Kava

Kava is one of the more complicated supplements on this site. It has a real efficacy signal for short-term relaxation - cleaner than most herbal anxiolytics - but the safety story has gone through three rewrites in 25 years and is still not fully settled. Read both sections before deciding.

The efficacy case rests on three pieces of evidence. The Pittler and Ernst Cochrane review (2003) pooled 11 trials with 645 participants and found kava extract significantly superior to placebo on the Hamilton Anxiety Scale, with adverse events described as mild, transient, and infrequent. The Witte 2005 meta-analysis of six WS 1490 (acetonic root extract) trials returned an odds ratio of 3.3 for clinical response in non-psychotic anxiety. The Volz 1997 25-week multicenter trial of WS 1490 in 101 outpatients with DSM-III-R anxiety disorders showed significant HAMA superiority over placebo starting at week 8, with no tolerance or dependence signal. So the European acetonic-extract literature is genuinely positive.

Sarris's Australian aqueous-extract program then added two trials with noble-cultivar water extracts. KADSS (2009, n=60 crossover) showed a ~10-point anxiety reduction on aqueous kava vs minimal placebo effect with no clinical liver toxicity at 250mg kavalactones/day for 3 weeks. The 2013 follow-up RCT (n=75, 120-240mg kavalactones/day for 6 weeks) showed a moderate effect on diagnosed GAD (Cohen d=0.62) with no LFT signal. Both reinforced that aqueous noble-root preparations look efficacious and well-tolerated short-term.

Then the larger 2020 K-GAD trial (Sarris et al, n=171, 16 weeks of aqueous noble extract delivering up to 240mg kavalactones/day in diagnosed GAD) was flatly negative. Remission was 17.4% on kava vs 23.8% on placebo. Kava produced more memory complaints, tremor, and LFT abnormalities (though no clinical liver injury). This is the largest, longest, and most rigorous kava-for-GAD trial ever run, and it failed. So the honest read is: short-term, situational, mild-to-moderate anxious feelings - probably real benefit. Long-term treatment of diagnosed GAD - no.

Now the safety side. The FDA issued a consumer advisory in March 2002 after liver-injury case reports out of Germany, Switzerland, the UK, and the US. Several European countries banned kava in 2002; most have since lifted those bans. The 2007 WHO expert review concluded that traditional aqueous extracts of noble cultivars, as used in the Pacific for centuries, carry very low hepatotoxicity risk, while the western case-cluster involved predominantly acetonic and ethanolic extracts, non-noble Tudei cultivars, aerial plant parts (which contain higher concentrations of pipermethystine and flavokavain B, two suspected hepatotoxins), and patients on co-medications. That "Pacific paradox" framing dominated for a decade.

Teschke et al revisited the question in 2011-2012 and complicated the picture: aqueous extracts in New Caledonia, Australia, Germany, and the US have produced occasional hepatotoxicity case reports too, and the team proposed contamination - aflatoxins, ochratoxin A, mold from poorly stored raw material - as a more plausible primary driver than extract chemistry per se. The current best-guess synthesis: extraction method matters less than raw material quality (noble vs Tudei, root vs aerial, fresh and well-handled vs moldy), individual susceptibility (alcohol, hepatotoxic co-medications, pre-existing liver disease), and dose-duration. Idiosyncratic liver injury at low rates remains possible even with clean material.

Practical bottom line: if you want occasional evening calm or to take the edge off a stressful week, a noble-cultivar root extract or traditional grog is a reasonable, evidence-backed choice for short-term use. Buy from a brand that discloses cultivar (noble, not Tudei) and plant part (root only, not aerial). Limit to 4-8 weeks of regular use. Do not combine with alcohol or hepatotoxic drugs. Do not take it if you have liver disease. Do not drive on it. And do not expect it to treat clinical GAD - the largest trial that tested that hypothesis came up empty.