Kava (Piper methysticum)

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 short version

The Schedule 4 entry reaches kava in therapeutic preparations outside the Register carve-out, and the reason is hepatotoxicity [8]. Kava sold as a food is exempt from the Poisons Standard and carries the Food Standards Code warnings instead [8,71]. Within the therapeutic carve-out a label warning is required above 25 mg of kavalactones per dose, and "May harm the liver" is one of the four statements it requires [9,94]; both instruments that carry that wording were read at first instance in September 2026 and are set out in section 12. The Australian case of kava-associated liver injury that this monograph sets out in full ended in transplantation and death (15.3) [5]; a second Australian case report of kava-associated hepatic harm was published in Pathology in 2026 and its full text is on order, so the record is no longer a single case and 15.3 is to be re-read against it (20.5). Reports of that severity are rare, and rarity is not the same as absence: kava is dispensed on the assumption that the liver is watched. One further thing belongs at the top of a safety section. A two-year gavage bioassay found liver tumours in mice of both sexes given a kava extract: hepatoblastoma rising with dose in males, which the National Toxicology Program graded clear evidence of carcinogenic activity, and hepatocellular adenoma or carcinoma in females at the two lower doses, which it graded some evidence [83,84]. On that animal evidence and inadequate evidence in people, the International Agency for Research on Cancer classifies kava extract as possibly carcinogenic to humans, group 2B [85]. The doses were one to two orders of magnitude above human use and the extract cannot be matched to a marketed product; 15.3b sets out the qualifications in full. One more belongs at the top. Kava is contraindicated in pregnancy and in lactation by every source this monograph has read, and the single published human report of maternal use through pregnancy describes a newborn whose kava-type rash lost him enough fluid through the skin to provoke a hypernatraemic seizure on his fourth day (15.4, 15.7a) [53].

15.2 Adverse effects in trials

In placebo-controlled trials, standardised kava extracts have generally been well tolerated, with adverse events mild, transient and similar in nature and frequency to placebo [5,18]. The size of that reassurance has a published denominator. The European committee’s assessment counts 8136 adults exposed to oral kava products in clinical trials, and records that only 509 of them were in trials that assessed liver function; its conclusion is that the trials were neither designed nor powered to detect a rare hepatic reaction and that the data cannot exclude hepatotoxicity [80]. The raised liver enzymes seen in the trials that did test are not, on that committee’s reading, evidence of hepatotoxicity in themselves [80]. A systematic review of eight placebo-controlled trials at 55-240 mg kavalactones/day for 2 days to 24 weeks found the commonest events in both arms were gastrointestinal symptoms, tiredness, restlessness, tremor and headache [5,55]. Post-marketing surveillance: 4049 patients on WS 1490 150 mg/day (105 mg kavalactones) for 7 weeks, adverse events in 1.5 per cent; 3029 patients on a second commercial extract for 4 weeks, 2.3 per cent; and a Brazilian surveillance study of 850 patients on WS 1490 100 mg three times daily (210 mg kavalactones/day), 16.7 per cent, most commonly fatigue, nausea, confusion and gastrointestinal upset [3,5]. The two sources AIP holds disagree on the dose in the second of these. The WHO monograph describes it as 800 mg of a 30 per cent extract, equivalent to 240 mg kava pyrones daily [3]; Barnes and colleagues name the product (Antares 120) and put it at 120 mg kavalactones daily [5]. The event rate, 2.3 per cent in 3029 patients, is the same in both accounts. AIP records the discrepancy and does not settle it.

The two Australian aqueous-extract trials add specific signals: poorer memory (36 vs 23 events, p = 0.044) and tremor or shakiness (36 vs 23, p = 0.024) in the 16-week trial, with liver function test abnormalities significantly more frequent in the kava group though no participant met criteria for herb-induced hepatic injury [12]; and more headaches in the 6-week trial with no difference in liver function tests [13]. The 6-week trial also looked specifically for withdrawal and addiction and found none, and found that kava significantly increased female sexual drive on an ASEX sub-domain (p = 0.040) with no negative effect in males [14].

15.3 Hepatotoxicity

The signal. Spontaneous reports of hepatotoxicity emerged in 2000-2001, first in Switzerland, then Germany. By July 2002, 68 reports of liver toxicity associated with kava had reached regulators in Canada, France, the UK, the USA and Switzerland and Germany [5]. To the end of 2005 the WHO Uppsala Monitoring Centre’s Vigisearch database held 91 reports, describing 189 reactions, for products reported to contain P. methysticum as the only active ingredient, from nine countries; 55 of those reactions were liver and biliary disorders, including 13 hepatitis, 8 jaundice, 6 hepatocellular damage, 6 raised hepatic enzymes, 4 bilirubinaemia, 3 hepatic failure and 2 hepatic coma [5]. An Australian case is on record: a 56-year-old woman developed fatigue, nausea and jaundice after about 10 weeks of a combination product stated to contain 60 mg kavalactones, Passiflora incarnata and Scutellaria lateriflora per tablet three times daily; biopsy showed non-specific severe acute hepatitis with pan-acinar necrosis and collapse of the hepatic lobules. She underwent liver transplantation on day 17 after admission, the procedure was complicated, and she died of progressive blood loss and circulatory failure; examination afterwards confirmed massive hepatic necrosis [5]. The product’s composition could not be confirmed: the Scutellaria was not identified in the product on analysis [5]. That case is the one published Australian fatality attributed to a kava-containing medicine, and it was a combination product taken for about ten weeks at 180 mg kavalactones a day, inside the daily ceiling Australia now permits and well outside the three-month duration limit the reference texts set.

