Buchu (Agathosma betulina)

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 The shape of the safety question

Buchu leaf has a benign clinical record and a constituent with a well-characterised hepatotoxicity. Those two statements are both true and the tension between them is the whole of this section. Mills and Bone record no reports of poisoning in the published literature and only in vitro toxicology data [5]. Barnes and colleagues record no documented side effects for buchu and note that the oil contains pulegone, a known hepatotoxin [6]. The Australian regulator has resolved the tension by capping the constituent: pulegone must be present and must not exceed 4 per cent of the medicine [9].

15.2 Adverse effects from human use

Both of the first two rows come to the same place through different books. Mills and Bone attribute the gastrointestinal effect to the British Herbal Compendium 1992 rather than to observation of their own [5], and Barnes and colleagues attribute the renal and gastrointestinal irritation to the volatile oil rather than to the leaf [6]. Neither is a case series. Apart from the one case below, in which buchu was one of four herbs, the human adverse-effect record for this drug is a handful of sentences passed between reference books.

The one published case. Ding and Perera describe a two-year-old girl on holiday in England from Germany, brought to a paediatric emergency department in Brighton with two days of painful urination [49]. She passed about 400 mL of colourless urine with a specific gravity of 1.000 and an osmolality of 53 mOsm/kg, against normal serum sodium and osmolality, and was described as a "big drinker", which raised the question of diabetes insipidus. Only on the third history did her parents mention a tea of buchu, couch grass, marshmallow and plantain, bought from a London health shop the day before for a suspected urinary infection. The tea was stopped and the leucocytes in her urine were put down to vulvovaginitis. Two days later she came back with fever and worse dysuria; her urine was now positive for leucocytes, nitrites and blood, and she was treated with trimethoprim; the culture of this sample, like that of the first, grew a pure growth of coliforms [49]. The authors read the dilute urine as the herbs acting as a strong diuretic, and note that frequent voiding, made worse here by that diuresis, leaves too little time in the bladder for bacteria to turn nitrate into nitrite, so that a dipstick, already less sensitive under the age of three, can miss the infection [49]. AIP’s reading of causality: buchu cannot be singled out, since four herbs were in the tea and the volume drunk is not reported, and a large volume of any tea dilutes urine. The clinical lesson holds whichever herb was responsible. A herbal "flush" given for a suspected urinary infection in a small child can hide the infection from the first test and delay its treatment.

Tolerability in the one trial of a buchu-containing product. The 16-week Swinburne trial of a multivitamin containing 10 mg of buchu in the men’s tablet reported no significant adverse reactions and minor gastrointestinal symptoms only [48]. At that dose it says nothing about buchu (section 13.2).

No AIP clinic or dispensary adverse-event record for this herb exists.

TGA Database of Adverse Event Notifications (DAEN). The attempts of 27/09, 07/10 and 20/10/2026 got no result set, because the search is a Power BI report embedded in the page. On 26/10/2026 Medicine reached it through the terms-of-use page in a scripted browser and it returned results, with data to 12/10/2026 [50]. "buchu", "diosphenol" and "pulegone" match no medicine. "Agathosma" matches four products in the Swisse Men’s Ultivite range, with 15 reports (12 naming a single suspected medicine); "Barosma" matches the Nature’s Own Fluid Retention Formula and one more Ultivite product, with 6 reports (4 single suspected). A control search for "Hypericum" returned 150 reports, so the filter was working. That is 21 reports from 1998 to 2022, and no deaths (Figure F10n). Every product is a multi-ingredient formula: the Ultivite tablets list 46 to 53 actives each, and the fluid-retention formula six (bearberry, buchu, juniper, parsley, potassium sulfate and pyridoxine). Ten reports carry a gastrointestinal term (nausea, abdominal pain, vomiting, diarrhoea, constipation, flatulence), in line with the occasional gastrointestinal intolerance the texts above record. Two reports need naming. A 1999 report of cholestatic hepatitis with hepatic necrosis in a 22-year-old woman listed eight suspected products, greater celandine and the fluid-retention formula among them, so buchu cannot be singled out. A 1998 report in a 75-year-old woman, with indapamide and the fluid-retention formula both suspected, records low sodium, potassium and chloride; indapamide can do that on its own, and the report bears on the diuretic row in 15.5. A DAEN entry is an unverified report, and the TGA states that inclusion does not mean a medicine caused the event [50]. AIP reads the set as a very-low-certainty signal with no causality assessed, consistent with the record in 15.1. The case listing as read is kept in the evidence record behind this monograph.

