The CDC Published a Cannabis Number This Month — We The People of the U.S. of A.

I went looking for anything new from the CDC on cannabis or hemp, because that is what doing homework looks like: you check the agencies whose numbers get quoted at you. There is something new, it is eight days old, and the number in it is going to travel.

The report

Published September 3, 2026

"Notes from the Field: Cannabis Detection Among Overdose Deaths — United States, 2021–2025," in the CDC’s Morbidity and Mortality Weekly Report, volume 75, number 34, pages 539–541. Four authors from the CDC’s Division of Overdose Prevention.

It uses a system called SUDORS, which collects death certificates, medical examiner and coroner reports, and postmortem toxicology from 31 states and the District of Columbia. Nebraska is one of them.

21.0%

Of 209,166 overdose deaths from January 2021 through June 2025, cannabis was detected in 43,880. The percentage held steady across those years.

The part that will get dropped

Detected is not the same as caused

Two more numbers appear in the same paragraph, and they are the ones that decide what the first number means:

Measure	Share of all overdose deaths	Count
Cannabis detected in toxicology	21.0%	43,880
Cannabis listed as causing the death	0.8%	1,687
Cannabis the only drug involved	0.004%	9

Nine deaths, in four and a half years, across 32 jurisdictions. The CDC states the conclusion plainly, so I do not have to reach for it:

MMWR, September 3, 2026, page 540 “This evidence indicates that overdose deaths caused by cannabis are rare, and detection of cannabis among overdose deaths is primarily indicative of polysubstance use.”

Among the deaths where cannabis was found, 73.5% also involved illegally manufactured fentanyls and 60.4% involved stimulants such as cocaine or methamphetamine.

So the honest sentence is long and the dishonest one is short. “Cannabis found in one in five overdose deaths” fits in a headline. “Cannabis is usually present alongside the fentanyl that did the killing” does not. Watch for the short one. It will be quoted at a hearing near you.

The finding inside the finding

The youngest group has the highest share

Among decedents aged 12–17, cannabis was detected in 42.7% of overdose deaths — 439 of 1,028. The trend inside that group moved: 49.2% in 2021, down to 35.9% in 2024, then back up to 44.6% in the first half of 2025.

The CDC notes this is the age group with the lowest self-reported cannabis use. And in that same group, 88.6% of the deaths where cannabis was detected also involved illegally manufactured fentanyls.

By contrast, adults 50 and over had the lowest detection rate, 16.7%, rising slightly from 14.8% to 18.1% over the period.

The hemp angle

Delta-8, and a testing limit worth knowing

Hemp-derived delta-8 THC shows up in the footnotes, in two ways that matter to anyone arguing about hemp products:

Among deaths with cannabis detected, delta-9 THC was found in 73.1% and delta-8 in 1.1%. Another 26.6% were recorded only as cannabis, not as a specific compound.
The routine test cannot tell delta-8 from delta-9. The CDC says the usual screening method “cannot distinguish delta-8-THC and delta-9-THC” and that separating them “necessitates confirmatory testing.”

That is a real limit, stated by the agency, on any claim about which product was involved. The CDC also says its own count is probably low: because cannabinoids rarely cause death, screening for them is not universal, so “the identification of cannabis use among overdose deaths is likely underestimated.”

What I found by accident

Two CDC pages are behind their own report

I looked for this report where you would expect to find it: the CDC’s own page listing its cannabis MMWR articles.

CDC page	Last updated	What that means
“MMWR Articles,” Cannabis and Public Health	February 15, 2024	Its newest listed article is from July 2023. The September 2026 report is not on it.
“Cannabis Facts and Stats”	March 7, 2025	Still cites 2021 use: 19%, or 52.5 million people. The new report cites a 2024 figure of 64.2 million.

That is not a scandal, and I am not calling it one. Agencies are big and web pages age. But it is a lesson I keep relearning and will keep publishing: the summary page is not the source. If I had trusted the CDC’s own index, I would have concluded there was nothing new since 2023, and been wrong by three years.

Go to the dated report. Every number above comes from the PDF of the report itself, not from a page about the report.

What it does not say

Four things I cannot tell you from this

Nothing about Nebraska by itself. Nebraska is one of 32 jurisdictions in the pool. The report publishes no state-by-state breakdown, so anyone who tells you “in Nebraska, one in five…” is going past the evidence.
Nothing about medical cannabis patients. The data is postmortem toxicology from overdose deaths. It does not describe people using a legal medical program.
Nothing about cause for those 43,880 people. Detection is a finding on a lab report, not a determination by a coroner.
Nothing that changes a page on this site. Our Nebraska pages report what the Legislature, the courts and the Commission did. This is national context, and it stays in its lane.

Why I read the whole thing

Because the first number in that report is going to be repeated at a city council, a legislature, maybe a hearing room I am sitting in on Monday. When it is, I want to be the person who can say what page it is on, what the next two numbers were, and what the authors themselves concluded.

Would you believe me if I told you a true number can still leave you with a false impression? That is the whole reason to read the source. It costs an afternoon, and the CDC put it online for free.

Primary sourceThe MMWR report (PDF) Nebraska’s recordThe dates page Every entryHomework, Published Live