What the pooled international case series contains. The European committee’s assessment gathers the regulatory record in one place, and its figures are larger than the Uppsala counts above because they include national files as well as spontaneous reports [80]. To July 2002 the WHO identified 93 cases of hepatic events worldwide associated with kava preparations, including liver failure leading to 14 liver transplants and 7 deaths, with the possibility of a small number of duplicate entries [80]. Causality on the WHO scale: 8 of the 93 (8.6 per cent) probable, 53 possible, 28 unassessable for want of data [80]. Those three categories account for 89 of the 93; the assessment does not say how the remaining four were graded, and Figure 31 shows them as not stated rather than folding them into a category. In the United Kingdom the regulator held 68 cases of adverse liver reaction when it prohibited unlicensed kava products in 2002 and a further 42 by 2005, 110 in all; nine had a fatal outcome, seven of them from the hepatotoxicity and two during or soon after a transplant, and a further nine people were transplanted and survived [80]. Its expert working group concluded that kava carried an unacceptable risk of idiosyncratic hepatotoxicity that no regulatory measure short of removing the products could prevent [80].

Causality when the cases are read one by one. Every pooled series shrinks when individual causality is assessed, and the assessment report sets out four such reviews [80]. Schmidt’s analysis of 82 reports drawn from eight regulators, the literature and two newspaper stories (one of the 82 came from the Australian TGA) discarded 20 as plainly unrelated, found a potentially hepatotoxic concomitant medicine in 21, considerable doubt in 7 and data too fragmentary to assess in 31, leaving three where a hepatotoxic effect of kava could be established and only one where kava had been taken within the German Commission E dose and duration [80]. From that residue Schmidt calculated an incidence of more than 0.02 cases per million daily doses, fewer than one case in 50 million days of use, and set it beside a figure of one case per 472,000 days for diazepam [80]. Stickel’s series of 29 new cases with 7 already published found hepatic necrosis or cholestatic hepatitis with both alcoholic and acetonic extracts, in 27 women and 9 men, with 9 developing fulminant liver failure, 8 transplanted and 3 dead, and complete recovery in everyone else after kava was withdrawn [80]. Teschke’s independent reading of 19 German cases found 2 probable, with 80 per cent of those patients having taken an overdose or self-medicated for longer than three months [80]. AIP comment (Medicine): two things have to be held at once. Spontaneous reporting under-counts, so an incidence calculated from it is a floor and not a rate, and the WHO made the point that a case coded unassessable cannot be reclassified as unrelated [80]. The true incidence is unknown and only a proper epidemiological study will give it [80]. What the causality work does establish is that severe cases cluster in overdose, prolonged use and co-medication, which is exactly where a prescriber can act.

The Australian fatality, with a causality score. The case set out above appears in the same assessment as Gow and colleagues 2003, scored RUCAM +8, a probable adverse drug reaction [80]. That is the strongest causality attribution any Australian kava case carries in the literature AIP has read.

Two further cases matter to the Australian position because they involve the preparation Australian law permits. Barnes and colleagues record two women in their fifties in New Caledonia who developed icterus and raised liver function tests four to five weeks after beginning kava prepared in the traditional way as an aqueous preparation [5]. Until 2003 it was thought that kava liver injury belonged to the ethanolic and acetonic extracts alone; those reports are the reason that view no longer holds [5].

Frequency and causality. None of the clinical trials or post-marketing surveillance studies reported hepatotoxicity as an observed adverse event, and seven of the trials in the Cochrane review monitored liver function and found no abnormality [5,18]. A clinical survey and critical analysis of 26 suspected German cases excluded causality in 18; of the remaining eight, only one patient had adhered to the regulatory conditions of no more than 120 mg kavalactones/day for no more than 3 months. The authors concluded that kava taken as recommended is associated only rarely with hepatotoxicity, and that excessive dose, prolonged treatment and co-medication raise the risk [4]. The validity of the causality scales used by the German authority has been challenged: the WHO and Naranjo scales lack liver specificity, and when the liver-specific CIOMS scale is applied the likelihood of causality falls substantially [4].

What causes it: four hypotheses, none proven.