TGA safety alerts. The TGA site was queried in September 2026 for "buchu" and "Agathosma". No safety alert, recall action or advisory naming this ingredient was located. Medicine repeated the query in October 2026; the TGA site search answered the automated session with a bot-verification page and its safety-alert listing returned no item naming buchu or Agathosma. Medicine ran both again in October 2026: the site search again served a bot-verification page, and the safety-alert listing for "buchu" showed six recent alerts, none naming buchu, Agathosma or pulegone, although AIP could not confirm that the keyword filter had been applied. The DAEN search in October 2026 was again the embedded Power BI report, with no result set returned to an automated session. On 26/10/2026 the alert listing gave the same six alerts for "buchu" as for a nonsense keyword, so the filter is not applied to a scripted session and the alert search still needs a person at a browser. The query was run from an automated session and AIP does not claim it was exhaustive; what can be said is that nothing turned up, and that no alert appears in any of the four practitioner texts either.

15.3 Toxicology

Cell viability. Moolla and colleagues tested methanol–dichloromethane (1:1) extracts of Agathosma species in the MTT assay; extracts of A. betulina and A. crenulata were not toxic up to 100 µg/mL [3]. The essential oils of both species were another matter: Viljoen and colleagues report IC₅₀ values of < 0.0001 µg/mL [3]. Brendler and Abdel-Tawab add, from Moolla’s 2005 thesis, that buchu essential oils at high doses were hepatotoxic in rats and affected liver and uterine function, attributed to R-(+)-pulegone depleting glutathione and producing centrilobular hepatocellular necrosis [3]. The same authors make a point that matters more in practice than any of the numbers: pulegone is not water soluble, so the caution attaches to the oil and to alcoholic extracts and not to a water infusion [3]. Mills and Bone, working from the same body of work in Graham cells (transformed human kidney epithelium), put it as the extract showing no toxicity at the concentrations tested (IC₅₀ > 100 µg/mL) while "the oil was highly active in disrupting the viability of test cells" [5]. AIP records the < 0.0001 µg/mL figure as published and notes that an IC₅₀ four orders of magnitude below the extract’s is an extraordinary result for a terpene mixture and would benefit from independent replication before it is built on.

Pulegone: the constituent that sets the safety margin. The EFSA FEEDAP Panel’s assessment is the most complete public toxicology on it that AIP holds in full text [7].

Menthofuran is present in buchu leaf oil in its own right at 0.22 to 0.37 per cent [7]. Methyleugenol, at 0.09 to 0.17 per cent, is the one constituent with experimentally proven genotoxicity and carcinogenicity in rodents; EFSA cites a BMDL₁₀ of 22.2 mg/kg bw per day from hepatocellular carcinoma in male rats [7]. α- and β-thujone are present at 0.003 to 0.010 and 0.015 to 0.038 per cent. The BMDL₁₀ of 8 mg/kg bw per day that EFSA retains is for α-thujone specifically, recalculated from 11 mg/kg bw per day derived from long-term rodent studies using clonic seizures as the response [7]. EFSA applies that same reference point to β-thujone as a worst-case assumption, because β-thujone is much the less neurotoxic of the two, and the calculation below follows EFSA and sums the two isomers against it [7].

Putting those numbers against a herbal dose. At the top of the practitioner range, 6 g of dry leaf a day [5,6], and taking the leaf at 2 per cent essential oil [5], the theoretical maximum oil intake is about 120 mg a day. The pulegone content of the oil is the variable. EFSA’s six commercial batches ran 2.13 to 7.81 per cent with a mean of 5.03, against a trade specification of 1.5 to 8 per cent [7]. At the mean that is roughly 6 mg of pulegone a day, about 0.086 mg/kg for a 70 kg adult, some 110-fold below the 14-week rat NOAEL of 9.375 mg/kg. At the highest batch EFSA measured it is about 9.4 mg a day, 0.13 mg/kg, and the margin falls to about 70-fold. Methyleugenol at the mean 0.12 per cent of the oil would be about 0.14 mg a day, roughly 0.002 mg/kg, some 10,800-fold below the rodent BMDL₁₀ of 22.2 mg/kg; at the top of its range, 0.17 per cent, about 7,600-fold. The thujones together, at their measured maxima, come to about 0.06 mg a day, roughly 0.0008 mg/kg, some 9,700-fold below the thujone BMDL₁₀ of 8 mg/kg [7].

A 70-fold margin on a subchronic rodent NOAEL is not a large one. Standard practice would divide by 10 for species difference and 10 again for human variability before calling a dose acceptable, and 70-fold does not cover that.

And the oil content is a second variable the margin is sensitive to. The 2 per cent figure above is a mid-range one. Barnes and colleagues give buchu leaf volatile oil at 1.0 to 3.5 per cent [6]. Recalculated at 3.5 per cent oil, the same 6 g daily dose carries about 210 mg of oil, which at the mean pulegone content is about 10.6 mg of pulegone a day, 0.15 mg/kg, and the margin falls from 110-fold to about 62-fold; at 3.5 per cent oil and EFSA’s highest-pulegone batch it is about 16 mg a day, 0.23 mg/kg, and 40-fold. So the honest statement of the margin for correctly identified A. betulina at the top of the dose range is a band of roughly 40- to 110-fold, not a single number, and the two things that move it are the oil yield of the batch and the pulegone content of the oil. Neither is on an Australian certificate of analysis today.

