Finnish and American researchers enrolled 29,133 male smokers aged 50 to 69 in a randomized, double-blind, placebo-controlled trial. They expected to prevent lung cancer. Diets rich in carotenoids had tracked with lower rates for years, and beta-carotene was the obvious candidate.
The men who got beta-carotene developed lung cancer 18 percent more often than the men who did not, with a confidence interval running from 3 to 36 percent. Their total mortality was 8 percent higher. The trial, published in 1994 in the New England Journal of Medicine, is still one of the best-run nutrition experiments ever conducted.
That pattern is why this subject rewards careful reading. The large supplement trials are unusually good: randomized, blinded, enormous, run for years. And they have repeatedly found the opposite of what everyone expected.
What follows is not a list of good supplements. It is a method for reading any claim about any of them, built from the trials already run.
The task force conclusion almost nobody has read
The US Preventive Services Task Force is the federal advisory body whose recommendations shape what clinicians screen for and what insurers cover. On June 21, 2022, it issued a final recommendation statement on vitamin and mineral supplements for preventing cardiovascular disease and cancer in community-dwelling, nonpregnant adults.
It reached three conclusions. The first carries a grade of D: “The USPSTF recommends against the use of beta carotene or vitamin E supplements for the prevention of cardiovascular disease or cancer.”
The second and third carry a grade of I, the Task Force’s label for insufficient evidence. For multivitamins, it concluded that current evidence is insufficient to assess the balance of benefits and harms for preventing cardiovascular disease or cancer, and it said the same of single or paired nutrient supplements other than beta-carotene and vitamin E.
The three findings are different, and the difference matters. One is a recommendation against. Two are statements that the question has not been answered, which is neither “supplements don’t work” nor “supplements work.”
Four large trials that found harm
Harm here is not theoretical and not a matter of contamination. These four trials used clean products at known doses, and the harm came from the nutrient itself.
Beta-carotene in smokers
The Finnish trial above, known as ATBC, tested alpha-tocopherol at 50 mg a day and beta-carotene at 20 mg a day in a four-arm design, with five to eight years of follow-up.
A second trial, known as CARET and run on a different population, produced the same verdict. Published in 1996, it enrolled 18,314 smokers, former smokers and workers exposed to asbestos, and gave them 30 mg of beta-carotene plus 25,000 IU of retinol daily or a placebo. After a mean of four years, the treated group had a relative risk of lung cancer of 1.28 (95 percent CI 1.04 to 1.57), of death from any cause of 1.17, and of death from lung cancer of 1.46. The investigators stopped the randomized trial 21 months early.
Two trials, two countries, two doses, the same direction of harm, in people who smoked.
Vitamin E and prostate cancer
SELECT randomized 35,533 men at 427 sites in the United States, Canada and Puerto Rico between August 2001 and June 2004. The doses were 200 micrograms a day of selenium as L-selenomethionine and 400 IU a day of vitamin E as all-rac-alpha-tocopheryl acetate.
The primary report, published in 2008 at a median 5.46 years of follow-up, found nothing. Prostate cancer hazard ratios were 1.13 for vitamin E and 1.04 for selenium, neither statistically significant. The vitamin E signal sat at P = .06.
Three years later the same investigators published again, with 54,464 additional person-years and 521 additional cancers. Now 620 men in the vitamin E group had developed prostate cancer against 529 on placebo: hazard ratio 1.17, 99 percent CI 1.004 to 1.36, P = .008. The harm did not appear in the primary report. It appeared only when the men were followed longer, and anyone citing SELECT has to cite both papers or they are describing a different trial.
The absolute figure keeps it in proportion: 1.6 extra prostate cancers per 1,000 person-years.
Selenium and diabetes
The same mineral produced a different signal elsewhere. A 2007 analysis in Annals of Internal Medicine followed 1,202 dermatology-clinic patients without diabetes at baseline, recruited in low-selenium areas of the eastern United States, who took 200 micrograms a day of selenium or placebo for an average of 7.7 years.