The aqueous-versus-solvent argument. Organic-solvent extracts are markedly more cytotoxic to HepG2 cells than water extracts, and in one bioassay-guided study the water extracts of every plant part showed no toxic effect at all [4,36]. Glutathione, present in water-based extracts and not in solvent extracts, has been proposed as protective [4]. Against that, Teschke and colleagues point out that liver injury has been reported after traditional aqueous extracts in the South Pacific as well as after acetonic and ethanolic extracts in the West, suggesting the solvent plays no major causative role, and that cultivar and plant part may matter more [29,30]. A 2026 in vitro study found neither an aqueous nor an ethanol extract cytotoxic to human hepatocytes, while flavokavain B alone was [42]. The European assessment adds causality-scored cases on the same side of the argument: after 2002 it records two cases of acute liver injury at RUCAM +5 and +6 and one of toxic hepatitis at RUCAM +10, all three after traditional aqueous kava preparations, together with an icteric hepatitis at RUCAM +7 after an unknown preparation and the fatal Australian case at +8 after a combination product [80]. Cases in which causality was demonstrated therefore exist for the aqueous route, which is the route Australian law permits.

Population-level data from the Pacific. Surveys have repeatedly failed to find kava hepatotoxicity in heavy traditional users: a Samoan survey in 2002 found no case; an Arnhem Land community study found mild reversible liver-enzyme changes, probably enzyme induction and not injury; a Fijian cross-sectional study of 101 participants found significantly raised GGT and ALP in kava consumers, strongly associated with total lifetime consumption, but no association with clinically significant adverse health effects; a Hawai’i study in a predominantly Tongan population found GGT elevated in 65 per cent of 31 drinkers versus 26 per cent of 31 controls (p = 0.005) and ALP in 23 versus 3 per cent (p = 0.053) [4]. Liver function changes in users of moderate aqueous kava were reversible and began returning to baseline after 1-2 weeks of abstinence [4].

AIP position. Kava hepatotoxicity is real, rare, idiosyncratic and of unresolved mechanism. The evidence is genuinely conflicting on cause, and this monograph says so. What is not in doubt is that the Australian regulator treats it as serious enough to schedule the herb, and that a practitioner using kava should treat a change in appetite, nausea, abdominal discomfort, dark urine or jaundice as a reason to stop the herb at once.

15.3b Carcinogenicity and genotoxicity

The WHO monograph reports only negative genotoxicity work: a standardised extract of 70 per cent kava pyrones up to 600 mg/kg body weight produced no increase in micronucleated polychromatic erythrocytes and no change in the polychromatic to normochromatic ratio, and there was no increase in revertants in Salmonella typhimurium TA98, TA100, TA1535, TA1537 and TA1538 with or without metabolic activation at up to 2.5 mg per plate [3]. The two-year bioassays came later and the WHO text predates them.

The two-year bioassays, read at first instance. Earlier drafts of this monograph took the bioassay result through the European committee’s assessment report [80]. Medicine has now read the three primary and synthesis sources themselves: the National Toxicology Program’s own technical report [83], the peer-reviewed publication of the same studies by Behl and colleagues [84], and the evaluation by the International Agency for Research on Cancer [85]. Reading them changed three things on this page, and one of them was a finding in female mice that the earlier text recorded as absent. What was tested. One kava extract, 98.04 per cent pure on high-performance liquid chromatography with ultraviolet detection, containing 27 per cent kavalactones (3 per cent methysticin, 6.3 per cent kavain, 7 per cent dihydrokavain, 4 per cent yangonin, 3 per cent dihydromethysticin), given by gavage in corn oil, five days a week: to B6C3F1 mice for 105 weeks at 0, 0.25, 0.5 and 1.0 g/kg/day, and to F344/N rats for 104 weeks (males) or 105 weeks (females) at 0, 0.1, 0.3 and 1.0 g/kg/day, 50 animals of each sex per group in mice and 49 or 50 males and 50 females per group in rats [83,84,85].

What the toxicology programme itself concluded, in its own four grades. The National Toxicology Program does not grade a substance as a human carcinogen; it grades the strength of the animal evidence, sex by sex and species by species, and its verdicts on this extract were: clear evidence of carcinogenic activity in male mice, on the hepatoblastomas; some evidence in female mice, on the combined hepatocellular adenoma or carcinoma; equivocal evidence in male rats, on the marginal Leydig cell adenomas; and no evidence in female rats [83]. It also recorded non-neoplastic lesions in the liver, forestomach, kidney, eye and pancreas of rats of both sexes, the liver of mice of both sexes and the forestomach of female mice, and judged the extract non-mutagenic, so that the carcinogenic activity in mice is most likely non-genotoxic, through enzyme activation or free-radical generation [83].

