Golden seal (Hydrastis canadensis)

Free to every reader. The safety section of every monograph is open to everyone, signed in or not. In an emergency call 000. For poisoning, call the Poisons Information Centre on 13 11 26.

15. Safety

15.1 Adverse effects

There is a lack of clinical safety and toxicity data for golden seal [9]. No adverse effect was reported from the herb itself in the four healthy-volunteer interaction studies, which between them exposed about 60 people to standardised extract for 14 to 28 days [30,31,32,33].

Barnes records that the alkaloid constituents are potentially toxic and that symptoms of golden seal poisoning include stomach upset, nervous symptoms and depression, with large quantities possibly fatal; high doses of hydrastine are reported to cause exaggerated reflexes, convulsions, paralysis and death from respiratory failure; and the root may cause contact ulceration of mucosal surfaces [9]. Those statements rest on older toxicological literature and on a 1974 book on human poisoning from plants, and no modern case series supports them.

Mills and Bone note that golden seal is often described in the popular literature as toxic in large doses or as suitable only for short-term use, and suggest this may come from a misreading of the homeopath Edwin Hale’s mid-nineteenth-century provings, whose results do not necessarily translate to herbal practice [7].

Australian spontaneous reports (DAEN, searched in September 2026). The TGA Database of Adverse Event Notifications, medicines, was searched by Medicine in September 2026 by selecting every product in the DAEN medicine list whose active ingredients include hydrastis canadensis or golden seal root dry. Eight products matched: Hydrastis Canadensis (golden seal root dry), an extemporaneous MediHerb herbal formula not on the ARTG, ImmuneForce (AUST L 299066), Nutrition Care Formula 33 SE (AUST R 22572), Yeast Stat (AUST L 55942) and three Quick Cleanse detox programs. Those eight products together appear in 14 reports, 12 of them with a golden seal-containing product as the single suspected medicine, with no reported death, spread over 49 reaction rows. The most frequent reaction terms were diarrhoea, vomiting and rash (3 reports each), then abdominal pain, pain and seizure (2 each); the rest were single reports, among them ascites, colitis, nausea, drug interaction, increased lacrimation and ocular hyperaemia [66].

Read that for what it is. A spontaneous-report database gives a count of what someone chose to report, with no denominator, no causality assessment and no comparison group, and every one of these products contains other herbs as well. Fourteen reports in a register covering 678,700 reports is consistent with either a safe herb or an under-reported one, and proves neither. The single "drug interaction" term is the only entry that lines up with the pharmacology in 15.5, and one report is not a signal.

Sources conflict on the size of the risk, and the conflict is between old secondary statements and the absence of modern case reports, not between two bodies of evidence.

15.2 Toxicology

The National Toxicology Program 2-year feed study [41]. Goldenseal root powder was fed to F344/N rats and B6C3F1 mice for 2 weeks, 3 months and 2 years, with genetic toxicology in Salmonella typhimurium, Escherichia coli, mouse bone marrow cells and mouse peripheral blood erythrocytes.

Mechanism [42]. In human HepG2 cells, berberine, followed by palmatine, was the most potent DNA-damage inducer of the five golden seal alkaloids tested by comet assay and γ-H2A.X expression. Berberine and palmatine suppressed topoisomerase I and II; gene silencing showed the DNA damage was associated with topoisomerase II inhibition and not topoisomerase I. Commercial golden seal extracts also caused DNA damage, and the extent correlated positively with berberine content.

AIP comment. The two findings pull in opposite directions: no mutagenicity in the standard battery, but a topoisomerase-II mechanism for DNA damage in hepatoma cells and liver tumours at high feed concentrations. The honest reading is that a mechanism exists, that the doses producing tumours are far above human exposure, and that the relevance to humans taking 0.7-2 g/day is unknown. Medicine ruling (23/09/2026): this is a reason for the precautionary six-week course in 12.2 and for caution in anyone with liver disease or on a hepatotoxic drug, and it is not a reason to call golden seal a human carcinogen. No human liver injury attributable to golden seal was found in the searches for this monograph, and the Australian spontaneous reports in 15.1 contain no liver-injury term other than a single report of ascites in a multi-ingredient product.

