Article Text


Do herbs increase the risk of herb–drug interactions for patients with arthritis?
  1. M Thomsen1,
  2. M Schmidt2,
  3. L Vitetta3,
  4. A Sali3
  1. 1HerbResearch Pty Ltd, 29 Macfarlane St, Sth Hobart, 7004 Australia
  2. 2Herb Research, Germany
  3. 3Graduate School of Integrative Medicine, Swinburne University, Victoria, Australia
  1. Correspondence to:
    Mr M Thomsen
  1. W Holden4,
  2. J Joseph5,
  3. L Williamson6
  1. 4Nuffield Orthopaedic Centre, Windmill Road, Headington, Oxford, UK
  2. 5Nicosia Polyclinic, Nicosia, Cyprus
  3. 6Great Western Hospital, Swindon, UK

Statistics from

In a recent article by Holden and colleagues, echinacea and other herbs are identified as dangerous for patients taking arthritis drugs.1 Is there evidence to support this claim?

The article states that devil’s claw, ginkgo, and garlic may have antiplatelet or anticoagulant effects, potentially exacerbating the risk of gastrointestinal bleeding from non-steroidal anti-inflammatory drugs or corticosteroids. No direct evidence supports these claims. A review of the references indicates that only one is based on an actual case report, a rare and unusual event of excessive garlic ingestion causing a spontaneous spinal epidural haematoma, which would not be considered typical. The other references are experimental studies or theoretical discussions.

Another questionable claim is that arthritic patients are at an increased risk of hepatotoxicity from the use of echinacea. Referenced is an article by Miller2 stating: “If used beyond 8 weeks, echinacea could cause hepatotoxicity and therefore should not be used with other known hepatotoxic drugs, such as anabolic steroids, amiodarone, methotrexate, and ketoconazole”. This is unsubstantiated as no studies or case reports of hepatotoxicity caused by echinacea have been published. Furthermore, Miller comments that “echinacea lacks the 1,2-unsaturated necine ring associated with hepatotoxicity of pyrrolizidine alkaloids”, confirming that echinacea is not hepatotoxic. Moreover, Holden and colleagues fail to report that Fugh-Berman immediately contested the suggestion that echinacea might be hepatotoxic.3

A review of the safety of echinacea demonstrates that although there have been reports of allergic reactions, no deaths, dose dependent adverse effects, or overdoses have been reported. Contraindications are presently only theoretical and interactions have not been reported. The most convincing argument for safety comes from the ratio of reported serious adverse effects to the estimated number of courses of treatment. Fewer than 100 reported adverse effects were reported during over 10 million courses of treatment.4 Animal experiments have failed to demonstrate significant toxicity, with oral doses >15 g/kg and intravenous doses >5 g/kg failing to kill rodents. The researchers concluded that a lethal dose could not be found, and hence that the LD50 value (50% lethal dose) was incalculable.5

Holden concludes “11% of patients were taking remedies that might interact with conventional drugs”. This statement is misleading, as the concomitant intake of herbs and conventional drugs is not equivalent to observation of herb-drug interactions. Did any of the 238 patients actually experience any interactions? Did any physicians detect any problems?

In a recent review, Butterweck and colleagues demonstrated that beliefs about herb–drug interactions are mainly theoretical considerations, and not clinically observed facts.6 Herb–drug interactions do occur but, equally, common foods such as broccoli, grapefruit juice, alcohol, and cigarette smoking may cause interactions.

Holden and colleagues assume that with the UK drug legislation changes and the transfer of uncontrolled food supplements into registered drugs, thousands of adverse event reports will suddenly surface. Although underreporting is likely from the uncontrolled marketing of food supplements, the expectation of a flood of adverse effect reports seems exaggerated. In countries such as Switzerland and Germany the herbs in question have long been registered drugs, and subject to pharmacovigilance with little cause for concern.

There is justified concern about potential herb–drug interactions. Any reporting of adverse or beneficial effects must be subject to true scientific rigour, otherwise the true nature of potential adverse events may be misleading and underreported.


Authors’ reply

We thank Dr Thomsen and colleagues for their comments. We accept some points but must object to several issues. While we agree that reporting of adverse effects should be subjected to scientific rigour, publication in peer reviewed journals does not necessarily guarantee this rigour. A recent review of 193 randomised controlled trials of non-drug treatments for patients with arthritis, all published in journals with a high impact factor, found that fewer than half of the trials reported adverse effects associated with the treatments.1

We did not imply that licensing of herbal remedies would lead to an increase in the reporting of adverse effects. Legislation affecting the licensing of herbal remedies in Britain is likely to decrease rather than increase their use and so reduce the number of reported interactions.

Our study was not designed to detect potential interactions between herbal remedies and conventional drugs, merely to highlight the numbers of patients at risk of harmful interactions. It is dangerous to assume that a paucity of reports of interactions means that interactions are not occurring, and the purpose of our study was to emphasise the need for doctors to ask about herbal remedies in the history and for patients to ask for advice before using these preparations.

The lack of direct evidence of an interaction between herbal remedies that inhibit platelet aggregation, such as garlic and ginkgo biloba, and non-steroidal anti-inflammatory drugs causing an increased risk of bleeding—for example, gastrointestinal haemorrhage, obviously does not mean that such a risk can be ignored. Lack of space did not permit us to include other case reports of bleeding associated with herbal remedies2–9 including one that was fatal.10

The “yellow card” scheme in Britain is a method by which healthcare professionals can report suspected adverse drug reactions to the Medicines and Healthcare Products Regulatory Agency, an agency of the Department of Health. Although doctors in Britain are notoriously bad at reporting adverse drug reactions, Drug Analysis Prints on the UK government website— (accessed 12 August 2005)—lists 67 total suspected adverse reactions for ginkgo biloba up to June 2005, including four cases of intracerebral bleeding. Fifty six cases of suspected adverse effects due to echinacea, including abnormalities in liver function tests, have been reported. Clearly these substances are not as benign as animal experimentation might suggest, hence our conclusion that all types of drug taken should be rigorously recorded.

Rheumatology outpatients often take anti- inflammatory drugs for many years and are clearly at high risk of gastrointestinal complications. We found that 44% of our sample had been taking herbal or over the counter remedies during the past 6 months.11 Doctors rarely ask whether patients are taking herbal remedies12 and patients only infrequently divulge this fact. In one series, 92% of patients taking warfarin had not discussed their use of herbal remedies with a doctor.13 Given this lack of information it is highly likely that patients with inflammatory arthritis admitted, for example, with gastrointestinal bleeding attributed to anti-inflammatory drugs may well have also been taking a herbal remedy that might have contributed to this event and been missed in the history taking.

In conclusion, our study raises the alarm, proposes vigilance, and reminds us that we should be inquiring directly about herbal remedies when recording a patient’s history. This is especially important in patients with inflammatory arthritis, with their associated comorbidity and polypharmacy making adverse effects and drug interactions more likely. This will allow the true incidence of any adverse effects and interactions to be recorded accurately as the authors correctly state in their last paragraph.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.