Most Americans know exactly one interaction between something edible and something prescribed, and it is not a supplement. It is grapefruit juice. The Food and Drug Administration explains the reason in plain terms: an enzyme in the wall of the small intestine, CYP3A4, breaks down many drugs before they ever reach the bloodstream, and grapefruit juice blocks it. Less breakdown means more drug in the blood, staying there longer.
Now run that backwards. St John’s wort pushes the very same enzyme in the opposite direction. It prompts the body to manufacture more of the enzyme, so a drug gets dismantled faster and less of it arrives where it is needed. Same enzyme, opposite result. The second problem is far harder to notice than the first, because nothing feels wrong. A medicine that has stopped working does not announce itself the way a side effect does.
That pair is the whole idea in miniature. The thing worth carrying around is not a memorized list of products. It is a short catalog of mechanisms, because there are only a handful of physical ways a supplement and a medicine can interfere with each other. Learn the shape of each and you can spot a plausible risk in a combination nobody has studied, which is most of them.
One correction first. If you picture interactions as a herb problem, you have the wrong picture.
Minerals, not herbs, dominate the record
In 2012, researchers led by Hsiang-Hui Tsai published a systematic review in the International Journal of Clinical Practice that pulled together 85 primary research papers, six books and two websites to catalog 1,491 distinct interaction pairs, covering 213 supplement entities and 509 medications. The five supplements carrying the most documented interactions were St John’s wort, magnesium, calcium, iron and ginkgo.
Three of those five are minerals. They are the ordinary contents of a multivitamin, a bone-health product or a prenatal, and they interfere with medicines by a mechanism that gets almost no public attention. The same review found that warfarin, insulin, aspirin, digoxin and ticlopidine drew the most interaction reports, and that drugs acting on the heart and nervous system were overrepresented.
Two caveats belong with those numbers. The review dates from 2012, and ticlopidine has since largely fallen out of American practice. A count of documented interactions also partly measures how hard anyone has looked, not only how often something happens.
Enzymes, sped up and shut down
Speeding up takes about ten days
The body has no accelerator pedal for an enzyme. It has to build more enzyme protein, which takes time, and that delay explains a great deal of what goes wrong.
A 1999 study by Andreas Johne and colleagues in Clinical Pharmacology and Therapeutics shows the clock running. Healthy volunteers took digoxin to a steady blood level, then added either placebo or a St John’s wort extract. After the first dose of the extract, nothing measurable happened. After ten days, digoxin exposure had fallen by 25 percent and peak concentration by 26 percent, and the authors wrote that the effect grew more pronounced through the tenth day.
Two consequences follow, neither intuitive. The reassurance of having taken something for a week without trouble is worth little, because the mechanism had not finished switching on. And stopping a supplement is itself a change: as the extra enzyme clears, a drug dose calibrated against a faster metabolism becomes a larger effective dose.
FDA puts numbers on how big this effect must be before a drug gets labeled for it. A strong inducer cuts the blood exposure of a sensitive test drug by at least 80 percent. A moderate inducer cuts it by 50 to 80 percent, a weak one by 20 to 50 percent. On the other side, a strong inhibitor multiplies exposure at least fivefold.
In FDA’s table of examples, St John’s wort sits in the strong inducer column, one of a small number of non-drug substances listed alongside pharmaceuticals. The footnote attached to it is the most useful sentence available on this subject. FDA writes: “The effect of St. John’s wort varies widely and is preparation dependent.”
That is a federal regulator stating, on the record, that the size of the interaction depends on which bottle you bought.
What induction has done
The documented consequences are specific. A 2000 letter to The Lancet from Frank Ruschitzka and colleagues described acute rejection in two heart transplant recipients whose cyclosporine levels fell after they began taking St John’s wort. That is a two-patient report rather than a series, but the cyclosporine interaction itself is treated as established by the National Center for Complementary and Integrative Health and by LiverTox, the federal liver-injury reference.
Contraceptive failure is better measured. A 2003 study by Stephen Hall and colleagues gave 12 women a combination oral contraceptive across three cycles, adding St John’s wort during the second and third. The half-life of ethinyl estradiol fell from about 23 hours to about 12. Breakthrough bleeding occurred in two of 12 women during the control cycle and seven of 12 during the herb cycle. A probe drug tracked alongside showed the induction was happening in the gut and on first pass through the liver, not in general circulation.