Phototoxicity. Three studies from the same laboratory establish this clearly.

This is the safety point that bears directly on Australian practice, because the MediHerb guide gives directions for an eye bath and a topical cream [14], and because far north Queensland UV levels are extreme. Medicine ruling (23/09/2026), confirmed: a golden seal eye bath, gargle or topical cream is used at night, or with the treated area kept out of sunlight until it has been washed off, and the dispensing direction carries that caution in writing. The reasoning is that the concentration Chignell and colleagues estimate for an infusion, about 1 mM berberine, is forty times the 25 µM that severely damaged human lens epithelial cells under UVA, and a tincture is higher again [45]; the eye bath at the MediHerb dilution of 5 mL of a 1:3 in 70-80 mL of saline still sits well above that cell-culture concentration. The same authors found oral preparations unlikely to cause ocular phototoxicity [45], so this caution is about the topical and ophthalmic directions only. This is a precautionary ruling from in vitro data, not from a clinical study of eye baths, and it is labelled as such.

15.3 Contraindications

15.4 Pregnancy and lactation

Mills and Bone assign Category C to berberine-containing plants: plants that have caused or are associated with a substantial risk of causing harmful effects on the fetus or neonate without causing malformations, on the basis of the presence of berberine and related alkaloids. On that basis they consider golden seal best avoided during pregnancy, except for short-term use to assist labour [7]. Hydrastine 500 mg induced labour when taken orally by pregnant women [7].

The animal data are more reassuring than the classification suggests.

• A reproductive screening test gave female rats 65 times the human dose of golden seal by gavage on gestation days 1-8 or 8-15. There was no increase in pre- or post-implantation loss, no change in fetal body weight, and no difference in external or internal malformations. The same extract was toxic to gestation-day-10.5 embryos cultured in rat serum for 26 h. The authors concluded that poor absorption from the small intestine probably explains the discrepancy, that serum concentrations from oral treatment are unlikely to reach the in vitro lowest-observed-effect level, and that golden seal at the prescribed human dose is unlikely to be absorbed enough to be unsafe in pregnancy: while stating that pharmacokinetic studies are needed to confirm it [46].
• A reduction in average fetal body weight per litter was seen in the offspring of mice fed golden seal root powder at 7.7 g/kg/day from days 6 to 17, with no significant developmental toxicity below that dose and increased maternal liver weight above 2 g/kg/day without histopathological lesions [7].
• Mills and Bone offer an alternative mechanism for the pregnancy concern: berberine displaces bilirubin from protein binding, with about a tenfold greater effect than phenylbutazone in vitro, and chronic administration to rats reduced bilirubin serum protein binding. That, rather than any uterine-contracting effect, may be what the contraindication is really about [7].

AIP position, confirmed by Medicine 23/09/2026. The contraindication stands, for three reasons that have nothing to do with which mechanism is right: the Australian ARTG entries carry it as a condition of listing [6,15], the manufacturer’s own guide carries it [14], and the reference texts carry it [7,9]. The wording AIP uses is: golden seal is not used in pregnancy or during breastfeeding. No qualifier about labour is added. Mills and Bone’s exception for short-term use to assist labour [7] is recorded here as what their text says, and AIP does not adopt it: the only human observation behind it is that 500 mg of hydrastine induced labour [7], which is an account of a pharmacological effect rather than evidence that the practice is safe, and a listed medicine in Australia carries the "do not use if pregnant" statement regardless. The animal data in the three points above are reassuring on malformation and are not a reason to soften the position, since no pharmacokinetic study in pregnancy has been done, as the authors of the reproductive screening test themselves said was needed [46]. Lactation rests on the reference texts and on the neonatal-jaundice concern, with no human milk data either way. Anyone who is pregnant, trying to become pregnant or breastfeeding and is considering this herb should consult a degree-qualified herbalist with competencies and adequate training in pharmacognosy.

15.5 Herb-drug interactions

This is the best-evidenced thing about golden seal, and it is a harm.