Type 2 diabetes developed in 58 selenium recipients and 39 placebo recipients, 12.6 against 8.4 cases per 1,000 person-years, hazard ratio 1.55 (95 percent CI 1.03 to 2.33). Among participants in the highest third of baseline blood selenium, the hazard ratio was 2.70. The authors named their own limitation: diabetes was a secondary outcome in a trial designed to study cancer. That makes it a signal, not a settled finding.
High-dose folic acid and the colon
Between 1994 and 2004, researchers at nine centers randomized 1,021 adults with a recent history of colorectal adenomas to 1 mg a day of folic acid or placebo. At the first surveillance colonoscopy cycle, adenoma rates were essentially identical, 44.1 against 42.4 percent. At the second cycle, advanced lesions appeared in 11.6 percent of the folic acid group against 6.9 percent on placebo, a risk ratio of 1.67 with a 95 percent confidence interval of 1.00 to 2.80 and P = .05.
The lower bound touches exactly 1.00, which makes this the weakest of the four results, and the authors said so, concluding that further research was needed into whether folic acid might raise the risk of colorectal neoplasia.
Note the dose. One milligram a day is more than twice the amount in the folate recommendation described below, in a population selected for existing colon lesions. Same nutrient, different amount, different people, opposite conclusion.
Where supplements clearly work, and exactly for whom
The population is not a detail attached to the claim. The population is the claim.
Folic acid before and during early pregnancy. In 1991, a Medical Research Council trial at 33 centers in seven countries randomized 1,817 women who were at high risk of a neural tube defect because of a previous affected pregnancy. Among 1,195 completed pregnancies with a known outcome, there were 27 neural tube defects, 6 in the folic acid groups and 21 in the others: a 72 percent protective effect, relative risk 0.28 (95 percent CI 0.12 to 0.71). A mixture of seven other vitamins showed no significant effect.
A Hungarian trial published in 1992 answered the other half, in women planning a pregnancy, in most cases their first. Among 4,156 known outcomes there were six neural tube defects in the trace-element comparison group and none in the group taking a supplement with 0.8 mg of folic acid (P = 0.029), with overall congenital malformations at 22.9 against 13.3 per 1,000. The same trial found no reduction in cleft lip.
On August 1, 2023, the Task Force gave folic acid supplementation a grade of A, its strongest, for anyone planning or capable of pregnancy, at 0.4 to 0.8 mg daily. That is the only grade A in this subject.
Vitamin D and calcium, in one specific group. A 1992 French trial gave 3,270 ambulatory women with a mean age of 84 either 1.2 grams of elemental calcium plus 800 IU of vitamin D3 daily, or a double placebo, for 18 months. Hip fractures were 43 percent lower and nonvertebral fractures 32 percent lower in the treated group.
Now set that beside the Task Force. In April 2018, USPSTF gave a grade of D to daily supplementation with 400 IU or less of vitamin D and 1,000 mg or less of calcium for primary prevention of fractures in community-dwelling postmenopausal women, and a grade of I for men, premenopausal women, and higher doses. The Task Force notes the topic is currently being updated.
Those findings look contradictory and are not. One studied women averaging 84 years old at 800 IU and 1.2 grams; the other evaluated community-dwelling postmenopausal women at 400 IU or less. Different people, different amounts, different answers. Article 327 covers vitamin D as a nutrient.
Iron, and the precision the word requires. “Iron works” is not a true sentence. “Iron corrects documented iron deficiency” is.
On August 20, 2024, the Task Force concluded that current evidence is insufficient to assess the balance of benefits and harms of either screening for iron deficiency or routine iron supplementation in asymptomatic pregnant adolescents and adults, grading both I. The scope language is the useful part: the statement explicitly does not apply to pregnant people who are severely malnourished, who have symptoms of iron deficiency or iron deficiency anemia, or who have certain hematologic conditions. A conclusion about routine supplementation in people without symptoms says nothing about treating a documented deficiency. The two get collapsed constantly.
Persistent fatigue with breathlessness, new numbness or tingling in the hands or feet, and unusual bruising all overlap a dozen conditions, and the test that tells them apart is ordered by a clinician, not purchased. And if you currently smoke or recently stopped, ATBC and CARET were run in exactly your population, which belongs in a conversation with whoever manages your care. Interactions with prescription medicines are covered in article 677.