Here the evidence stops being tidy, though the disagreement is narrower than it sounds. FDA classifies St John’s wort as a strong inducer of CYP3A, and a 2016 systematic review in Contraception, by authors at the Centers for Disease Control and Prevention and FDA, opens by calling the herb a known strong inducer of that same enzyme. Where the review differs is magnitude for one drug class: it concluded the pharmacokinetic evidence suggests weak to moderate induction of contraceptive metabolism specifically. Both stand, and the reconciliation is probably in FDA’s footnote, since the review noted the one study finding no pharmacokinetic difference had used a product low in hypericin.
The ovulation question is also unresolved. A 2005 study in Contraception by Patricia Murphy and colleagues reported increased follicle growth and probable ovulation. A 2003 randomized trial in the British Journal of Clinical Pharmacology by Andrea Pfrunder and colleagues found no evidence of ovulation. Both are peer reviewed and both are small.
Blocking works faster
Inhibition is the mirror image and tends to act quickly, because blocking an existing enzyme requires building nothing. Grapefruit is the familiar case. Goldenseal is the supplement version. NCCIH reports that goldenseal inhibits CYP2D6 and CYP3A4, two enzymes responsible for metabolizing more than half of pharmaceuticals in current use, and that metformin levels fell about 25 percent in healthy adults given goldenseal extract with it.
Note the direction of that last result. Inhibition usually raises drug levels, yet goldenseal lowered metformin, a drug that moves largely by transporter. Mechanisms do not sort themselves neatly by product.
The pumps at the gut wall
Enzymes are not the only machinery. Cells lining the intestine carry efflux pumps, the best known being P-glycoprotein, which push absorbed molecules back into the gut before they reach circulation. More pump means less drug absorbed.
NCCIH describes St John’s wort as a potent inducer of both cytochrome P-450 enzymes and intestinal P-glycoprotein, which is why it behaves worse than a single-mechanism agent. Johne’s digoxin finding was a transporter result, not an enzyme one: the authors concluded the likely mechanism was induction of the P-glycoprotein drug transporter.
Garlic is the clean case of a pump effect with no enzyme effect at all. A 2017 review in American Family Physician by Gary Asher and colleagues found garlic had no measurable effect on CYP1A2, CYP2D6 or CYP3A4, yet reduced concentrations of drugs that P-glycoprotein carries. Green tea extract runs the other way. The same review reports that human trials found no effect on CYP2D6 or CYP3A4, the two enzymes it examined for that product, but possible pump inhibition that may explain higher simvastatin levels.
One structural detail is worth noticing. FDA’s interaction tables have columns for transporter substrates and transporter inhibitors. There is no column for transporter inducers. The regulatory apparatus for pump induction is less developed than it is for enzymes, which is part of why this mechanism stays obscure.
When no metabolism is involved at all
Everything above changes how much drug is present. A different class of interaction leaves drug levels untouched and adds effects instead. The American Family Physician review states the distinction that matters: this kind of interaction happens when a supplement affects the drug’s mechanism directly, and adjusting the dose usually does not fix it.
Bleeding is the best-documented example. The NIH Office of Dietary Supplements states that vitamin E can inhibit platelet aggregation and antagonize vitamin K-dependent clotting factors, which is two mechanisms in one sentence, the second being the same one warfarin itself uses. ODS notes that the amounts needed for clinically significant effects are unknown but probably exceed 400 international units a day.
Sedation stacks the same way. The 2017 review describes caution with kava alongside central nervous system depressants such as benzodiazepines or alcohol, because of increased drowsiness and reduced motor reflexes. So does serotonin: NCCIH states that combining St John’s wort with certain antidepressants can lead to serotonin syndrome, describing effects ranging from tremor and diarrhea through confusion and muscle stiffness to death.
Then the counterexample, which matters for calibration. Fish oil looks like it should add to bleeding risk, and ODS says it can have antiplatelet effects at high doses, though less potently than aspirin. But ODS also reports that most research indicates 3 to 6 grams a day does not significantly affect anticoagulant status in people taking warfarin, and that a 2014 review concluded omega-3s do not change the risk of clinically significant bleeding. The confirmed high-dose signal turned out to be something else: two large trials found 4 grams a day for several years slightly increased atrial fibrillation risk among people with cardiovascular disease or at high risk of it. A plausible mechanism is a hypothesis, not a finding.
