Half of What Toddlers Swallow by Accident Is Now an Edible. The 25-Year Study Wants Child-Proof Packaging.

Researchers at Nationwide Children's Hospital pulled 25 years of poison center records and found 41,612 cases of a child under 6 swallowing a recreational drug nobody meant for them to find. Marijuana edible preparations account for 47.9% of them. The ingestion rate per 100,000 children rose 3,307% between 2000 and 2024, the figure that will appear in every hemp ban floor speech between now and November, and the least useful figure in the paper. The 2000 baseline was 1.05 per 100,000. Multiply a near-zero by a large number and you get a large number.

The findings that matter to anyone who buys edibles as medicine sit further down. Of these ingestions, 96.4% happened in a residence and 57.6% involved children younger than 3. What the authors ask for is better packaging. If you keep edibles in the house, buy the boring child-resistant container and store it the way you store a prescription bottle. Read this study as a packaging argument, because that is the argument its authors made.

One limit is worth stating before any of these numbers get used against your dispensary. The National Poison Data System does not record where a product came from. A dispensary gummy and a gas station gummy enter the record as the same thing.

What did the 25-year study actually find?

Between 2000 and 2024, US poison centers logged 41,612 unintentional ingestions of recreational drugs by children under 6. Cannabinoids accounted for 84.7% of them, marijuana products for 77.0%, and marijuana edible preparations alone for 47.9% of every case in the dataset. The overall rate climbed from 1.05 per 100,000 children in 2000 to 35.91 in 2024.

The curve is flat for thirteen years and then it is not. Almost all of the movement comes after 2013, when the rate jumps 2,286.2% from 1.51 to 35.91. That is also the year the first edible marijuana ingestion appears, which the authors flag as partly a coding artifact: the generic poison center code for marijuana edibles was not created until 2016, so earlier exposures existed and were invisible to the database.

Semi-synthetic cannabinoids, the paper's label for the delta-8 family, account for 7.0% of ingestions (2,897 cases). That category is the only window the dataset offers into hemp-derived products, and it is a narrow one.

Are edibles the most dangerous thing a toddler can find?

No. Compared with all other substances in the study combined, marijuana edible ingestions were less likely to produce a major effect (RR 0.55, 95% CI 0.50 to 0.61), a medical admission (RR 0.90) or a critical care unit admission (RR 0.74). Opioid ingestions carried a major-effect risk ratio of 6.52. Edibles drive the volume. Opioids drive the severity.

Both halves are worth holding at once. Marijuana edibles produced 555 major effects and 4,982 medical admissions, 32.8% of all major effects and 44.2% of all admissions. A category that is half the cases lands in the hospital more often than anything else in raw counts while being the gentlest thing per incident. Cannabinoid ingestions ran a 3.8% major-effect rate against 27.6% for opioids, 11.0% for dissociative agents and 9.1% for stimulants. All six deaths in 25 years involved fentanyl (four) or cocaine (two).

The newest line in the dataset is one nobody tracked five years ago. Psychedelic ingestions rose 827.3% between 2018 and 2024, psilocybin and psilocin mushrooms up 1,466.7%, driven by the mushroom gummy aisle that appeared in smoke shops without anyone deciding it should. Independent testing of those products has turned up kavalactones, pregabalin and acetylpsilocin alongside the mushrooms. A 25-year epidemiological paper logs the entire copycat-candy era in the tone of an inventory sheet, and the flatness is the most damning thing about it.

Where are children finding these products?

At home. The study found 96.4% of ingestions occurred in a residence, 57.6% involved children younger than 3, and 95.9% involved a single substance. The mechanism the authors describe is ordinary: products that resemble candy or food, packages that hold many servings at once, and adults consuming in front of children who copy what they see.

Hannah Hays, medical director of the Central Ohio Poison Center and a co-author, put the storage guidance as keeping products "stored up, away and out of sight and reach of children," preferably locked. Writing about the study for CNN on September 10, emergency physician Leana Wen was blunter about the usual half-measure: "Putting something on a high shelf is not sufficient." Wen also named what makes edibles hard on a caregiver, which is that onset runs an hour or more behind ingestion, so a child can finish several pieces before anyone knows there is a problem. Our guide to overconsumption covers the same delay from the adult side.

If it happens, the number is Poison Help, 800-222-1222, and the call is free and staffed around the clock. Make it before symptoms start rather than after.

What do the researchers actually want changed?

Packaging law. The prevention paragraph asks for "opaque, child-resistant packaging" with limits on the total dose per serving and the number of servings per package, single-serving formats, and adequate labeling. Nothing in the paper calls for prohibition. The senior author, Gary Smith, framed the finding as increases "especially pronounced for edible marijuana and psilocybin products."

Here is the part that gets left out when the 3,307% goes to Congress. Most adult-use states already impose that entire list on licensed operators, with per-piece dose caps, per-package totals, opaque child-resistant containers and universal THC symbols, enforced through license conditions and recalls. The product the study describes is more often the one sold outside that system. When Missouri's governor signed the state's intoxicating hemp bill in April, the Missouri Independent reported smoke shop products carrying as much as 1,000 mg of THC per package, a hundred times a standard dispensary serving, in a container nobody licensed.

That is the gap the federal hemp ban closes with a blunt instrument, capping hemp products at 0.4mg total THC per container starting December 11, with non-natural cannabinoids losing hemp status November 12. A dose cap per container is a packaging rule wearing a prohibition costume. The researchers asked for the rule. Congress wrote the costume.

What the numbers cannot tell you

The National Poison Data System runs on voluntary reporting, so it undercounts, and the authors say so plainly. Cases lost to follow-up introduce bias. The risk ratios are unadjusted. "Marijuana edible preparations" is a category a poison center specialist assigns during a phone call, not a lab result. And the database captures no product source, so no line in this paper separates a regulated dispensary edible from a hemp gummy from something homemade.

Which is why the 3,307% will be misused within the week. It is a real number attached to a real problem, and it cannot carry the argument people will hang on it, because the dataset behind it is blind to the one variable that argument turns on. What survives scrutiny is narrower: half of what small children accidentally eat is now an edible, almost all of it at home, and the fix the researchers name is a container. Regulators have moved slowly on far better-documented harms than this one, so the drawer in your house will stay ahead of the rulemaking for a while.