Galantamine is a single, well-characterized molecule. It is sold two ways: as a generic prescription drug for Alzheimer’s disease, and as a dietary supplement marketed for memory. Same chemical, two aisles, two rulebooks.
In 2024, a team from Cambridge Health Alliance in Massachusetts and Sciensano, the Belgian public health institute, published in JAMA what happens when you measure both. Across 11 generic galantamine drug products, the measured content ran from 97.5 percent to 104.2 percent of what the label declared. All 11 landed within 10 percent of their label. Across 10 galantamine supplements, the measured content ran from less than 2 percent to 110 percent of label. One of the 10 was within 10 percent.
Three of the 10 supplements, 30 percent, were contaminated with Bacillus cereus carrying enterotoxin genes associated with diarrheal illness. Those three contained 60, 62 and 75 percent of their labelled galantamine.
That one study does something no other study in this field manages. It shows a potency failure and a purity failure in the same products, with a control group made of the identical molecule under a different regulatory regime. The variable is not chemistry. It is which rulebook applies.
Potency and purity are two different failures
Potency is about presence and amount: is the declared ingredient there, and is it there in the declared quantity? Purity is about absence: is something in the bottle that the label never mentioned?
They fail for different reasons and they are caught by different tests. A product can pass one and fail the other. The galantamine supplements failed both at once, which is unusual enough to be worth noticing.
A companion article in this library covers contamination as a subject in its own right, including undeclared pharmaceuticals, heavy metals and microbes. This one stays with the measurement itself: what laboratories find when they weigh what is in a product against what the label says, and what the rules permit that result to be.
What the surveys measured
The most-cited recent potency survey appeared in JAMA Network Open in July 2023. Researchers bought 63 sports supplements online whose labels declared one of five botanical ingredients with claimed stimulant or anabolic effects. Six were excluded because the physical label did not list one of the five. The remaining 57 were reconstituted in methanol and run through liquid chromatography mass spectrometry.
Twenty-three of the 57, 40 percent, contained no detectable amount of the ingredient their label declared. Among those that contained any, the measured quantity ran from 0.02 percent to 334 percent of the labelled amount. Six of the 57, 11 percent, came within 10 percent of their label. Seven products, 12 percent, contained at least one ingredient the Food and Drug Administration prohibits in supplements; one product contained four of them.
The authors’ summary line is blunt: 89 percent of the labels did not accurately declare the ingredients found in the products.
The published table is more instructive than the summary, because it shows the spread inside a single ingredient. Twenty-one products declared the caffeine-like compound methylliberine. One labelled 40 mg per serving measured 134 mg, or 334 percent of its claim. Another labelled 100 mg measured 61 mg. A third labelled 100 mg measured exactly 100 mg. Same ingredient, same market, three different relationships between the printed number and the powder.
The purity side of the same dataset is narrower and sharper. Five different prohibited compounds turned up across the 57 products: three once sold as medicines in Europe, one marketed in Russia and never approved in the United States, and one never approved as a medication anywhere. None appeared on a label. A laboratory had to go looking for them.
The same paper supplies its own historical comparison, which is more interesting than the headline. A study conducted before FDA banned ephedra found 6 of 12 products, 50 percent, within 10 percent of their labelled ephedra. A later study of caffeine content found 9 of 20, 45 percent, within 10 percent. The accuracy of the newest fringe ingredients is markedly worse than the accuracy of the older mainstream ones.
Now the limits, which the authors state themselves. The sample was small. Only one sample of each brand was analysed, so nothing is known about variation between batches of the same product. Only five targeted ingredients were examined. And the sampling frame matters: these were products whose own labels claimed novel compounds, a group you would expect to be unrepresentative.
That last point generalizes. No study in the published literature estimates the label-accuracy failure rate of the US supplement market as a whole. Every figure you will read, including the ones above, describes a category, a channel or an ingredient list that somebody chose. A reader who takes 89 percent as a statement about supplements in general is reading a number about 57 sports products.