Those margins are reassuring and they are also fragile, and the reason is species. Every figure above assumes the drug is A. betulina. Substitute A. crenulata, at the Kaiser figure of 50 per cent pulegone in the oil [3], and the same 6 g daily dose delivers about 60 mg of pulegone a day, about 0.86 mg/kg — still under the rat NOAEL, but only about eleven-fold under it, which the interspecies and interindividual factors above consume entirely. Collins and colleagues put A. crenulata pulegone at 31.6 to 73.2 per cent [3]; at the top of that range the daily intake is about 88 mg, 1.3 mg/kg, seven-fold under a rodent NOAEL, and at that point nothing is left of the hundred-fold that the standard factors ask for. At the 1 to 4 g/day working range section 12.3 settles on, a mistaken A. crenulata at the Kaiser figure still comes to 0.14 to 0.57 mg/kg, 16- to 65-fold under the same NOAEL, so the species question does not go away by dosing low; it only gets less acute. A hybrid, where Mills and Bone say pulegone "varies enormously" [5], is unquantifiable. This is why the identity tests in section 10 are a safety matter as much as a purity one, and it is a fair reading of why the Australian Determination made pulegone a mandatory, capped component [9].

Interindividual variability. Zárybnický and colleagues measured the acute toxicity of R-pulegone and R-menthofuran in precision-cut human liver slices from different donors and found substantial interindividual variability, with associated miRNA expression changes, comparing the effect to paracetamol [17]. The practical point is that a population margin is not an individual margin: a person with depleted glutathione — from fasting, alcohol, paracetamol, or malnutrition — is not the average rat. The buthionine-sulfoximine result gives the size of that effect: around ten-fold [7].

15.4 Contraindications

Mills and Bone list no contraindications "on current evidence" and give their pregnancy and lactation position separately [5]. The table below sets out each text’s position beside AIP’s reading, and says where AIP is more cautious than its sources and why.

15.5 Interactions

Barnes and colleagues’ position is the one AIP adopts as the general statement: none documented, but the potential for buchu preparations to interact with concurrent medicines with similar or opposing effects should be considered [6]. One practical addition that is not in any of the texts and follows from the chemistry in 15.3: pulegone’s detoxication runs through glutathione, so anything that depletes it — a paracetamol overdose, heavy alcohol use, prolonged fasting, malnutrition — moves a person off the population margin. That is a reason to keep the dose modest in those situations rather than a reason to avoid the herb.

15.6 Overdose and occupational exposure

No reports of poisoning appear in the published literature [5,6]. The excess-dose caution in every text — Barnes: "excessive doses of buchu should not be taken in view of the potential toxicity of the volatile oil" [6] — is a precaution from constituent chemistry, not from cases.

The essential oil is a different material from the leaf and should be handled as one. EFSA concluded that buchu leaf oil "should be considered as irritant to skin and eyes, and as a skin and respiratory sensitiser", and added that when the oil is handled, exposure of unprotected users to methyleugenol cannot be excluded and should be minimised [7]. Anyone handling the neat oil in a dispensary should use gloves and eye protection and work with ventilation. In Australia the neat oil carries no schedule of its own. A preparation containing more than 4 per cent d-pulegone is a Schedule 6 poison, with Appendix E first-aid requirements and, for containers of 200 mL or less, a child-resistant closure [10]. A neat buchu oil at 2 to 8 per cent pulegone straddles that line, and a dispensary holding it should know which side of it their bottle sits on.

15.7 Australian regulatory safety position, summarised

Read against the instruments in force on 20/10/2026, the Poisons Standard October 2026 (F2026L01327, registered 30/09/2026, in force from 01/10/2026) and the Therapeutic Goods (Permissible Ingredients) Determination (No. 3) 2026 (F2026L01382, registered 13/10/2026, in force from 16/10/2026). They replaced the June 2026 Poisons Standard and the No. 2 Determination, which refs 9 and 10 also record. Every requirement below is worded as it was in the earlier editions; only some item numbers moved, and buchu leaf oil is now item 978.

• The herb is not scheduled [10].
• It is permitted as an active, excipient and homoeopathic ingredient in listed medicines, with the pulegone requirement at item 423 [9].
• Buchu leaf oil is permitted only as an excipient, as a flavour or fragrance, with the flavour at not more than 5 per cent and the fragrance at not more than 1 per cent of the medicine [9]. Australia does not permit buchu oil as an active in a listed medicine.
• d-Pulegone above 4 per cent, and methyleugenol above 1 per cent, are Schedule 6 [10].

That is a coherent regulatory position and, so far as AIP can tell, a stricter and better-targeted one than any other country’s. It is worth saying plainly because it runs against the usual assumption that Australian herbal regulation is looser than Europe’s: on this drug, it is not.