Where "group 2B" actually comes from, and a correction to this monograph. The classification of kava extract as possibly carcinogenic to humans, group 2B is the International Agency for Research on Cancer’s, made by a working group that met in Lyon in June 2013 and published in volume 108 of its monograph series in 2016: inadequate evidence in humans, sufficient evidence in experimental animals, overall group 2B [85]. The European committee’s assessment report attributes those words, group number included, to the National Toxicology Program [80]. It is the wrong body: group 2B belongs to the cancer agency’s scale and appears nowhere in the toxicology programme’s own grading, which is quoted above. Earlier versions of this monograph followed the assessment report and repeated the misattribution, in this section and in the summary at 15.1. Both are corrected here, and the misattribution in the assessment report is recorded rather than passed on.

Cancer in people: one study, and the working group set it aside. The cancer agency found a single epidemiological study of cancer and kava. Steiner (2000) compared cancer incidence in men in six South Pacific countries against a surrogate measure of exposure built from the number of kava plants under cultivation in each country, on the assumption that all kava grown before 1990 was consumed locally and mostly by men. An inverse correlation was reported between all-cancer incidence in men and estimated exposure, with no confidence interval and no test of statistical significance; the working group judged the study uninformative on its ecological design, its crude exposure and outcome measures and its failure to assess the role of chance, and concluded that the evidence in humans is inadequate [85]. AIP comment (Medicine): that is the whole of the human cancer evidence on kava, and it points the pleasant way without carrying any weight. It is neither reassurance nor alarm, and nobody should quote the inverse correlation as a protective effect.

How to read it. Five qualifications travel with this finding, and every one of them is in the primary sources.

1. The extract is characterised by its kavalactone pattern and its chromatographic purity and by nothing else. No drug-to-extract ratio and no extraction solvent are recorded, so the European assessor attributed the result to that preparation alone, and this monograph does the same [80]. Whether the material resembles an aqueous Australian tablet is unknown.
2. The doses are far above therapeutic use. The lowest mouse dose, 0.25 g/kg/day of a 27 per cent extract, delivers about 67 mg of kavalactones per kg of body weight on a dosing day, against about 3.6 mg/kg for the Australian daily ceiling of 250 mg in a 70 kg adult. AIP comment (Medicine): that is roughly 19 to 1 at the lowest dose tested and about 75 to 1 at the top. Two things narrow it and are not applied here: dosing was five days a week, so a seven-day average is about five-sevenths of those figures, and no interspecies scaling has been applied. The arithmetic is AIP’s, worked on the doses and the kavalactone per cent the primary report states; it is not a measurement.
3. The female mouse dose-response does not rise with dose, and the report explains why it may not. Liver tumours were significant at 0.25 and 0.5 g/kg and not at 1.0 g/kg, where body weight was 11 per cent below control. The toxicology programme fitted a model adjusting for body weight, survival, housing and route and found the predicted incidence in the absence of any chemical effect no different from what the top-dose females showed, which is why it graded female mice some evidence rather than clear [83]. A non-monotonic result with a plausible mechanical explanation is weaker evidence than a monotonic one, and it is not no evidence.
4. The male rat finding rests on an unusually clean control group. A 76 per cent control incidence of Leydig cell adenoma sits at the bottom of a historical range of 76 to 94 per cent, and the dosed groups at 88 to 98 per cent sit inside or barely above it. The toxicology programme said so itself, in these words: "The broad range of incidences in the historical control data together with the lower than expected incidence of neoplasms in the concurrent vehicle controls, poses uncertainty in determining whether the testicular adenomas were treatment related." That is why the grade is equivocal [83]. AIP comment (Medicine): this is the row a reader is most likely to over-read, and on the primary report it is the weakest of the four.
5. Hepatoblastoma is rare in mice and has no established human counterpart here. It has a low spontaneous background, about a quarter metastasise, and the incidences in this study lie outside the historical control range, which is what made the male-mouse grade clear [83,84]. Its timing differs between species and no mechanism is established for this preparation [80,83].

AIP position (Medicine). A two-year gavage bioassay in two species found liver tumours in mice of both sexes at doses one to two orders of magnitude above anything a person takes, from an extract nobody can match to a marketed product, and the cancer agency’s reading of the human evidence is that there is effectively none. None of that is a reason to tell a patient that kava causes liver cancer, and saying so would misrepresent every source above. It is a reason the liver stays the organ of interest in every decision about this herb, and a reason a two-year carcinogenicity result belongs in a safety section rather than in a footnote.

15.4 Skin: kava dermopathy

A dry, scaly, ichthyosiform pigmented skin condition is the best-characterised adverse effect of heavy kava use, first reported by Cook’s expedition members [4]. It is dose-related, appears at long-term intakes of 400 mg kavalactones/day or more, and regresses quickly on cessation [4]. A clinical pharmacological evaluation of dihydromethysticin found 300-800 mg/day produced a scaly rash in a high proportion of subjects [4]. The cause is unresolved; interference with cholesterol metabolism has been proposed, and a trial of 100 mg/day nicotinamide failed to correct it, which argues against simple niacin deficiency [4].