More than the label says is usually on purpose
Here is the part that reframes the subject. When a test finds more of an ingredient than the label declares, the most likely explanation is not sloppiness. It is design, and it is lawful.
The regulation is 21 CFR 101.36(f)(1), which governs how compliance with supplement labelling is determined. It contains this sentence: “Reasonable excesses over labeled amounts are acceptable within current good manufacturing practice.”
Manufacturers add extra of an ingredient that degrades, so that the product still meets its claim at the end of its shelf life. The industry calls this overage. It is routine, it is written into release specifications, and it has been measured.
The measurement that counts came from the Agricultural Research Service of the US Department of Agriculture working with the National Institutes of Health Office of Dietary Supplements, published in the American Journal of Clinical Nutrition in 2017. Adult multivitamin and mineral products were bought under a national sampling plan and analysed in qualified laboratories against certified reference materials. Thiamin was the single exception. For every other ingredient, average measured content landed above what the label promised.
The size of the excess varied by nutrient. Copper, manganese, magnesium, niacin, phosphorus, potassium, folic acid, riboflavin and vitamins B-12, C and E ran 1.5 to 13 percent above label. Selenium and iodine ran about 25 percent above, regardless of how much the label claimed. Calcium ran from 7.1 to 29.3 percent above, depending on the labelled amount.
The authors also noted that for 12 of the 18 nutrients, most products already carried labelled amounts at or above the Recommended Dietary Allowance. Their conclusion was that nutrient overexposure from supplements combined with ordinary food intake may have unintended consequences, a question they said needed further examination. Two other articles in this library deal with combining products and with the documented harms of excess.
The regulatory asymmetry explains why overage is one-directional. FDA’s own Dietary Supplement Labeling Guide, citing 21 CFR 101.9(g)(3) and (g)(4), says a product must contain 100 percent of the volume or weight declared on the label for ingredients the manufacturer added, with an allowance for deviation attributable to the analytical method. For dietary ingredients that occur naturally in the raw material, the threshold is 80 percent of the declared value.
Notice what is missing. There is a floor and no ceiling. A manufacturer’s only one-sided legal risk is being under. A product measured at 334 percent of its label is not obviously out of compliance in the way a product at 74 percent is.
Whether plus or minus 10 percent is the right test
The studies above judge accuracy against a window of plus or minus 10 percent of label. Where does that window come from, and is it the right one?
It comes from pharmaceuticals. It is roughly the band generic drugs are expected to sit inside, and the galantamine generics show what that expectation produces in practice: all 11 inside it, clustered between 97.5 and 104.2 percent. Researchers applying it to supplements are applying the standard used for medicines, deliberately, to products sold for similar reasons.
The industry disputes the choice. When the gummy study by the same research group was published in 2023, the Council for Responsible Nutrition, the supplement industry’s main trade association, responded that the research conflated the overage standards used for pharmaceutical products with how overages are handled for dietary supplements, and that the authors had used FDA’s drug standard for overages and shortages. The association’s position is that supplement rules require a product to contain at least its labelled amounts through shelf life, that extra is therefore added to compensate for degradation, and that doing so within good manufacturing practice is permitted.
Both positions are accurately stated and neither is a factual error. One is a clinician-researcher applying the yardstick used for prescription medicines to products containing molecules that are prescription medicines elsewhere. The other is a trade association representing the companies that set the release specifications, arguing that the legal standard those companies are held to is a floor, not a band.
What a reader can take from the disagreement is narrower and more useful than either side’s conclusion. A potency test result is uninterpretable without knowing whether an overage was intended and how large it was meant to be. That information sits in a manufacturing record that is not public.
What the testing rule obliges a manufacturer to do
The federal rule most people assume covers product testing is 21 CFR 111.75. It is worth reading precisely, because the obligations are narrower than the reputation.