What FDA does before a supplement reaches a shelf
Dietary supplements sit under the umbrella of foods in US law, not under the drug approval system. FDA states the consequence plainly on its own question-and-answer page for consumers: “Under DSHEA, FDA does not have the authority to approve dietary supplements before they are marketed.”
Elsewhere FDA puts it shorter still: the agency does not approve dietary supplements for any purpose, and companies are responsible for making sure their products are safe and accurately labeled.
One partial exception gets misread often. A genuinely new ingredient, one not marketed in the US before October 1994, generally requires the manufacturer to notify FDA before selling it. FDA’s draft guidance on new dietary ingredient notifications, dated October 2026, supersedes the agency’s August 2016 revised draft and its March and April 2024 revisions. Inside that 2026 document sit two older numbers: as of December 2014, FDA had received and completed evaluation of just over 750 such notifications since the first arrived in 1995, and in 2012 the agency estimated 55,600 dietary supplements on the market, against roughly 4,000 in 1994.
Those are 2012 and 2014 figures published in a 2026 guidance, and they carry their own dates. The arithmetic people perform on them, dividing 750 into 55,600 to produce a non-compliance rate, is an inference drawn by readers, not a finding by FDA. The two numbers are not comparable, because most products on the market contain only ingredients sold before 1994, which need no notification at all.
Three current examples show how unsettled the boundary is, and each is a different kind of case.
NMN. Before September 29, 2025, FDA’s position was that nicotinamide mononucleotide was excluded from the dietary supplement definition because it had been investigated as a drug first. In letters dated September 29, 2025, responding to two citizen petitions, the agency reversed that, concluding NMN is not excluded while continuing to treat it as a new dietary ingredient. Not excluded is not the same as cleared, and every sentence about NMN needs that date attached.
NAC. FDA says N-acetyl-L-cysteine is excluded from the supplement definition, because it was approved as a new drug before being marketed as a supplement or food. In final guidance issued August 1, 2022, the agency said it intends to exercise enforcement discretion for certain NAC products labeled as supplements. They are on shelves because FDA has chosen not to act, not because they qualify.
Tianeptine. FDA’s consumer warning, current as of May 8, 2025, says tianeptine does not meet the statutory definition of a dietary ingredient and is not approved for any medical use. No discretion, and active warnings. These three cases get lumped together in coverage, and they are not alike.
A seal on the bottle is not a government approval
Sort every standard, seal and claim in this field into four buckets, and the distribution tells you most of what you need.
Defined in regulation: the 1994 statute, the federal manufacturing rules for supplements, the new dietary ingredient notification requirement, and the legal duty to report serious adverse events.
Run as a government certification program: nothing. There is no federal certification program for US dietary supplements, and that empty bucket is itself the finding.
Private third-party certification: USP Verified and NSF Certified for Sport. USP’s consumer page says its mark indicates a product contains the ingredients listed on the label at the declared potency and amounts, holds no harmful levels of specified contaminants, breaks down and releases within a specified time, and was made according to FDA current good manufacturing practices. NSF’s Certified for Sport page lists four points: absence of 290 banned substances, contents matching the printed label, no unsafe contaminant levels, and manufacture at a GMP-certified facility audited annually or bi-annually. A different NSF page gives the banned-substance count as 280. Both figures are NSF’s, and the discrepancy is theirs to reconcile.
Undefined: “FDA approved” on a supplement, which does not exist.
Notice what none of those points claims. Not one says the product does anything. They certify identity, content, contaminants and manufacturing. The NIH Office of Dietary Supplements, in a consumer fact sheet last updated January 4, 2023, says directly that these seals do not guarantee a product is safe or effective.
They are also optional. Nothing requires a company to apply, so a product without a mark has not failed a test. It may simply never have taken one.
How harm gets counted after the fact
With no premarket safety review, surveillance is what remains, and its limits matter before you trust a number from it.
Federal law requires the responsible person, meaning the manufacturer, packer or distributor named on the label, to report serious adverse events to the Department of Health and Human Services within 15 business days of receiving them. The statute defines serious as death, a life-threatening experience, inpatient hospitalization, persistent or significant disability, a congenital anomaly or birth defect, or an event requiring intervention to prevent one of those.
