The tejocote that was never tejocote

Tejocote is the Mexican hawthorn, Crataegus mexicana. It is a real plant with a real culinary tradition, and for about a decade its root has been sold online as a weight-loss supplement, mostly through social media and third-party marketplaces.

Yellow oleander is a different plant entirely. The FDA calls it Thevetia peruviana; the CDC and most current botanical sources call it Cascabela thevetia, which is the accepted name for the same species. It is an ornamental shrub, it is native to roughly the same part of the world, and every part of it contains cardenolides: cardiac glycosides that act on the heart the way digoxin does, because they are chemically the same kind of molecule.

The two plants are not related. They are not in the same family. They do not look alike as whole plants. Chopped into brown pieces in a bag, or milled into a capsule, they look like each other and like everything else.

What FDA's alert actually says

The FDA safety alert, last updated 17 September 2026, states the finding plainly:

FDA analysis has determined that certain products labeled as tejocote (Crataegus mexicana) root or Brazil seed are adulterated because they contain yellow oleander (Thevetia peruviana) instead of the labeled ingredients.

The word doing the work there is instead. This is not a contamination story, where a correct ingredient picked up something it should not have. The labelled ingredient is absent and a toxic species is in its place. In regulatory language that is economically motivated adulteration or, more bluntly, substitution.

Three products were added on 17 September. Two are Niwali-brand tejocote products that the FDA tested itself. The third is the one that matters, and I have not seen a single news write-up lead with it: Easy Forte Natural Fiber, which is not sold as tejocote at all. It is sold as a fibre supplement. It was tested by a state laboratory under the FDA's Laboratory Flexible Funding Model, and it contained yellow oleander.

The FDA says so directly, in a line added to the alert some time ago and easy to skim past:

Additional investigation by FDA has determined that some dietary supplements that are not labeled to contain tejocote root may also contain toxic, yellow oleander.

That single sentence changes the shape of the problem. If the substitution were confined to products labelled tejocote, a consumer could avoid it by avoiding one word. It is not confined there. A shopper reading labels has no way to know which botanical weight-loss product happens to have been filled from the same supply.

I counted the table on the alert page on the day I wrote this: 28 rows, listing more than thirty product names, several of which repeat across different sellers and marketplaces. Coverage of the September update reported the list as 33 products, which is a defensible count of the name entries rather than the rows. Either way, the list has been growing since 2023 and the FDA states that it will keep adding to it.

The finding underneath the alert

The FDA's investigation did not start with the FDA. It started with a phone call to a poison centre.

On 8 September 2022 an emergency physician called the New Jersey Poison Information and Education System about a 23-month-old child who had swallowed a tejocote root product his mother had bought as a weight-loss supplement. The child was bradycardic (heart rate 90 beats per minute against a normal range of 98 to 135) and hypotensive at 71/60. The electrocardiogram showed sinus bradycardia, frequent premature ventricular complexes and scooped ST segments, which is the classic picture of cardenolide toxicity.

The serum digoxin assay came back at a low level, which the poison centre read as cross-reactivity with a non-digoxin cardiac glycoside rather than as a real digoxin exposure. The child was given digoxin-specific antibody fragments, the standard digoxin antidote. The ECG and blood pressure normalised. Twelve hours later the ECG showed cardenolide toxicity again, a second dose was given, and it normalised again.

That case, written up by Berland, Kababick, Santos and Calello in MMWR in September 2023, is the report the FDA's alert page cites as the origin of its investigation. But the case is not the part of that paper that should change how you shop. This is:

The poison centre bought ten products labelled tejocote online in December 2022, each from a separate listing, and shipped them to Flora Research Laboratories in Oregon. Using ultra-high pressure liquid chromatography with accurate-mass time-of-flight mass spectrometry, and comparing against an authenticated tejocote reference obtained with the help of an ethnobotanist, they got this result:

Nine of 10 products labeled as tejocote were yellow oleander, with no evidence of tejocote root.