Recent Australian dermatology literature has sharpened the picture. A 2024 detailed histopathological analysis of kava-induced dermatitis, a 2025 case series describing acute kava dermatitis with DRESS-like features and sebotropic inflammation, and a 2026 Medical Journal of Australia case report on diagnosing acute kava dermopathy all describe an acute inflammatory presentation distinct from the chronic ichthyosiform picture [50,51,52]. A 2024 report describes neonatal and maternal ichthyosiform dermopathy after kava use through pregnancy, and Medicine read it at first instance in September 2026 [53]. A 41-year-old woman had been drinking kava and kratom daily from a kava bar for about a year before she conceived. About three months into the pregnancy she moved to a commercial tonic of kava root extract and kratom leaf, and drank 9-10 two-ounce bottles of it a day; from 32 weeks of gestation she cut back to four bottles a day, on a buprenorphine wean her prescriber had started for the kratom half and which she could only tolerate at 1 mg sublingually twice daily. She was also taking sertraline and levothyroxine and vaping nicotine daily. From her sixth month she had a dry, scaly, non-pruritic, non-erythematous rash on the shoulders and lower limbs, and it cleared about three weeks after she stopped the tonic following delivery [53].

The neonatal course runs more severely than the adult picture, and that is the part a prescriber needs. The infant was delivered at 36 weeks and 6 days, needed positive-pressure ventilation for the first four minutes of life, and went to intensive care at three hours of age with generalised hypertonia and jitteriness, managed as neonatal opioid withdrawal. Diffuse flaking ichthyotic lesions appeared on day 3 and progressed to skin taut enough to stop him closing his mouth or eyes, with the lips and the lower eyelids held everted. On day 4 he had a generalised tonic-clonic seizure, treated with intravenous lorazepam and phenobarbital, and was intubated for the desaturations that came with it; the neonatology team attributed the seizure to a serum sodium of 155 mmol/L, driven by insensible fluid loss through the rash. A blood culture grew group B Streptococcus and Serratia marcescens, treated with 14 days of intravenous penicillin G and cefepime. He was extubated on day 6, the rash had resolved by day 22, he went home on day 25 on a morphine taper, and at nine-month follow-up he was meeting his milestones with no recurrence of the rash [53].

AIP comment (Medicine). The authors rest causality on the timing, on the rash clearing in mother and infant once exposure stopped, and on the absence of a better explanation, and they set out their own limits plainly: no residual product was kept for assay, they could find no major clinical laboratory that tests samples for kava constituents, the rash ran through a period of critical illness, and no genetic testing for a congenital ichthyosis was done, though spontaneous resolution without recurrence makes one unlikely. Kratom was taken alongside the kava throughout, and their argument for looking past it is that kratom has been linked to photodistributed hyperpigmentation and not to an ichthyosiform rash [53]. One report carries no frequency and this one cannot isolate kava. What it does put on the record is that a neonatal kava dermopathy can be severe enough to lose fluid through the skin and provoke a hypernatraemic seizure, which is a different order of harm from the reversible adult rash, and it is one of the reasons the position at 15.7 is stated as flatly as it is.

Other skin reactions on record: systemic contact-type dermatitis after oral kava; two cases of a drug eruption in sebaceous-gland-rich areas after 3 weeks of systemic kava, one with a strongly positive patch test and one with significant proliferation of peripheral blood lymphocytes to kava extract; urticaria; delayed-type hypersensitivity; and one case of dermatomyositis [4].

15.5 Neurological and other effects of heavy use

Mathews and colleagues surveyed 39 kava users and 34 non-users in a coastal Arnhem Land community, reported in 1988 [32]. Twenty respondents (27 per cent) were very heavy users at a mean 440 g/week and 15 (21 per cent) heavy users at 310 g/week. Findings in users: poor self-rated health, a "puffy" face, the typical scaly rash, slightly increased patellar reflexes; very heavy users were 20 per cent underweight with greatly increased GGT; albumin, plasma protein, urea and bilirubin decreased; HDL cholesterol increased; haematuria more frequent, urine poorly acidified and of low specific gravity; increased red-cell volume, decreased platelet volume, decreased lymphocyte count; shortness of breath with tall P waves on ECG suggesting pulmonary hypertension.

Clough and colleagues’ cross-sectional study of 101 adults enrolled in an east Arnhem Land community in March 2000, published in 2003, confirmed the core of it: dermopathy more frequent (p < 0.001), raised GGT and ALP (p < 0.001), lower lymphocyte counts (p < 0.001), with fibrinogen, plasminogen activator inhibitor-1 and neurocognitive tests not different between kava categories [33]. A second case-control study from the same programme found no association between kava use and pneumonia (crude OR 1.26, 0.74-2.14; adjusted OR 1.98, 0.63-6.23), while cannabis (OR 2.27, 1.18-4.37) and alcohol (OR 1.95, 1.07-3.53) were significantly associated with pneumonia [35a]. A third studied kava use and ischaemic heart disease in the same communities [34a]; its result is not stated here, because the paper is a two-page short report held in PubMed Central only as a scanned, PDF-only record with no abstract and no machine-readable text, and the Education and Research reference audit of 23/09/2026 could not open the full text. The Librarian is asked to obtain it. What can be said on the cardiovascular question from a source AIP has read is narrower: in the cross-sectional study, markers of cardiovascular risk were raised across the whole community and were not higher in kava users [33].