Everything else is voluntary. FDA’s own description of its adverse event reporting system says reporting is voluntary for beverages, conventional foods, dietary supplements and cosmetics, that generally only a small fraction of adverse events is reported, that a causal relationship cannot be established between a product and the reactions in a report, and that the data cannot be used to estimate how often these events occur.
How small is the fraction? FDA scientists examined 15,430 supplement adverse event reports from 2004 through 2013 and estimated the reporting rate at roughly 2 percent.
For an actual population estimate, researchers at CDC and FDA used surveillance data from 63 emergency departments covering the same decade. Their annual figures: 23,005 emergency department visits traced to supplement-related adverse events, and 2,154 of those visits ending in a hospital admission. Excluding unsupervised ingestions by children, 65.9 percent of single-supplement visits involved herbal or complementary products and 31.8 percent involved vitamins and minerals. Weight-loss and energy products together caused 71.8 percent of visits for palpitations, chest pain or racing heartbeat.
Those categories line up with the contamination data. A 2018 analysis of FDA’s tainted-products database found 776 adulterated supplements from 146 companies between 2007 and 2016, concentrated in sexual improvement (353), weight loss (317) and muscle building (92). Sibutramine, a withdrawn weight-loss drug, turned up in 84.9 percent of the adulterated weight-loss products. Adulteration is covered in article 315.
Six questions that replace “does it work”
Put any supplement claim through these. They are the questions the trials above answer, in the same order.
- In whom? Smokers, postmenopausal women, 84-year-olds, people with a documented deficiency? Change the population and the answer changes.
- For what outcome? A blood level, a symptom score, or a disease event? The first is cheap to move and the last is what matters.
- At what dose? Folic acid at 0.4 mg before pregnancy and 1 mg a day in people with colon adenomas are two different experiments.
- Compared with what? Placebo, a different form, or nothing at all?
- Followed for how long? SELECT’s harm was invisible at a median 5.5 years and clear once the men were followed longer.
- Funded by whom? A 2017 Cochrane review of 75 studies found industry-sponsored research reported favorable results more often than independent research, a risk ratio of 1.27, and reached favorable conclusions more often still, a risk ratio of 1.34. That review covered drugs and devices; no equivalent analysis built from supplement trials could be located, which is a real gap.
Where this evidence lands
The strongest evidence in this entire subject is narrow and specific: folic acid, before conception, for neural tube defects, graded A. Almost everything else sits in one of two places. Either the question has not been answered, which is what a grade of I means and what the Task Force concluded about multivitamins. Or the answer turned out to be the opposite of the expectation, which is what happened to beta-carotene in smokers and vitamin E in healthy men.
There is a reason the good news is so narrow. Trials this large are rare, expensive and slow, and when they finally run they tend to dismantle the observational associations that justified them. The few settled answers all come with a named population attached.
The question is never whether a supplement works. It is: in whom, for what outcome, at what dose, over how long, measured how, and paid for by whom.
Sources
- US Preventive Services Task Force. “Vitamin, Mineral, and Multivitamin Supplementation to Prevent Cardiovascular Disease and Cancer: Preventive Medication (final recommendation, June 21, 2022).” uspreventiveservicestaskforce.org
- US Preventive Services Task Force. “Folic Acid Supplementation to Prevent Neural Tube Defects: Final Recommendation Bulletin (August 1, 2023).” uspreventiveservicestaskforce.org
- US Preventive Services Task Force. “Vitamin D, Calcium, or Combined Supplementation for the Primary Prevention of Fractures in Adults: Preventive Medication (final recommendation, April 17, 2018).” uspreventiveservicestaskforce.org
- US Preventive Services Task Force. “Iron Deficiency Anemia in Pregnant Women: Screening and Supplementation (final recommendation, August 20, 2024).” uspreventiveservicestaskforce.org
- US Food and Drug Administration. “Questions and Answers on Dietary Supplements.” fda.gov
- US Food and Drug Administration. “Facts About Dietary Supplements.”
