Efficacy of Kava Extract for Treating Anxiety: Systematic Review and Meta-Analysis Pittler, Max H. MD; Ernst, Edzard MD, PhD, FRCPC (Edin) Department of Complementary Medicine, School of Postgraduate Medicine and Health Sciences, University of Exeter, Exeter, United Kingdom __________________________________________________________________________________________________________ Abstract Synthetic anxiolytic drugs are effective for treating anxiety, but they are burdened with adverse effects. Constraints on resources and time often render therapies such as psychologic interventions impracticable. Thus, an effective oral medication with few adverse effects would be a welcome addition to the therapeutic repertoire. This systematic review and meta-analysis was aimed at assessing the evidence for or against the efficacy of kava extract as a symptomatic treatment for anxiety. Systematic literature searches were performed in the computerized databases MEDLINE, EMBASE, BIOSIS, AMED, CISCOM, and the Cochrane Library (all from their respective inception to June 1998). The search terms used were kava, kawa, kavain, Piper methysticum, and Rauschpfeffer (German term for Piper methysticum). Experts on the subject were contacted to provide further information. There were no restrictions regarding the language of publication. Double-blind, randomized, placebo-controlled trials of oral kava extract for the treatment of anxiety were included. All publications were blinded before assessment by a person not involved in the study. Data were extracted in a standardized, predefined fashion independently by the two reviewers. The methodologic quality of all trials was assessed. Superiority of kava extract over placebo was suggested by all seven reviewed trials. The meta-analysis of three trials suggests a significant difference in the reduction of the total score on the Hamilton Rating Scale for Anxiety in favor of kava extract (weighted mean difference, 9.69; 95% confidence interval, 3.54-15.83). These data imply that kava extract is superior to placebo as a symptomatic treatment for anxiety. Therefore, kava extract is an herbal treatment option for anxiety that is worthy of consideration. ___________________________________________________________________________________________________________ Introduction Anxiety disorders are one of the most common types of psychiatric conditions.1 The U.S. National Comorbidity Survey, for instance, suggests a 1-year prevalence of anxiety disorders of 17% and a lifetime prevalence of almost 25%.2 In the majority of cases, anxiety is treated by general practitioners.3-5 Benzodiazepines are most commonly used, but are associated with serious adverse effects (AEs) such as dependence, sedation, and memory impairment.6-8 Constraints on resources and time may render other therapies such as psychologic interventions impracticable. Drug therapy will therefore most likely remain the primary treatment.9, 10 Recent data from the United States suggest that patients with anxiety frequently use alternative therapies.11 One plant-based therapeutic option for treating anxiety is kava. Kava is the beverage prepared from the rhizome of the oceanic kava plant (Piper methysticum Forst.).12 Throughout the South Pacific, extracts of kava have been used for recreational and medicinal purposes. Traditionally, it was used to treat a variety of ailments such as gonorrhea and to induce relaxation and sleep, but it was also used to counteract fatigue.13, 14 The rhizome of cultivated P. methysticum is used today as raw material for the production of kava extract.15 In 1998, it was among the top-selling herbs in the United States, with an annual turnover of approximately $8 million and a growth rate of 473%.16 Uncontrolled studies suggest that kava may be beneficial for the treatment of anxiety.17, 18 However, these studies cannot differentiate between nonspecific effects and specific therapeutic effects.19 Furthermore, nonrandomization may lead to a substantial overestimation of the effect size.20 This systematic review and meta-analysis draws from randomized, placebo-controlled, double-blind trials to assess the efficacy of kava extract for treating anxiety. Methods Systematic literature searches were performed to identify all randomized, controlled trials (RCTs) of kava for the treatment of anxiety. Computerized databases that were searched included MEDLINE, EMBASE, BIOSIS, AMED (British Library), CISCOM (Research Council for Complementary Medicine, London), and the Cochrane Library (all from their respective inception to June 1998). The search terms used were kava, kawa, kavain, Piper methysticum, and Rauschpfeffer (German term for P. methysticum). A manual search was performed using the bibliographies of studies and reviews identified by the computer search and by scanning our own files. In addition, experts on kava and leading manufacturers were contacted and were asked to contribute published and unpublished