Acute neurological presentations on record: generalised choreoathetosis involving limbs, trunk, neck and facial musculature with marked tongue athetosis in a 27-year-old Aboriginal man after kava bingeing, on three occasions, other causes excluded [4]; vertigo, vomiting and ataxia resolving in 4-6 h [4]; three cases of meningismus in France, transient, in related patients using possibly adulterated material [4]; acute urinary retention in a 61-year-old Tongan man after 1 L of kava [4]. Four cases of apparent dopamine antagonism, including acute dystonia, oral and lingual dyskinesia and aggravation of Parkinson’s disease, have been described by German neurologists [4]; against that, kava extract inhibited haloperidol-induced catalepsy in rats, and an open-label study of 42 patients on neuroleptics found kava significantly reduced extrapyramidal signs [4]. A 2025 report describes severe kava withdrawal managed with phenobarbital in a 45-year-old man [54]. It sits against the negative withdrawal and addiction finding of the Australian 6-week trial [14], and it carries the same limitation as the pregnancy case in 15.7a: the man had been using a beverage containing both kava and kratom heavily and had stopped several days before presenting, and he had already been started on buprenorphine for suspected kratom withdrawal, so kava is not isolated as the cause. The treating team’s reading was that a GABAergic withdrawal was running underneath the opioid one [54].

Visual and oculomotor effects. Reduced near-point of accommodation and convergence, increased pupil diameter and disturbance of ocular muscle balance after 500 mL of traditionally prepared beverage, maximal 30-40 min after drinking [3,4]. In intoxicated Australian Indigenous kava drinkers (mean 205 g per person, about 100 times a clinical dose, over 14.4 h) there was ataxia, tremor, sedation, blepharospasm, raised GGT and ALP, saccadic dysmetria and saccadic slowing, with complex cognitive functions performed normally [4].

15.6 Interactions

15.7 Contraindications

• Pre-existing liver damage or liver disease [3,4].
• Pregnancy and lactation. Sourced and graded in 15.7a below. AIP position: do not use in pregnancy or lactation.
• Endogenous depression: Commission E [3,4]. The European assessment gives the reason: the risk of hepatotoxicity rises when kava is taken with antidepressants [80].
• Existing liver disease, and alcohol abuse: carried as contraindications in the European assessment from a European scientific co-operative monograph [80]. The liver-disease contraindication is already in the first bullet; alcohol abuse is an addition to the list this monograph previously carried.
• Warning signs that mean stop. Unusual fatigue, weakness, loss of appetite, unintended weight loss, yellowing of the whites of the eyes or of the skin, dark urine or colourless stools can all be signs of liver damage [80].
• Children under 12: the Australian TGA position as recorded by Mills and Bone [4].
• Parkinson’s disease, and caution in older people, because of the dopamine-antagonism reports [4].
• Driving and machinery. WHO warns that motor reflexes and the ability to drive or operate heavy machinery may be adversely affected even within the recommended dose range [3]. The controlled studies at 210-240 mg kavalactones/day found no impairment of safety-related performance [4], and a driving-simulator arm of the acute crossover trial found no impairing effect of 180 mg kavalactones on any driving outcome against placebo, where oxazepam 30 mg slowed braking reaction time against both placebo (p = 0.002) and kava (p = 0.003) and produced more lapses of concentration than kava (p = 0.033) [80]. A European scientific co-operative monograph likewise held that kava did not impair driving [80]. The WHO warning stands against all of that, and the Australian 16-week trial found more tremor and poorer memory on kava [12].

15.7a Pregnancy and lactation, source by source

AIP position (Medicine, 23/09/2026). Every source above that contraindicates kava in pregnancy or lactation does so for want of data, and the two texts that grade their own evidence both grade the human evidence at their lowest level. That does not soften the conclusion. A herb whose principal recognised harm is idiosyncratic liver injury, whose lactones are lipophilic enough to be expected in milk, and which now has one published report of a neonatal dermopathy severe enough to cost the baby enough fluid through the skin to provoke a hypernatraemic seizure after maternal use of a kava-containing product, has nothing to recommend it in pregnancy or lactation and a statutory Australian label warning against it. Do not use it. Where a pregnant or breastfeeding woman is already taking kava, the question of what to do next belongs with a degree-qualified herbalist with competencies and adequate training in pharmacognosy working alongside her treating practitioner.

