Abstract
Background and Aim: Little is known about the use of herbal medicines by people living with cancer in the UK. This systematic review aimed to estimate the prevalence of herbal medicine use by this group, the characteristics of users, factors motivating use, and attitudes towards herbal remedies.
Design and Methods: Fifteen electronic databases were searched. People who were research-active in the field were contacted and asked about further published or unpublished work. All studies identified as relevant to the purpose of the review were assessed. Searches were not restricted by publication type or date.
Results: Of 1288 unique references identified, 11 met the eligibility criteria. Studies were excluded where research had been conducted outside the UK; where information on herbal medicine use was not differentiated from that relating to complementary and alternative therapies more broadly, and where neither prevalence of use nor information on user characteristics was included. Prevalence estimates ranged from 3.1 to 24.9%. Most studies did not obtain information specifically on herbal medicines and only one examined the characteristics and motivations of users of herbal medicines as distinct from complementary and alternative therapies in general.
Conclusions: The high degree of heterogeneity of methodology, sample selection and characteristics, and research design resulted in a wide range of estimates of prevalence. Well-designed research is needed to define the evidence base about the herbal medicines taken by people with cancer in the UK, the reasons for use, knowledge about possible effects and potential risks, and where people seek information.
Background
The use of herbal remedies and ‘natural’ medications taken by mouth is widespread and becoming increasingly common in the UK.1–3 In some circumstances, herbal medicines can present significant risks, particularly to those undergoing treatment for other illnesses such as cancer. For example, by inducing the activity of metabolizing enzymes, St John's Wort (Hypericum perforatum) can reduce the time that the anti-cancer treatment Imatinib remains in the body by 44%.4 Black cohosh (Cimifuga racemosa) is contra-indicated for women with oestrogen-receptor positive breast cancer due to uncertainty about possible oestrogenic activity,5 as are Dong Quai (Angelica sinensis) and ginseng (Eleutherococcus senticosus), which have been shown to stimulate cell growth in a human breast cancer cell line.6 Herbal medicines may cause health problems unrelated to cancer; black cohosh has been associated, albeit rarely, with serious liver problems.5 Despite the potential, both to enhance the effects of some conventional medications and to reduce the effectiveness of others, people nevertheless tend to believe that ‘natural’ remedies are by definition harmless.7,8
People with cancer are generally advised to tell the professionals treating them if they are taking any form of medication, including herbal medicines and supplements.9 However, many do not disclose herbal medicine use,2,10 and healthcare professionals often feel unable to give informed advice about the efficacy or safety of herbal remedies.11 The most common source of advice on herbal medications is not the medical profession, but friends, relatives, support and self-help groups.2,12
Systematic reviews typically treat complementary and alternative therapies as a common entity, rarely provide information specifically related to herbal medicines, and usually take an international approach.13–15 The use of information from other countries to inform UK policy or practice regarding herbal medicines is problematic as cultures and practices vary. Even across Europe, there is heterogeneity; a single medicinal plant may be variously defined as a food, a functional food, a dietary supplement or a herbal medicine, depending on the regulations applying to different medicines in each country. A World Health Organization (WHO) report on the use and regulation of traditional medicines indicated that in 50 out of 192 responding member states, herbal remedies are sold on prescription.16 In contrast, in the UK, herbal medicine is seen as outside mainstream medical practice.1 Cultural approaches may also differ; both China and India have highly developed systems of traditional medicine based largely on herbs, and the beliefs and principles guiding these (as well as the types of herbal remedies used) differ from the biomedical model prevalent in the UK.17
Little is known about the use of herbal medicines among people living with cancer in the UK. This review aimed to estimate the prevalence of herbal medicine use by cancer patients in the UK, the characteristics of users, the factors motivating use, knowledge of, and attitudes towards herbal remedies among users.
Methods
There is no agreed nomenclature in the field of complementary and alternative therapies. For clarity, standardized terminology is used throughout this review. ‘Complementary and alternative therapies’ describes all modalities, including mind/body techniques such as reiki, biologically based therapies such as vitamins and food supplements, and physical therapies such as acupuncture. ‘Complementary and alternative medicines’ includes substances taken orally as part of complementary or alternative treatment. ‘Herbal remedies’ includes all herbal preparations, whether taken orally, applied topically or by any other method, whereas ‘herbal medicine(s)’ includes only herbal preparations taken orally, and conforming to the definition outlined in European Directive 2004/24/EC (article 1, para.30) as ‘any medicinal product, exclusively containing as active ingredients one or more herbal substances or one or more herbal preparations, or one or more such herbal substances in combination with one or more such herbal preparations’.18
Search strategy
The search strategy aimed to identify all studies examining the prevalence of, beliefs about, knowledge of and attitudes to herbal medicine use by cancer patients in the UK. Scoping searches indicated that this information was often subsumed into studies examining complementary and alternative therapies as a whole; the search strategy therefore remained broad (Table 1).