material. RCTs were included if they had been conducted in a double-blind, placebo-controlled manner. Trials not performed using kava monopreparations were excluded. No language restrictions were imposed. Article identifiers (names of authors, institutions, journals, and addresses) were removed by a person not involved in the study before the assessment. Data extraction and evaluation were performed independently by the two reviewers using standardized, predefined criteria. Articles in languages other than English or German were translated in-house. Three trials reported results that were suitable for meta-analysis. The common outcome measure in these studies was the total score on the Hamilton Rating Scale for Anxiety (HAM-A). The mean change compared with baseline was used to assess differences between the kava and placebo groups. Methodologic quality was assessed using the scoring system developed by Jadad and associates Other criteria for data extraction are shown in Table 2. Disagreements in the assessment of data were resolved by discussion, and consensus was reached in all such cases. A meta-analysis of data was performed using standard computer software (RevMan 3.01, Update Software Ltd., Oxford, England). The summary estimate of the treatment effect was calculated using the weighted means of the within-study treatment effects. Weighted mean differences and 95% confidence intervals were calculated using a random-effects model. Results Fourteen double-blind RCTs on kava were retrieved.22-35 Seven studies were excluded because they either were duplicate publications 22, 23 or were performed using kavain.24-27 One study 28 was excluded because it was conducted in combination with benzodiazepine treatment. Seven double-blind RCTs met the aforementioned criteria and were included in this systematic review and meta-analysis.29-35 Six trials scored at least 3 of 5 points on the scoring system to assess methodologic quality (Table 3). Table 2 summarizes the key data of all trials ranked in a hierarchical order according to their methodologic quality. Three trials 29-31 assessed a common outcome measure and could therefore be included in the meta-analysis (Fig. 1). These studies reported the HAM-A total score as their main outcome measure and included patients only if the total score on the HAM-A scale at base-line was 19 or greater. Additionally, 51% of the patients in trials included in the meta-analysis received a diagnosis according to DSM-III-R criteria.31 All trials revealed weighted mean differences that favored kava extract over placebo. Their 95% confidence intervals did not overlap the line of zero effect size, indicating significant difference. The meta-analysis of these data shows a significant difference in the reduction of the HAM-A total score from baseline in favor of kava extract compared with placebo (weighted mean difference, 9.69; 95% confidence interval, 3.54-15.83). Four studies 32-35 did not report data suitable for statistical pooling. All of these trials demonstrated a significant reduction of anxiety in patients treated with kava extract (Table 2). Three trials 33-35 reported significant inter-group differences in favor of kava extract, and one 32 stated beneficial effects compared with baseline findings. The medication used differed in terms of kavapyrone content per daily dose (240 mg,33 210 mg,29-31 150 mg,35 and 60 mg 32,34). This reflects differences in the total composition of the extracts. The meta-analysis was performed using trials that were homogenous in terms of drug quality (Table 2). Five of seven trials reported AEs experienced by patients receiving kava extract.30-32, 34, 35 Stomach complaints, restlessness, drowsiness, tremor, headache, and tiredness were reported as AEs (Table 2). Two trials comprising 31% of the studied patients reported the absence of AEs while taking kava extract. Discussion This systematic review and meta-analysis of double-blind RCTs suggests that, compared with placebo, kava extract is an efficacious symptomatic treatment for anxiety. This finding corroborates the results of previous analyses on less extensive data.14, 36-38 Constraints on resources and physicians' time, the preference of oral drug treatment,39 and the inclination of patients with anxiety to use alternative therapies 11 may render kava extract an attractive, phytotherapeutic option. The meta-analysis of three trials suggests a significant reduction in the HAM-A total score of approximately 10 points in favor of kava extract. Similarly, results of trials that were not suitable for a quantitative analysis suggest significant beneficial effects in favor of kava (Table 2). The pooling of trial data in meta-analyses can be criticized for several reasons.40, 41 In particular, results from large RCTs do not always agree with results obtained from meta-analyses of relatively small sets of data.41 However, in this meta-analysis, the findings are supported by the results