A 10-point scoring system for interaction probability was devised. This scale has not been validated but we provide it only as a guide for assessing whether reports of herb–drug interactions contain adequately reliable information.

Three clinical studies have examined interactions of psychoactive herbs with alcohol. One study found that mixed valepotriates from Valeriana officinalis reduced the adverse effect of alcohol on concentration [18]. Another demonstrated that Panax ginseng enhanced alcohol clearance in humans [19]. A clinical study of an infusion of Piper methysticum found that kava potentiates impairment when combined with alcohol [20].

The effect of Catha edulis on the pharmacokinetics of single-dose ampicillin and amoxicillin has been studied. The bioavailability of the latter was reduced significantly during a khat-chewing session, but no effect was noted with amoxicillin [21].

Licorice (Glycyrrhiza glabra) contains glycyrrhizin, which inhibits 11β-dehydrogenase, 5α-reductase, and 5β-reductase. Glycyrrhizin is metabolized largely to glycyrrhetinic acid (an even more potent inhibitor of 5α-, 5β-reductase and 11β-dehydrogenase). Liquorice has been shown to increase plasma concentrations of prednisolone [22, 23]. It also potentiated the cutaneous vasoconstrictor response of hydrocortisone [24]. Licorice is a common herb in Chinese and Japanese herbal mixtures. The effect of several such mixtures on prednisolone concentrations has been tested. Both Sho-saiko-To and Xiao Chai Hu Tang decreased the plasma AUC of prednisolone [25] whereas Saiboku-To increased it. Sairei-To did not affect it [26]. Both Sho-saiko-To and Rikkunshi-to were found not to affect the pharmacokinetics of a single oral dose of ofloxacin [26].

St John's wort affects the clearance of many drugs, including cyclosporin, antidepressants (predominantly SSRIs), digoxin, indinavir, and phenprocoumon. The underlying mechanism appears to be multifactorial. There is evidence of a strong interaction with P-glycoprotein (PgP), an ATP-dependent drug efflux transporter known to pump drugs out of the cell membrane, thus decreasing intracellular concentrations [27]. St John's wort increases expression of duodenal PgP/MDR1 [28]. The herb also induces the activity of an important form of cytochrome P450, CYP3A4. However, results of studies of the effect of St John's wort on CYP3A4 are conflicting. While three enzyme marker studies indicated a potent inducing effect of St John's wort (300 mg three times daily × 14 days) on CYP3A4 activity [29, 30], two others found no effect (both 300 mg three times daily, one for 3 days [31], the other for 8 days [32]) on CYP2D6 or 3A4 activities. It is possible that the negative trials were simply too short in duration. Alternatively, differences in quality of the HMPs used might explain the discrepancy. One study examining the effect of St John's wort on CYP1A2 activity (using a caffeine/dextromethorphan probe) found that it (300 mg three times daily for 8 days) had no effect on 17 DMX/caffeine ratios [33], indicating that this HMP has a low potential for drug interactions involving CYP 1A2.

Cases of serotonin syndrome could arise if St John's wort increased serotonin levels. However, it cannot be considered an SSRI, because in vitro, St John's wort inhibits the uptake of serotonin, noradrenaline and dopamine only at high concentrations with IC₅₀s of 2.4, 4.5, and 0.9 µg ml⁻¹, respectively [34] which are unlikely to be achieved with oral dosing. However, enough cases of interactions with SSRIs have been reported that a serotonergic (or serotonin-amplifying) effect of St John's wort could occur.

There is reasonable documentation of interactions between coumarin anticoagulants and St John's wort, danshen, dong quai, ginseng, and ginkgo [35]. Most of these case reports are probably not true interactions but result from additive anticoagulant effects. Dong quai contains coumarins, and would be expected to augment the effects of a coumarin-derived anticoagulant. Ginkgo and garlic interfere with platelet function, and have been associated with bleeding even in the absence of warfarin or other anticoagulant treatment. Danshen also interferes with platelet function but appears to decrease the elimination of warfarin at least in rats [37]. In a recent case report, the INR of a patient taking warfarin who drank 0.5–1 gallon of green tea (Camellia sinensis) was seen to decrease [36].

Inadequate reporting makes it difficult to determine whether a herb–drug interaction has occurred. Authors should be required to document all relevant information (see above). The adverse event should be clearly described, alternative explanations should be explored and a rechallenge should be considered. The HMP should be analysed to ascertain the contents of the product. Even well-documented case reports can only serve as a critical early warning system.