Minerals in the gut, where timing is the answer
This is the mechanism most readers have never heard explained, and it is the one behind three of the five most-documented supplements.
Divalent minerals bind drug molecules in the digestive tract, forming complexes too large and too insoluble to absorb. ODS puts it plainly on its magnesium page: magnesium can form insoluble complexes with tetracyclines. Nothing metabolic happens. The drug never gets in.
Two features set this apart. First, the loss runs both ways. ODS states that quinolone and tetracycline antibiotics can interact with zinc in the gut in a way that inhibits absorption of both the zinc and the antibiotic. Separating them is not a concession to the drug; it is how each one works.
Second, because the problem is contact in the gut rather than anything in the blood, the agencies answer it with timing rather than avoidance. And the intervals are not uniform, which is why generic advice to space things out falls short. The Office of Dietary Supplements writes on its magnesium sheet that these antibiotics should be taken at least 2 hours before or 4 to 6 hours after a magnesium-containing supplement, and says on its zinc sheet that the same interval minimizes the interaction there. For calcium and quinolone antibiotics it reports that taking the antibiotic two hours before or two hours after prevents the effect. For calcium carbonate and the thyroid medicine levothyroxine, the FDA-approved drug label instructs patients taking that supplement to avoid levothyroxine within four hours.
Symmetric in one case, asymmetric in two, and a four-hour window in another. Nobody derives these from first principles, which is the argument for asking a pharmacist rather than guessing.
Minerals also compete with one another, and here the agency hedges. ODS says calcium might interfere with iron absorption, immediately adding that the effect has not been definitively established. That hedge is part of the finding, and popular coverage routinely drops it.
Consistency, not avoidance
Vitamin K and warfarin is the most clinically important single pairing in the subject, and also the one most often described backwards.
Warfarin works by blocking vitamin K’s role in building clotting factors. The vitamin is the thing the drug was designed to oppose, so intake of it directly offsets the drug. ODS states that people taking warfarin and similar anticoagulants need to maintain a consistent intake of vitamin K from food and supplements, because sudden changes in intake can increase or decrease the anticoagulant effect, and consistency is not the same thing as a low intake or none at all.
A warfarin dose is calibrated against a person’s habitual vitamin K intake, so a sudden drop pushes the effect up and a sudden spike pushes it down. Swearing off leafy greens in week three of therapy is a change, and change is the problem.
Why the same change matters more with some drugs
A 25 percent drop in digoxin exposure is a small number. It matters because of what digoxin is.
FDA describes a small set of medicines with a much tighter therapeutic window than most, called narrow therapeutic index drugs. For these, small differences in dose or blood concentration can mean the drug stops working or becomes harmful. FDA’s summary is blunt: a small change in the dose can be toxic. The two examples FDA names are warfarin and levothyroxine.
Now look at the convergence. Warfarin, levothyroxine and digoxin turn up repeatedly across every mechanism above, and two of the three sat near the top of the 2012 review’s most-reported list. The reason is not that supplements single these drugs out. It is that a percentage change in exposure a wide-margin medicine absorbs harmlessly lands outside the usable range for these.
A seventh pathway runs the opposite way and lands squarely in this category. Drugs deplete nutrients, and nutrients can lower drug levels. ODS reports that antiepileptic medicines including valproic acid, carbamazepine and phenytoin accelerate the breakdown of vitamin B6, producing low blood levels of its active form. The reverse happens too: ODS states that pyridoxine supplementation at 200 milligrams a day for 12 to 120 days can reduce blood concentrations of phenytoin and phenobarbital, adding that whether smaller doses do anything is not known. A vitamin lowering a seizure medicine is the tightest illustration available of why interaction size and interaction consequence are different quantities.
Where the information lives, and what to ask for
Four federal resources carry real interaction content, and three are free to the public.