15.8 Overdose and abuse

Excessive kava drinking causes pupil dilation, reduced light reflexes, photophobia, bloodshot eyes and poor attention to diet [4]. The Australian experience in Arnhem Land is set out in 15.5. Intake in those communities most commonly ran 100-500 g/week and reached 900 g/week [4,32]. The European assessment puts wider figures on the same question: weekly per capita consumption in Arnhem Land estimated at 145 g of powder for 1989-90 and 368 g for 1990-91, against a literature range across all surveys of 39 to 1840 g of powder, or 4.1 to 188.6 g of lactones, per week [80]. It also records a case of rhabdomyolysis after a large quantity of kava in a 34-year-old man, and the visual and oculomotor disturbance described in 15.5 after 600 mL of beverage drunk in 15 minutes; the WHO account of that same literature gives the volume as 500 mL, and AIP records both figures rather than choosing between them [3,80]. There are no reported cases of overdose with kava extracts as distinct from the beverage [80]. Three cases of sudden cardiac death in young Aboriginal men who drank kava the evening before a football match are on record, with no electrocardiographic or epidemiological evidence that sudden death is more frequent in kava-using communities than in other Aboriginal communities [4]. A 2025 case describes severe kava withdrawal managed with phenobarbital, in a man who had been using a combination kava and kratom beverage (15.5) [54]. AIP comment: much of the Arnhem Land morbidity is confounded by heavy alcohol use, malnutrition and the socioeconomic conditions in those communities, and several of the original observations have been questioned in the literature on exactly that ground [4]. The confounding does not make the observations disappear; it makes them hard to attribute.

15.9 Regulatory safety actions

TGA Database of Adverse Event Notifications (DAEN). Medicine attempted the DAEN medicines search in September 2026 and repeated it the same day. daen.tga.gov.au/medicines-search/ resolves, with a redirect to aems.tga.gov.au/daen-medicines-terms-condition/: the medicines search is now a Power BI report behind a terms-and-conditions gate on the AEMS portal and returns no data to a scripted request. The TGA site search at tga.gov.au answers a scripted request with a bot-verification interstitial rather than results. No DAEN figure is quoted in this monograph, because AIP has not read one. This is recorded as an open item: the search must be run by hand in a browser, for the ingredient Piper methysticum and for the trade names listed in section 12, and the report count, the date range and the reaction terms entered in section 20 before release. Nothing in this section rests on DAEN; the Australian case described in 15.3 comes from the published literature [5], and the Australian regulatory history in 15.9 from the instruments themselves [8,9] and from Mills and Bone [4].

TGA safety actions and alerts, re-checked in September 2026 (Medicine). The DAEN position is unchanged: the medicines search still redirects to a Power BI report behind a terms-and-conditions gate on the adverse-event portal and returns no data to a scripted request, and the TGA site search still answers with a bot-verification page rather than results. Both were tried again in September 2026. One thing did come back. The TGA publishes its safety alerts as a syndication feed, and that feed was retrieved in full in September 2026: 50 alerts spanning 04/12/2024 to 14/09/2026, none of which names kava or Piper methysticum. That is a real negative over that window and it is reported as exactly that much. The feed carries only the most recent 50 alerts, so it says nothing about earlier years, and the Medicines Safety Update archive is served through the same search application that refuses automated queries. The by-hand DAEN search stays an open item at 20.5, and the Australian regulatory actions in the table above are taken from the instruments in force and from the reference texts.

The TGA feed sweep widened, 28/09/2026 (Medicine). The syndication feeds that the regulator publishes were read again and the sweep was taken past safety alerts to every feed that could carry a kava action. Nine feeds returned, 50 items each: safety alerts (04/12/2024 to 14/09/2026), all alerts (16/12/2025 to 14/09/2026), market actions (02/12/2024 to 10/09/2026), safety updates, which is the Medicines Safety Update channel (19/07/2024 to 11/09/2026), notices (05/02/2024 to 18/09/2026), media releases (24/10/2025 to 18/09/2026), news (26/08/2025 to 21/09/2026), blog (13/02/2019 to 02/06/2026) and consultations (02/07/2024 to 11/09/2026). Not one item in any of them names kava or Piper methysticum. That matters more than a bare absence, because the safety-updates channel over the same window carries "Withania somnifera (Ashwagandha) and very rare risk of liver injury", published 07/07/2026: the regulator was publishing herbal liver-injury safety updates in this period and did not publish one on kava. The negative is still bounded by what the feeds hold – each carries only its most recent 50 items, so none of them reaches back before 2024 except the blog – and it is a negative about published safety actions, not about adverse-event reports. The Database of Adverse Event Notifications was tried again in September 2026 and is unchanged: the medicines search redirects to a business-intelligence report behind a terms-and-conditions gate on the adverse-event portal, and the site search answers a scripted request with a bot-verification page. No DAEN figure is quoted anywhere in this monograph, because AIP has not read one.