The incoming-ingredient obligation reads that a manufacturer must “Conduct at least one appropriate test or examination to verify the identity of any component that is a dietary ingredient.” Read each part of that sentence separately, because each part narrows it. At least one test. Identity only, which says nothing about purity or strength. And components that are dietary ingredients, which leaves out everything else in the capsule.
For other components, the rule offers a choice: run appropriate tests, or rely on a certificate of analysis from the supplier. The certificate route comes with conditions. The manufacturer must first qualify the supplier by confirming the reliability of its certificate through its own testing, keep documentation of how it did so, and periodically re-confirm it. The rule does not define how often that is.
For finished products, the rule reaches only a subset of batches, chosen by the firm under a statistical sampling plan the firm designs itself. Testing every batch is permitted. Testing a subset is also permitted. And the specifications those batches are checked against are the firm’s own. The companion article on manufacturing standards covers where those specifications come from, which is the single most-cited failure FDA finds.
Paragraph (d) permits a manufacturer to exempt a product specification from finished-batch verification where no scientifically valid method exists to test it at that stage, provided the firm documents why other controls will ensure the specification is met and quality control approves the reasoning.
Then there is paragraph (h), which lists the acceptable kinds of method. A test or examination must be an appropriate, scientifically valid method, and must include at least one of five types. The first type on the list is gross organoleptic analysis. Looking at the material and smelling it satisfies the rule.
That is not a loophole somebody found. It is the text, and for some checks it is a sensible one. It is also a long way from what most people picture when a bottle says the contents were tested.
Nothing requires a supplement to still be what the label says
Here is the gap that surprises people most. There is no federal requirement that a dietary supplement carry an expiration date, and no federal requirement for stability testing at all.
Part 111 runs to sixteen subparts covering personnel, plant, equipment, production controls, quality control, components, master manufacturing records, batch records, laboratory operations, manufacturing, packaging, holding and distribution, returns, complaints and recordkeeping. None of them requires a firm to establish how long its product holds its potency. A summary of the differences between the supplement and drug manufacturing rules, prepared by a regulatory consultancy and hosted by the Consumer Healthcare Products Association, puts it directly: there are no requirements for expiration dating, shelf life or best-if-used-by dating, and none for the stability testing that would support them. Drug manufacturers, by contrast, must establish expiration dates backed by a formal stability programme.
A date printed voluntarily is a different matter. Once a firm prints one, it has made a factual claim, and it needs its own data to support it. Those data are the firm’s property and are not published. Whether an inspector ever looks at them is a separate question, covered in the companion article on inspections.
The one product class where degradation has been measured over time is probiotics, because viable organisms can be counted. A 12-month study published in Frontiers in Nutrition in September 2026 stored commercial multi-strain probiotic supplements at ambient, refrigerated and frozen temperatures and enumerated viable counts. Ambient storage showed a trend toward the greatest losses. Compliance with label-declared counts declined toward the end of the stated minimum durability, and some products failed to meet declared values under every storage condition, which the authors read as instability built into the formulation rather than caused by storage. They also found inconsistent naming and limited strain-level labelling across all the products tested. That work was done at a Czech university on products sold in Europe, which is a real limit on applying it here.
For vitamins and minerals, the equivalent US study does not appear to exist. A search of the published literature found no American, peer-reviewed, longitudinal study measuring a supplement’s content at manufacture against its content at the printed expiry. That is a genuine gap, not a finding that nothing degrades.
How to read a test result you find online
Numbers about supplement content circulate widely and most of them arrive stripped of the context that makes them mean anything. Five questions put it back.
- Measured against what? A result compared to the label is a different claim from a result compared to a published standard or a studied dose. Most are compared to the label.
- Who bought the sample, and where? Products supplied by a manufacturer, products bought at retail and products bought online are three different samples. Some testing programmes exclude their own paying clients from their published failure rates.
- How many samples per brand? Nearly all of this research tests one sample of each product. Batch-to-batch variation inside a brand is largely unstudied.