NCCIH, at nccih.nih.gov, publishes consumer pages and provider summaries; its herb-drug interactions digest is dated December 2024. The Office of Dietary Supplements, at ods.od.nih.gov, publishes paired consumer and health-professional fact sheets per nutrient, and the health-professional versions, which carry the interaction and upper-limit detail, are public as well. LiverTox, on the NCBI Bookshelf, comes from the National Institute of Diabetes and Digestive and Kidney Diseases with the National Library of Medicine and the Drug-Induced Liver Injury Network. The Dietary Supplement Label Database, also from ODS, catalogs what is printed on more than 200,000 US product labels. Natural Medicines, frequently cited in clinical settings, is a subscription database reached through an institutional license, so a pharmacist can usually look something up in it and a member of the public cannot.
Then the honest part, which comes from NCCIH itself. A 2021 post on the center’s research blog asked what is known about drug-herb interactions and answered: “In most instances, unfortunately, the answer to these questions is ‘We’re not sure.'” The post explains there is not enough data to reach conclusions about interactions involving most herbs used for health purposes, that case reports are hard to interpret, and that laboratory mechanisms leave it unclear whether meaningful effects occur in people. NCCIH’s response has been to fund a research center on natural product and drug interactions, led from Washington State University, the University of Pittsburgh and the University of North Carolina at Greensboro.
That is why mechanism beats memorization. The list will stay incomplete for years. The mechanisms will not change.
A pharmacist can resolve most of these questions in minutes, given the right inputs. Five things determine the answer, and they follow directly from what the mechanisms above turn on:
- The exact product, not the plant. FDA’s footnote says the effect varies by preparation, so a standardized extract and a whole-herb capsule are not interchangeable inputs.
- Whether the prescription has a narrow margin. A pharmacist knows which ones do. It is the single biggest amplifier of everything else.
- How long the supplement has been in use. Induction matures over roughly ten days, so a recent start and a six-month habit are different situations.
- Whether anything was recently stopped. Stopping is a change in exactly the way starting is.
- The clock, not just the list. For minerals the question is frequently what time of day each thing is swallowed, and the intervals differ by mineral.
Signs that warrant a call rather than a question at the next appointment are the ones suggesting a drug level has already moved: unexpected bruising or bleeding in someone on an anticoagulant, breakthrough bleeding on an oral contraceptive, the return of symptoms a medicine had been controlling, unusual drowsiness after adding a sedating product, or the cluster NCCIH describes for serotonin syndrome, including tremor, diarrhea, muscle stiffness and confusion. Anyone on an immunosuppressant after a transplant, or on medicines for HIV, epilepsy or cancer, has reason to route a new supplement to the prescribing team first, since those are the classes where the documented failures happened.
The most useful takeaway sounds the least dramatic. Ask about timing, not only about safety. A large share of the interactions in the documented record involve ordinary minerals and ordinary antibiotics or thyroid pills, and the fix is a clock rather than a prohibition. Almost nobody asks that question, because almost everybody is busy worrying about herbs.
Keep exploring: sustainable harvesting medicinal plants.
Sources
- U.S. Food and Drug Administration (CDER). “Drug Development and Drug Interactions: Table of Substrates, Inhibitors and Inducers (inhibitor/inducer strength definitions; St. John's wort as a strong CYP3A inducer; transporter tables).” fda.gov
- National Center for Complementary and Integrative Health, National Institutes of Health. “Herb-Drug Interactions: What the Science Says (Clinical Digest, December 2024).” nccih.nih.gov
- National Center for Complementary and Integrative Health, National Institutes of Health. “Finding Answers to Questions About Drug-Herb Interactions (Research Blog, September 2, 2021).” nccih.nih.gov