- US Food and Drug Administration. “Dietary Supplements: New Dietary Ingredient Notifications and Related Issues: Guidance for Industry (draft guidance, October 2026).” fda.gov
- US Food and Drug Administration. “FDA Releases Final Guidance on Enforcement Discretion for Certain NAC Products.” fda.gov
- US Food and Drug Administration. “FDA Warns Consumers Not to Purchase or Use Any Tianeptine Product Due to Serious Risks.” fda.gov
- US Food and Drug Administration. “Response to Citizen Petition, Docket FDA-2023-P-1867 (September 29, 2025).” downloads.regulations.gov
- US Food and Drug Administration (openFDA). “Food Adverse Events (CFSAN Adverse Event Reporting System) API Reference.” open.fda.gov
- Cornell Law School Legal Information Institute. “21 U.S. Code Section 379aa-1 – Serious adverse event reporting for dietary supplements.” law.cornell.edu
- NIH Office of Dietary Supplements. “Dietary Supplements: What You Need to Know.” ods.od.nih.gov
- US Pharmacopeia. “What the USP Verified Mark Means.” usp.org
- NSF Certified for Sport. “Our Mark.” nsfsport.com
- PubMed. “The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. "The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers." New England Journal of Medicine, 1994. PubMed. DOI: 10.1056/NEJM199404143301501.” doi.org
- PubMed. “Omenn GS et al. "Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease." New England Journal of Medicine, 1996. PubMed. DOI: 10.1056/NEJM199605023341802.” doi.org
- PubMed. “Lippman SM et al. "Effect of selenium and vitamin E on risk of prostate cancer and other cancers: the Selenium and Vitamin E Cancer Prevention Trial (SELECT)." JAMA, 2008. PubMed. DOI: 10.1001/jama.2008.864.” doi.org
- PubMed. “Klein EA et al. "Vitamin E and the risk of prostate cancer: the Selenium and Vitamin E Cancer Prevention Trial (SELECT)." JAMA, 2011. PubMed. DOI: 10.1001/jama.2011.1437.” doi.org
- PubMed. “Stranges S et al. "Effects of long-term selenium supplementation on the incidence of type 2 diabetes: a randomized trial." Annals of Internal Medicine, 2007. PubMed. DOI: 10.7326/0003-4819-147-4-200708210-00175.” doi.org
- PubMed. “Cole BF et al. "Folic acid for the prevention of colorectal adenomas: a randomized clinical trial." JAMA, 2007. PubMed. DOI: 10.1001/jama.297.21.2351.” doi.org
- PubMed. “MRC Vitamin Study Research Group. "Prevention of neural tube defects: results of the Medical Research Council Vitamin Study." Lancet, 1991. PubMed. PMID: 1677062.” pubmed.ncbi.nlm.nih.gov
- PubMed. “Czeizel AE, Dudas I. "Prevention of the first occurrence of neural-tube defects by periconceptional vitamin supplementation." New England Journal of Medicine, 1992. PubMed. DOI: 10.1056/NEJM199212243272602.” doi.org
- PubMed. “Chapuy MC et al. "Vitamin D3 and calcium to prevent hip fractures in elderly women." New England Journal of Medicine, 1992. PubMed. DOI: 10.1056/NEJM199212033272305.” doi.org
- PubMed. “Geller AI et al. "Emergency Department Visits for Adverse Events Related to Dietary Supplements." New England Journal of Medicine, 2015. PubMed. DOI: 10.1056/NEJMsa1504267.” doi.org
- PubMed. “Timbo BB et al. "Dietary Supplement Adverse Event Report Data From the FDA Center for Food Safety and Applied Nutrition Adverse Event Reporting System (CAERS), 2004-2013." Annals of Pharmacotherapy, 2018. PubMed. DOI: 10.1177/1060028017744316.” doi.org
- PubMed. “Tucker J, Fischer T, Upjohn L, Mazzera D, Kumar M. "Unapproved Pharmaceutical Ingredients Included in Dietary Supplements Associated With US Food and Drug Administration Warnings." JAMA Network Open, 2018. PubMed. DOI: 10.1001/jamanetworkopen.2018.3337.” doi.org
- PubMed. “Lundh A, Lexchin J, Mintzes B, Schroll JB, Bero L. "Industry sponsorship and research outcome." Cochrane Database of Systematic Reviews, 2017. PubMed. DOI: 10.1002/14651858.MR000033.pub3.” doi.org