Nine out of ten. And read the published table for the tenth, because it is the detail everyone drops. The tenth product, a tejocote seed liquid drop, tested negative for yellow oleander and negative for tejocote root. It was not the honest one. It was the one that contained neither.

So of ten products sold as tejocote root, zero contained tejocote root.

The rule that already required this to be caught

Here is the part that should be uncomfortable for anyone in this industry, and I am in this industry.

Federal law already requires the test that would have caught every one of these. The dietary supplement good manufacturing practice rule, 21 CFR Part 111, has been in force since 2007. Section 111.75(a)(1)(i) says:

Conduct at least one appropriate test or examination to verify the identity of any component that is a dietary ingredient, unless you petition the agency under paragraph (a)(1)(ii) of this section and the agency exempts you from such testing.

Note what that paragraph does not allow. For every other kind of component, the next paragraph of the same section lets a manufacturer skip its own testing and rely on a certificate of analysis from the supplier, provided it first qualifies that supplier by confirming their results. For a dietary ingredient, that shortcut does not exist. Identity testing is on you, on every lot, unless you have petitioned the FDA and been granted a specific exemption, and that petition has to demonstrate "no material diminution of assurance, compared to the assurance provided by 100 percent identity testing."

The CDC authors put the same point in their opening line, and it reads like an understatement on purpose:

Regulations mandate that all ingredients used to manufacture dietary supplements be tested for identity and be free from reasonably anticipated contaminants. Despite these regulations, misbranded dietary supplements are frequently found to contain potentially dangerous substances.

So the honest framing of this story is not that a gap in the rules let a poison through. The rule is there, it is specific, and it is the strictest identity requirement in the whole section. What failed is enforcement reaching the sellers, most of whom are shipping through marketplace listings from outside any inspected supply chain. Look down the FDA's own table at the recall-status column: recall announcement, recall announcement, declined to recall, firm has not committed to a recall, FDA unable to contact firm, the selling firm has not responded to calls or emails from FDA. That column is the actual story.

Not medical advice. This is educational analysis, not a recommendation — a study is not a prescription. Talk to a qualified clinician before acting on anything you read here. Full disclaimer →

What an identity test actually is

"Identity testing" sounds like one thing. It is several, they do not have the same power, and the difference is exactly what this case turns on.

The weakest tier is organoleptic and macroscopic examination: someone looks at the material, smells it, checks that the pieces are the right colour and texture against a written description. For a whole, distinctive botanical this is not worthless. For brown chopped root in a sack it is close to it.

The middle tier is chromatographic fingerprinting, usually high-performance thin-layer chromatography, where the extract's band pattern is compared against a reference. This works when you have an authenticated reference of the right species, and it can fail quietly when you do not, because a fingerprint only tells you the sample does or does not match the thing you compared it to.

The two tiers that actually settled this case are DNA barcoding and mass spectrometry, and in 2021 a group led by Shin at Korea's Ministry of Food and Drug Safety ran both on this exact question, two years before the CDC report. Their conclusion:

Our results demonstrate that Alipotec samples contained ingredients from different species, yellow oleander instead of tejocote. The rpoB barcode region was able to differentiate between tejocote and yellow oleander species. Moreover, it was also observed that three compounds, including thevetin B, neriifolin, and digitoxigenin, clearly distinguish between tejocote and yellow oleander samples.

That is a complete, published, freely citable method. One plastid gene region (rpoB) separates the species genetically. Three cardenolide markers separate them chemically. This was in the peer-reviewed literature in 2021. The FDA's list kept growing through 2024, 2025 and 2026.

There is one real caveat on DNA methods that deserves stating, because I am not going to pretend barcoding is magic: heavily processed and extracted material can be degraded past the point where amplifiable DNA survives, which is why the serious labs run orthogonal methods rather than one. In this case the chopped-root and capsule formats are about as favourable for barcoding as botanical material gets, and the chemistry agreed with the genetics anyway.

