Astragalus root

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

Evidence grades used in this section: clinical (seen in people: trials, case reports with causality assessed), pharmacokinetic study, animal or in vitro, theoretical (inferred from pharmacology), and traditional (a rule of practice with no trial behind it).

Adverse effects in trials (clinical). The oral single-herb trials are small and mostly silent on safety. The telomere trial of a multi-ingredient product reported that all 40 participants completed and no adverse effects occurred over 6 months [30]. The knee, rowing, resistance-training and diastolic-function trials did not report adverse events at all [26–29], so their silence is no evidence of safety. In the heart-failure review, 7 of 19 trials reported on adverse reactions and 4 of those reported none; the reactions seen were hypotension, sinus bradycardia, nausea, dizziness and cough, 13.4 per cent with astragalus against 16.3 per cent with conventional care alone, and the authors say they usually settled with symptom management [24]. In the cancer-fatigue review, injected preparations were followed by fever, and in one trial rash, eczema and pruritus, mostly mild [23]. The Cochrane CKD review notes that astragalus injection has been associated with allergic reactions including allergic shock; 6 of its 22 trials reported no adverse effects and the other 16 did not report on them [22]. Bone and Mills: controlled studies suggest a low incidence of significant side effects, and "no adverse effects are expected if used as recommended" [5, pp. 382, 389]. Braun and Cohen: none known [7, p. 53].

Case reports (clinical). A 21-year-old man with atopic dermatitis developed oedematous erythema of both hands and feet about 18 hours after astragalus root was added to a Kampo decoction he had taken without trouble for three years; it cleared within 2 days of stopping, and he continued the original formula afterwards without recurrence. Astragaloside was negative on lymphocyte transformation testing and he declined rechallenge [48]. AIP causality assessment (WHO-UMC criteria): probable, on timing, positive dechallenge and tolerance of the base formula; the causative constituent is unknown. The authors cite two earlier eruption cases linked to astragalus-containing medicines, which were not read for this draft. A 2014 case report of raised CA19-9 with liver and kidney cysts attributed to astragalus is on order and is not assessed here until read.

Toxicology (animal). No adverse effects within 48 hours of single oral doses of 75 and 100 g/kg in mice [4, p. 55; 5, p. 389]. Intraperitoneal LD50 in mice reported as 40 g/kg (Bone and Mills) or 39.8 g/kg (Mills et al.); in another study 50 g/kg i.p. produced no significant toxicity [5, p. 389; 8, p. 39]. A 3-month study of astragalus given by injection found no distinct toxicity up to 39.9 g/kg in rats and 19.95 g/kg in dogs [5, p. 389]. A 90-day study of a root water extract at 47 g crude drug/kg/day in rats found no mortality or organ pathology; food intake, rectal temperature and rotarod latency fell and recovered within the 30-day recovery period, so 47 g/kg is a dose with transient effects and not a clean no-effect level [31]. An aqueous extract at 1.25 mg/mL modestly raised aberrant cells (16 per cent) in vitro; other extracts were non-mutagenic in the Ames test or antimutagenic [4, p. 56; 5, p. 389; 8, p. 39]. Intravenous astragaloside IV was maternally toxic at 1.0 mg/kg and fetotoxic above 0.5 mg/kg in rats, without teratogenicity in rats and rabbits [5, p. 389]. AIP comment: converted to human-equivalent doses by body-surface area [49], the highest oral animal doses above (100 g/kg once in mice; 47 g/kg/day for 90 days in rats) are about 18–19 times the top traditional human dose of 30 g/day, and the 75 g/kg mouse dose about 14 times (Figure 15). That is a comfortable margin for acute and subchronic use, and a much smaller one than the raw mg/kg figures suggest.

Immune effect at high dose (secondary report). Mills et al. state that doses above 28 g/day may suppress immunity, citing a secondary reference work [8, pp. 39–40]. The primary source was not read and the claim is ungraded; it sits at the top of the traditional dose range (9–30 g/day), and is a reason not to push decoction doses beyond it without cause.

Contraindications (traditional). Acute infection, on traditional grounds (Bone and Mills; MediHerb; Braun and Cohen) [5, p. 389; 6, p. 26; 7, p. 54]. No trial has tested this rule. Bone and Mills list no other special warnings [5, p. 389].

Herb–drug interactions.

No clinical or pharmacokinetic interaction study of the oral drug in people was among the sources read. A human P-glycoprotein study with fexofenadine (2013) is on order and is not cited here until read. MediHerb refers users to its herb–drug interaction chart [6].

Pregnancy (no human data). Bone and Mills place astragalus in category B1 of the Australian pregnancy categorisation (no increase in malformation from limited use in women; no fetal damage in animal studies of oral doses); this is the authors’ own assignment, not a TGA categorisation [5, p. 389]. Braun and Cohen: safety unknown, although no fetal damage has been reported in animal studies [7, p. 54]. Mills et al.: safety unknown, no report either way (their evidence level 5) [8, p. 38]. WHO: no data, so it should not be given in pregnancy [4, p. 56]. MediHerb: professional supervision suggested [6, p. 116]. Locoweed toxicity belongs to other Astragalus species; swainsonine has not been detected in this drug [5, p. 389; 8, pp. 38–39].

Lactation (no data). Bone and Mills: no data on safety in breastfeeding [5, p. 389]. Mills et al.: unknown (evidence level 5) [8, p. 39]. WHO: use not recommended, since excretion into milk is not established [4, p. 56]. MediHerb: professional supervision [6, p. 116].

AIP comment on pregnancy and lactation: there are no human data either way. The conservative course, which WHO takes, is to avoid it; where it is used, it should be only under the care of a degree-qualified herbalist with competencies and adequate training in pharmacognosy.

Children and older people. Children: limited information; adverse effects are not expected (Bone and Mills) [5, p. 389]. The Cochrane review found no placebo-controlled trial in children [21]. Older people: no source read gives advice specific to older adults; the theoretical interactions above (immunosuppressants, cardiac drugs) matter more in this group because they take more medicines.

Overdose. No incidents found in the published literature (Bone and Mills) [5, p. 389].

TGA safety alerts and DAEN. Searched in September 2026 by Medicine. TGA safety alerts and safety information: a search of the TGA website for "astragalus" found no safety alert, product recall or safety advisory; the one regulatory notice found was the cancellation of an ARTG listing (Bioglan Astragalus, ARTG 319762) under section 30(1)(c) of the Therapeutic Goods Act 1989, published among the TGA’s cancellations by sponsors, effective 16/06/2025 [50]. The TGA site’s own search page did not return results to automated access, so this rests on an indexed web search limited to tga.gov.au. The Database of Adverse Event Notifications opens only behind a terms-of-use page that could not be passed automatically, so DAEN has not been searched; this is to be done by hand before release.

Referral. For use alongside immunosuppressant, chemotherapy or cardiac medicines, in pregnancy or breastfeeding, or in serious illness, consult a degree-qualified herbalist with competencies and adequate training in pharmacognosy.