Mechanism. (−)-β-Hydrastine is a more potent time-dependent inhibitor of midazolam 1′-hydroxylation in human intestinal microsomes than berberine (K_I 8.48 µM and k_inact 0.041 min⁻¹ against >250 µM and about 0.06 min⁻¹). A physiologically based pharmacokinetic model incorporating both alkaloids predicted the observed midazolam interaction to within 20 per cent after acute and chronic dosing, and implicated hydrastine as the major precipitant, mainly through time-dependent inhibition of intestinal CYP3A [36]. Both alkaloids contain a methylenedioxyphenyl ring, a structural alert for time-dependent inhibition of cytochromes P450, and the ring forms a metabolic-intermediate complex with the enzyme [36]. The metformin effect appears to be inhibition of intestinal uptake transporters: mouse work showed decreased metformin systemic exposure through inhibition of intestinal uptake transporters [39], and a human enteroid monolayer study probed the same mechanism [40].

Predicted but not measured. A PBPK model predicted a moderate twofold increase in bosutinib AUC (5th-95th percentile 1.4-3.1) with clinically relevant doses of goldenseal extract, and little effect on imatinib (AUC ratio 1.0-1.2). The authors state the predictions need confirmation in controlled interaction studies [38]. This is a model output and not a measured interaction.

In vitro CYP findings. Golden seal decreased the activity of CYP2C8, CYP2C9, CYP2D6, CYP2E1 and CYP3A4 in vitro; when individual alkaloids were tested, hydrastine was substantially more potent than berberine, though both were active [7]. Golden seal root was identified as the most potent CYP3A4 inhibitor of 21 popular herbal products screened in vitro; that screen is Budzinski and colleagues 2000 [34a], reported as the starting point of the indinavir study [34]. A rat study found golden seal protective against paracetamol hepatotoxicity through CYP2E1 inhibition [7].

Other reported interactions.

15.6 Australian regulatory safety position

No poison schedule is recorded on the ARTG public summaries for MediHerb Golden Seal (423693), MediHerb Sinus Forte (450186) or Nature’s Sunshine Golden Seal (25442), all produced 23/09/2026 [6,15,16]. The Therapeutic Goods (Poisons Standard—June 2026) Instrument 2026 (Cth), F2026L00633, could not be searched by full text for this draft: the Federal Register of Legislation served the instrument only through a JavaScript application on 23/09/2026 and the document-download endpoint could not be reached [30a]. The statement "no Poisons Standard entry" therefore rests on the three ARTG records and not on a reading of the instrument, and Legal and QA are asked to confirm it directly before release. The same limitation applies to the item number for hydrastis canadensis in the Therapeutic Goods (Permissible Ingredients) Determination (No. 2) 2026 (Cth), F2026L00707 [5]; what is established from the ARTG records is that the ingredient is permitted in listed medicines, since a listed medicine may only contain permitted ingredients [6,15,16].

Database of Adverse Event Notifications, medicines, searched in September 2026 by Medicine. The DAEN medicines search is now delivered as an embedded report and cannot be queried as a data service; the search was run interactively. Every product in the DAEN medicine list whose active ingredients include hydrastis canadensis or golden seal root dry was selected, eight products in all. The result is 14 reports, 12 with a golden seal-containing product as the single suspected medicine, no reported death, over 49 reaction rows. The reaction terms are set out in 15.1 [66]. Inclusion in the DAEN does not mean the event was caused by the medicine, and all but one of the eight products contain other herbs.

TGA safety alerts. The TGA website was searched in September 2026 for "goldenseal", "golden seal" and "hydrastis". Each search returned no matching result, so no TGA safety alert, recall, product defect correction or safety advisory naming golden seal was found [66a]. That is a website search and not a search of an alerts register; if the TGA publishes alerts that its site search does not index, they would not appear here.

15.7 Referral

Golden seal is a practitioner-only medicine in its Australian liquid and tablet forms [6,14,15], it is contraindicated in pregnancy and lactation, and it changes the blood levels of common prescription medicines. Anyone taking prescription medicine who is considering golden seal should consult a degree-qualified herbalist with competencies and adequate training in pharmacognosy before starting it.