The clinical detail worth carrying

One finding from the CDC report is worth knowing even if you never touch a weight-loss supplement, because it is the kind of thing that decides an emergency-room hour.

A person poisoned by yellow oleander can show a positive serum digoxin result on an immunoassay, through cross-reactivity, and that number does not tell you how much cardenolide is on board. The authors state it directly: serum digoxin assays are not reliable for detection of thevetin B and cardiac glycosides other than digoxin, and laboratory-reported digoxin levels do not accurately reflect serum levels of other cardiac glycosides. They also note, citing the yellow-oleander poisoning literature, that higher doses of antibody fragments may be needed than in ordinary digoxin toxicity.

This is the same failure mode this beat keeps running into from the other direction. An instrument returns a number, the number has a name attached, and the name is not the quantity you needed. I wrote about the laboratory version of that in Report 153. The clinical version has a shorter fuse.

What I could not confirm

I did not test anything. No sample was purchased or analysed for this report. Everything here comes from the FDA's alert page, the CDC report and the published methods literature.

The Shin et al. paper was read as its complete author abstract, not in full. It is paywalled at the publisher. The abstract is the authors' own words and the quoted passage is verbatim from it, but I did not read their methods or their chromatograms, and I am relying on the CDC authors' citation of that work for the characterisation of the Alipotec result.

I cannot tell you the current composition of any specific product. The FDA's table records what specific sampled lots contained when they were tested, some of them years ago. A brand on that list may have changed suppliers; a brand not on the list has not been cleared, it has generally just not been tested. The FDA is explicit that it is continuing to sample and will keep adding.

I did not independently verify the product count. My count of 28 rows is my own reading of the page on 23 September 2026. The figure of 33 products comes from trade coverage of the same update and I have not reconstructed how it was reached.

I have not audited these firms' testing records. I am describing what 21 CFR 111.75 requires and what the analytical results show. I have no visibility into whether any particular firm on that list performed identity testing, performed it badly, or performed none, and the FDA's warning letters, not this report, are the record of that.

The signal

Three things to take from this, in order of how much they will actually protect you.

First, the specific one: if you have any botanical weight-loss product in the house that is sold as tejocote root, Raiz de Tejocote, Mexican hawthorn, Brazil seed or Semilla de Brasil, check it against the FDA's list, and understand that not being on the list is not the same as having been tested. If anyone has taken one of these, the FDA's advice is to contact a health care provider even if it was a while ago, and that advice exists because cardenolide effects are cardiac.

Second, the general one: species substitution is the failure mode that consumer-facing testing talk almost never covers. The conversation about supplement quality is usually about potency, whether the bottle has as much of the ingredient as it claims. Potency testing assumes identity. If you assay the wrong plant for the right compound, you get a clean-looking number for a material that should never have entered the building. Identity comes first, and it is the one test the federal rule refuses to let a manufacturer outsource to a supplier's paperwork.

Third, the one for anyone who makes this stuff: the question to ask a contract manufacturer is not "do you do identity testing." Everyone says yes. The question is which method, against which authenticated reference, on which lots. Organoleptic examination against a written description is a legal answer to that question. It would not have caught a single one of these ten products.

Disclosure, plainly: I founded and run Shroombiosis (a company I run), which formulates and sells functional-mushroom supplements. That gives me a direct commercial interest in how this industry is perceived and regulated, and it means a report arguing for stricter identity testing is a report arguing for a standard I also have to meet. Weigh it accordingly. Nothing here is sponsored, no link earns a commission, I have no relationship with any firm named on the FDA's list or with any of the researchers cited, and none of the products discussed compete with anything I sell; here's the full policy. A recommendation with no stake at all: for performance nutrition, Die Tryin Co. is a fellow combat-veteran-owned brand I recommend. I don't own it and earn nothing from the link.