- National Center for Complementary and Integrative Health, National Institutes of Health. “Depression and Complementary Health Approaches: What the Science Says (Clinical Digest, December 2021).” nccih.nih.gov
- U.S. Food and Drug Administration (CDER). “CDER Conversation: Setting and Implementing Standards for Narrow Therapeutic Index Drugs.” fda.gov
- U.S. Food and Drug Administration. “Grapefruit Juice and Some Drugs Don't Mix (intestinal CYP3A4 inhibition; drug transporters).” fda.gov
- NIH Office of Dietary Supplements. “Vitamin K: Fact Sheet for Health Professionals (consistent intake with anticoagulants).” ods.od.nih.gov
- NIH Office of Dietary Supplements. “Magnesium: Fact Sheet for Health Professionals (insoluble complexes with tetracyclines; antibiotic and bisphosphonate timing).” ods.od.nih.gov
- NIH Office of Dietary Supplements. “Zinc: Fact Sheet for Health Professionals (antibiotic absorption interference in both directions; timing windows).” ods.od.nih.gov
- NIH Office of Dietary Supplements. “Calcium: Fact Sheet for Health Professionals (quinolone and levothyroxine timing).” ods.od.nih.gov
- NIH Office of Dietary Supplements. “Iron: Fact Sheet for Health Professionals (calcium-iron competition, with the agency's hedge).” ods.od.nih.gov
- NIH Office of Dietary Supplements. “Vitamin E: Fact Sheet for Health Professionals (platelet aggregation and vitamin K-dependent clotting factors).” ods.od.nih.gov
- NIH Office of Dietary Supplements. “Omega-3 Fatty Acids: Fact Sheet for Health Professionals (anticoagulant status; atrial fibrillation at 4 g/day).” ods.od.nih.gov
- NIH Office of Dietary Supplements. “Vitamin B6: Fact Sheet for Health Professionals (antiepileptic drug interactions in both directions).” ods.od.nih.gov
- LiverTox, National Institute of Diabetes and Digestive and Kidney Diseases / National Library of Medicine. “St. John's Wort (CYP3A4 and 2C9 induction; no convincing evidence of direct hepatotoxicity).” ncbi.nlm.nih.gov
- American Family Physician (American Academy of Family Physicians). “Asher GN, Corbett AH, Hawke RL. "Common Herbal Dietary Supplement-Drug Interactions." Am Fam Physician. 2017;96(2):101-107.” aafp.org
- International Journal of Clinical Practice. “Tsai H-H, Lin H-W, Simon Pickard A, Tsai H-Y, Mahady GB. "Evaluation of documented drug interactions and contraindications associated with herbs and dietary supplements: a systematic literature review." Int J Clin Pract, 2012. PubMed. DOI: 10.1111/j.1742-1241.2012.03008.x.” pubmed.ncbi.nlm.nih.gov
- Clinical Pharmacology and Therapeutics. “Johne A, Brockmoller J, Bauer S, et al. "Pharmacokinetic interaction of digoxin with an herbal extract from St John's wort (Hypericum perforatum)." Clin Pharmacol Ther, 1999. PubMed. DOI: 10.1053/cp.1999.v66.a101944.” pubmed.ncbi.nlm.nih.gov
- Clinical Pharmacology and Therapeutics. “Hall SD, Wang Z, Huang S-M, et al. "The interaction between St John's wort and an oral contraceptive." Clin Pharmacol Ther, 2003. PubMed. DOI: 10.1016/j.clpt.2003.08.009.” pubmed.ncbi.nlm.nih.gov
- Contraception. “Berry-Bibee EN, Kim M-J, Tepper NK, Riley HEM, Curtis KM. "Co-administration of St. John's wort and hormonal contraceptives: a systematic review." Contraception, 2016. PubMed. DOI: 10.1016/j.contraception.2016.07.010.” pubmed.ncbi.nlm.nih.gov
- British Journal of Clinical Pharmacology. “Pfrunder A, Schiesser M, Gerber S, et al. "Interaction of St John's wort with low-dose oral contraceptive therapy: a randomized controlled trial." Br J Clin Pharmacol, 2003. PubMed. DOI: 10.1046/j.1365-2125.2003.02005.x.” pubmed.ncbi.nlm.nih.gov
- Contraception. “Murphy PA, Kern SE, Stanczyk FZ, Westhoff CL. "Interaction of St. John's Wort with oral contraceptives: effects on the pharmacokinetics of norethindrone and ethinyl estradiol, ovarian activity and breakthrough bleeding." Contraception, 2005. PubMed. DOI: 10.1016/j.contraception.2004.11.004.” pubmed.ncbi.nlm.nih.gov
- The Lancet. “Ruschitzka F, Meier PJ, Turina M, Luscher TF, Noll G. "Acute heart transplant rejection due to Saint John's wort." Lancet, 2000. PubMed. DOI: 10.1016/S0140-6736(99)05467-7.” pubmed.ncbi.nlm.nih.gov
- NIH Office of Dietary Supplements. “Dietary Supplement Label Database (DSLD).” dsld.od.nih.gov