The sweep taken to the whole published feed directory, 28/09/2026 (Medicine). The nine feeds above are part of a larger set, and on the fourth Medicine pass the regulator’s own list of syndication feeds was read and every one of the 32 feeds on it was retrieved and searched: 1,405 published items in all, and not one names kava or Piper methysticum. The whole sweep was repeated at the ninth Medicine pass on 28/09/2026, from the feed directory again rather than from a stored list: the directory now publishes 33 feeds, they held 1,455 items between them, and not one names kava or Piper methysticum either. The feeds roll, which is why the count moves; what does not move is the answer. Four of those channels bear directly on this herb’s Australian status and none of them had been read before. Scheduling decisions (final) carries 50 decisions back to 25/11/2020 and scheduling decisions (interim) 41 back to 17/05/2017, so no interim or final scheduling decision on kava has been published in those windows and the Schedule 4 entry described in 15.9 stands unamended over them. Scheduling submissions carries 50 back to 11/04/2018, so nobody has applied in that window to move kava. Laboratory testing reports carries 25 back to 26/02/2019, so no kava product has been named in a published TGA laboratory test in that period, and compositional guidelines carries 50 back to 01/02/2017 with no kava guideline among them. The comparator holds and widens: over the same period the safety-updates channel published "Withania somnifera (Ashwagandha) and very rare risk of liver injury" on 07/07/2026 and "New warning on hepatotoxicity risk for Veoza (fezolinetant)" on 16/10/2025, so the regulator was publishing liver-injury updates and published none on kava. The limits are the same ones: each feed holds only its most recent items, the window is stated for each, and this is a negative about published regulatory actions and publications, not about adverse-event reports, which remain unread for want of the by-hand DAEN search.

A TGA alert on kava does exist, and the feeds could never have shown it, 28/09/2026 (Medicine). Every sweep above is bounded by how far back a syndication feed reaches, which for the safety-alert channel is 04/12/2024. Searching the regulator’s published safety alerts by name instead of by feed returns one that names this herb: Kava practitioner alert, published 26/02/2002, still carried on the TGA’s safety-alerts pages and retrieved and read in full in September 2026. It is the regulator’s own contemporaneous account and four things in it belong in this section. It alerts health professionals to hepatotoxicity reported overseas, and states that reports from Germany and Switzerland had linked kava extracts with 30 serious cases of hepatotoxicity, including six of liver failure: one death, four transplants and one patient under evaluation for transplantation when the report was published, with liver damage following between two weeks and two years of use, and causality confounded in a number of cases by concomitant drugs themselves linked to liver problems. It records that no Australian case of liver failure or related problem from a kava medicine had been reported to the TGA’s Adverse Drug Reaction Unit, and asks practitioners to review cases of unexplained hepatotoxicity for a kava link. It draws the solvent distinction that 13.5 and 15.3 still turn on, in the regulator’s own words: the implicated products are those containing chemically extracted constituents of kava, as opposed to kava in herbal form, the aqueous extract prepared by traditional methods. And it advises consumers to stop a kava medicine and obtain medical advice if symptoms appear, to get professional advice before starting one, and to avoid kava medicines altogether where there is a predisposing factor for liver damage such as a history of liver disease.

Two corrections to this section follow from it. The first is a date. The table above credits the August 2003 expert committee with the 250 mg/day kavalactone ceiling, on Mills and Bone’s account [4]. The February 2002 alert already states all three of the present limits as being in force: 125 mg of kavalactones per tablet or capsule, 3 g of dried rhizome per tea bag, and a maximum daily dose of not more than 250 mg of kavalactones. The per-dose and per-day ceilings therefore predate the 2003 determination by at least eighteen months, and what 2003 added was the restriction to water extracts, dispersions and whole or peeled rhizome. The second is a source. The contraindication in children under 12 and in pregnant and nursing women, carried at 15.7 on Mills and Bone’s report of the TGA position, is stated by the regulator directly in this alert, which records that kava medicines were permitted to be marketed to all segments of the population except those groups. That contraindication now rests on the regulator and not only on a reference text. The alert also records 84 kava medicines on the Register at that date, against the 24 entries naming the species found on 23/09/2026 (section 12).

AIP comment (Medicine). A 2002 alert is not a current safety action and nothing here should be read as one; it is twenty-four years old and its ARTG count, its case count and its request for reports have all been overtaken. Two things about it still matter. It is evidence that the regulator has published on kava and liver injury, which the feed sweeps above could not have established either way, and it should be read as correcting the impression those negatives might otherwise leave. And its statement that no Australian case had then been reported is the starting point of the sentence at 15.1 about the Australian case record, which the 2026 case report still on order will move again (20.5).

15.10 Referral

Kava carries a real, if rare, hepatic risk, a two-year rodent carcinogenicity finding in the liver of mice of both sexes, a real interaction profile with central depressants, and a regulatory ceiling that is easy to exceed with imported product bought online. Anyone considering kava should consult a degree-qualified herbalist with competencies and adequate training in pharmacognosy.