of the largest RCT in the database.31 Furthermore, there is a reasonable overall agreement between RCTs, which indicates validity of the overall result. Nevertheless, several caveats exist. One relates to citation tracking, which cannot provide a guarantee for the completeness of the identified references. A recent investigation indicates that European journals may be underrepresented on MEDLINE.42 This is of particular interest regarding herbal medicine. Apart from one trial,33 all reviewed studies originated from Europe and were published by authors from Germany. In addition, there is a tendency for negative trials to remain unpublished, 43 particularly in journals of complementary or alternative medicine in which studies with positive findings may be overrepresented.44 Our search strategy included American and European data-bases as well as manual searching and the contact with experts and manufacturers. Nevertheless, it is conceivable that some trials may have been missed and that our overall result may be overly optimistic. However, we have no concrete evidence that this type of publication bias has, in fact, influenced our overall finding. To exclude our own possible bias, article identifiers were removed before the evaluation. A standard scoring system was used in an attempt to quantify the likelihood of bias inherent in the studies 21 (Table 1). Six of seven trials scored at least 3 of 5 points for methodologic quality (Table 3). Nevertheless, none of the studies were completely flawless. In particular, the sample size in most trials was small and lacked a proper power calculation. Double-blinding and randomization procedures should have been described in the published reports. Only four studies 29, 30, 32, 34 detailed their randomization procedures. Inadequate sequence generation in randomized studies has been suggested to yield larger estimates of treatment effects.20 Thus, the uncertain quality of the method to generate a randomized sequence may be another source of bias. The pharmacologic actions of kava extract supports the notion that it may be effective for treating anxiety. In vitro studies suggest that kavapyrones, the pharmacologically active components, act on the central nervous system. Kavapyrones are believed to mediate effects on [gamma]-aminobutyric acid-A receptors, particularly in the hippocampus and amygdala complex.45, 46 Animal experiments using electroencephalographic (EEG) measurements have demonstrated effects of kava extract on the corpus amygdala.47 Central nervous effects of kavain 48 and kava extract 49 have also been demonstrated in studies on human volunteers by using EEG measurements. New synthetic drugs that are currently being assessed for the treatment of anxiety are 5-hydroxytryptamine receptor agonists and the substance-P inhibitor MK-869.39 Other pharmacologic options include antidepressants and benzodiazepines. The latter may, however, cause AEs such as sedation, amnesia, and the development of tolerance, and it may carry an increased risk of road-traffic accidents.50-52 Although comparative studies suggest the absence of a significant difference between benzodiazepines and kavain or kava extract,53, 54 the latter seems relatively safe. This is supported by all of the above studies 29-35 and by the fact that two studies, representing 31% of the studied patients, did not report any AEs in patients treated with kava extract (Table 2). Two postmarketing surveillance studies,55, 56 each involving more than 3000 patients, found AEs in 2.3% and 1.5% of patients during treatment with kava extract given in a daily dose equivalent to 120 to 240 mg and 105 mg of kavapyrones, respectively. AEs reported most frequently were gastrointestinal complaints, allergic skin reactions, headache, and photosensitivity. In conclusion, the evidence presented here suggests that kava extract is relatively safe and more efficacious than placebo in the symptomatic treatment of anxiety. Important caveats exist, which prevent firm conclusions. The findings warrant further and more rigorous investigations of the risk-benefit relation of kava. _____________________________________________________________________________________________________________ Table 1. Scoring system used to measure the likelihood of bias. 21 Table 2. Double-blind, randomized, placebo-controlled trials of kava for anxiety Table 3. Methodologic quality of individual trials Fig. 1. Meta-analysis (random-effects model) of double-blind, randomized, placebo-controlled trials of kava extract versus placebo with the total score on the HAM-A as the common endpoint. The weighted mean difference (linear scale) is given with its 95% confidence interval. The vertical line (weighted mean difference = 0) represents the absence of difference between kava extract and placebo. The weighted mean difference to the left of the vertical line (smaller than zero) is in favor of placebo and to the right is in favor of kava extract.