023), when drug overdose deaths significantly accelerated, Post said. As such, preliminary federal data suggests a spike in overdose deaths in January 2025, sparking national concern and media coverage.
“That surge made headlines at the time, but it didn’t reflect reality,” Post said. “What we found was a mismatch between predictive models and a rapidly changing infection. CDC scientists did their best job with fewer people, more restrictions, and more people monitoring.”
Subsequent revisions to federal estimates corrected the overestimates and confirmed that overdose deaths continue to decline, Post said.
The United States relies on preliminary mortality estimates from the National Center for Health Statistics, which use statistical models to account for delays in reporting deaths. Those models performed well for years, but struggled when overdose deaths began to decline after a long period of rapid growth, Post said.
The study authors emphasize that this episode highlights a broader issue. This means that surveillance systems are most vulnerable during transitions when trends change direction.
why is it important
Accurate data is essential to public health responses. Misinterpreting trends can misdirect policymaking, undermine public trust and distort resource allocation, Post said.
Despite such episodes, researchers emphasize that federal mortality data remains the most reliable near-real-time source for tracking overdose deaths.
Why the model broke
The overdose crisis has been transformed by the synthetic opioid fentanyl, which has spread unevenly across the country and caused a spike in deaths from 2022 to 2023.
The models trained during that explosive growth were applied during subsequent periods of decline. result:
- Overestimation of deaths in early 2024 and 2025
- False signals of nationwide “spike”
- Confusion among policy makers, researchers, and the public
Looking to the future
The authors call for greater transparency in federal data systems, including advance notice of methodological changes and clear documentation of revisions.
“When the numbers change, people notice it,” Post said. “We need to help them understand why.”
Learn more about the research
Researchers used OD Pulse, Northwestern’s national, regional, and state dashboard designed to track drug overdose deaths from January 1999 to April 2025 with unprecedented timeliness and accuracy. They analyzed national overdose death data from 2015 to early 2025 and compared the observed numbers to multiple versions of federal projection estimates.
The title of the paper is “The 2025 drug overdose surge was not: neither politics nor data errors explain the anomaly.” Co-authors include researchers from the University of California, San Francisco, the University of Maryland, Baltimore, and the University of North Carolina at Chapel Hill.
sauce:
Reference magazines:
DOI: 10.2105/AJPH.2025.308412