- Was an excess intended? Above label is the normal result for a multivitamin. Below label is the result the rule forbids.
- Who paid, and who ran the lab? Government agencies, academic researchers, trade associations and companies that sell certification have all produced numbers in this field, and the funding is usually stated if you look.
What the measurement record supports
Strip the subject to its structure and it comes to this. The number on a supplement label is a target the manufacturer set. The law requires the product to reach that target and says nothing about how far past it the product may go. Verification is against the firm’s own specification, by a method that may be as simple as looking and smelling, on a statistically chosen subset of batches, with one identity test per dietary ingredient. Nothing requires the number to still be true in a year, and nothing requires a date to be printed at all.
Most of the time that system produces a multivitamin with slightly more in it than the label promises. Sometimes, in the corners of the market where novel compounds are sold, it produces a bottle with none of the declared ingredient and a prohibited drug instead.
If the amount of something matters clinically in your case, which is true for nutrients a clinician asked you to take, and true for anything taken alongside a medicine with a narrow safety margin such as warfarin, levothyroxine or lithium, then the accuracy of a label is a question for the person managing that treatment and not something a bottle can answer. Bring the actual container to the appointment. The label is the only record of what the product claims to be, and as this research shows, it is a claim rather than a measurement.
Sources
- Cornell Law School, Legal Information Institute. “21 CFR 101.36: Nutrition labeling of dietary supplements (compliance provisions at (f)).” law.cornell.edu
- Cornell Law School, Legal Information Institute. “21 CFR 111.75: What must you do to determine whether specifications are met?.” law.cornell.edu
- Cornell Law School, Legal Information Institute. “21 CFR Part 111: Current Good Manufacturing Practice in Manufacturing, Packaging, Labeling, or Holding Operations for Dietary Supplements (subpart list).” law.cornell.edu
- U.S. Food and Drug Administration. “Dietary Supplement Labeling Guide: Chapter IV, Nutrition Labeling (compliance questions, citing 21 CFR 101.9(g)(3) and (g)(4)).” fda.gov
- JAMA (PubMed). “Cohen PA, Jacobs B, Van Hoorde K, Vanhee C. "Accuracy of Labeling of Galantamine Generic Drugs and Dietary Supplements." 2024;331(11):974-976. DOI: 10.1001/jama.2024.0328.” doi.org
- JAMA Network Open (PubMed). “Cohen PA, Avula B, Katragunta K, Travis JC, Khan I. "Presence and Quantity of Botanical Ingredients With Purported Performance-Enhancing Properties in Sports Supplements." 2023;6(7):e2323879. DOI: 10.1001/jamanetworkopen.2023.23879.” doi.org
- American Journal of Clinical Nutrition (PubMed). “Andrews KW, Roseland JM, Gusev PA, et al. "Analytical ingredient content and variability of adult multivitamin/mineral products: national estimates for the Dietary Supplement Ingredient Database." 2017;105(2):526-539. USDA Agricultural Research Service with the NIH Office of Dietary Supplements. DOI: 10.3945/ajcn.116.134544.” doi.org
- Frontiers in Nutrition (PubMed). “Brousilova T, Vlkova E, Neuzil-Bunesova V, Musilova S, Kadlec-Modrackova N. "Storage temperature as a determinant of probiotic long-term stability in commercial dietary supplements." 2026;13:1933900. DOI: 10.3389/fnut.2026.1933900.” doi.org
- Sciensano (author institution copy of the JAMA research letter). “Accuracy of Labeling of Galantamine Generic Drugs and Dietary Supplements (full text with results tables and disclosures).” sciensano.be
- Consumer Healthcare Products Association (hosting a presentation by EAS Consulting Group). “Summary of Key Differences Between FDA Drug and Dietary Supplement GMPs.” CHPA.org
- Nutraceuticals World. “Ingredient Overages Widespread in Melatonin Gummies, Study Claims (25 April 2023), carrying the Council for Responsible Nutrition's response.” nutraceuticalsworld.com