Search strategy and electronic databases searched
| Text words | Index terms |
|---|---|
| Complementary medicine | CANCER |
| Complementary treatment | HERB |
| Alternative medicine | COMPLEMENTARY THERAPY |
| Alternative treatment | ALTERNATIVE THERAPY |
| England | UK |
| Wales | UNITED KINGDOM |
| Scotland | |
| Ireland | |
| Database | |
| PubMed | |
| Medline (1950–November 2008) | |
| Embase (1980–2009) | |
| Health Management Information Consortium (HMIC) | |
| PsycINFO (1967–March 2009) | |
| Allied and Complementary Medicine (AMED) | |
| OpenSIGLE | |
| Cumulative Index to Nursing and Allied Health Literature (CINAHL) | |
| SportDiscus | |
| Web of Knowledge | |
| National Library of Health Complementary and Alternative Medicine Specialist Library | |
| British Library Direct | |
| Research Council for Complementary Medicine (RCCM) | |
| British Nursing Index (BNI) | |
| Health Business Elite | |
| Text words | Index terms |
|---|---|
| Complementary medicine | CANCER |
| Complementary treatment | HERB |
| Alternative medicine | COMPLEMENTARY THERAPY |
| Alternative treatment | ALTERNATIVE THERAPY |
| England | UK |
| Wales | UNITED KINGDOM |
| Scotland | |
| Ireland | |
| Database | |
| PubMed | |
| Medline (1950–November 2008) | |
| Embase (1980–2009) | |
| Health Management Information Consortium (HMIC) | |
| PsycINFO (1967–March 2009) | |
| Allied and Complementary Medicine (AMED) | |
| OpenSIGLE | |
| Cumulative Index to Nursing and Allied Health Literature (CINAHL) | |
| SportDiscus | |
| Web of Knowledge | |
| National Library of Health Complementary and Alternative Medicine Specialist Library | |
| British Library Direct | |
| Research Council for Complementary Medicine (RCCM) | |
| British Nursing Index (BNI) | |
| Health Business Elite | |
To maximize the possibility of identifying all relevant publications, 15 electronic databases were searched, and citation searches undertaken where relevant publications were identified. In addition, people known to be research active in the field were contacted to ascertain their knowledge of further published, unpublished or ongoing studies. Bibliographies of all studies identified as relevant to the purpose of the review were examined. No restrictions were placed on searches either by publication type or date. Searches were limited to those written in English.
Eligibility, data extraction and quality assessment
All studies providing data on herbal medicine use by cancer patients in the UK were included. Those where data on herbal medicines could not be isolated from that relating to other therapies or preparations were excluded, as were study designs involving case studies with fewer than five participants or reporting expert opinion rather than primary research findings. Herbal medicines made from more than one plant were included, as were ginger (Zingiber officinale), garlic (Allium sativum) and similar plant-derived substances if used for medicinal purposes and taken as supplements rather than as flavouring agents. Studies which focused exclusively on homeopathic preparations, minerals and vitamins were excluded.
Citations were downloaded into Reference Manager (version 10) and duplicates removed. Two independent reviewers screened titles and abstracts for eligibility. Studies that could confidently be judged as ineligible on the basis of title/abstract alone were excluded. Full papers were obtained when a decision on inclusion/exclusion could not be reached through assessment of title and abstract alone. Data were extracted from eligible papers using a pre-defined data extraction sheet designed to record study size, design, methodology, patient demographics, prevalence of herbal medicine use, and the reasons for use. One reviewer undertook the data extraction, which was checked against the original paper by a second reviewer. Disagreements were resolved by discussion or through arbitration by a third reviewer. Each eligible paper was read by two reviewers and assigned a quality assessment score, based on criteria adapted from Critical Appraisal Skills Programme (CASP) tools for quality assessment (Table 2).19 Studies received one point for each of the eight criteria they were judged to meet successfully, giving a potential maximum score of eight points for study quality.
Quality assessment criteria and score for each included studya
| Assessment criteriab | Werneke et al. (2004)25 | McCann and Newell (2006)20 | Scott et al. (2005)29 | Molassiotis and Cubbin (2004)28 | Wilkinson et al. (2008)23 | Gami et al. (2003)21 | Catt et al. (2006)26 | Hunter et al. (2004)27 | Rees et al. (2000)22 | Downer et al. (1994)24 | Corner et al. (2006)7 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Were the aims of the research clearly stated? | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 |
| Was the chosen methodology appropriate to address the aims? | 1 | 0 | 0 | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 1 |
| Are the important characteristics of the study population specified? | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| Is the response rate above 70%? Or if not, can inferences be made about study population representativeness? | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 0 | 0 |
| Is a definition of the disease and treatment (i.e. cancer; herbal medicines) stated? | 1 | 0 | 1 | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 0 |
| Is age and gender specific prevalence reported? | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 |
| Was data analysis sufficiently rigorous? | 0 | 0 | 0 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| Is there a clear statement of findings? | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| Total score by study | 5 | 2 | 4 | 3 | 6 | 7 | 4 | 5 | 7 | 6 | 6 |
| Assessment criteriab | Werneke et al. (2004)25 | McCann and Newell (2006)20 | Scott et al. (2005)29 | Molassiotis and Cubbin (2004)28 | Wilkinson et al. (2008)23 | Gami et al. (2003)21 | Catt et al. (2006)26 | Hunter et al. (2004)27 | Rees et al. (2000)22 | Downer et al. (1994)24 | Corner et al. (2006)7 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Were the aims of the research clearly stated? | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 |
| Was the chosen methodology appropriate to address the aims? | 1 | 0 | 0 | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 1 |
| Are the important characteristics of the study population specified? | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| Is the response rate above 70%? Or if not, can inferences be made about study population representativeness? | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 0 | 0 |
| Is a definition of the disease and treatment (i.e. cancer; herbal medicines) stated? | 1 | 0 | 1 | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 0 |
| Is age and gender specific prevalence reported? | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 |
| Was data analysis sufficiently rigorous? | 0 | 0 | 0 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| Is there a clear statement of findings? | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| Total score by study | 5 | 2 | 4 | 3 | 6 | 7 | 4 | 5 | 7 | 6 | 6 |
aAdapted from Critical Appraisal Skills Programme (CASP) appraisal tools for quality assessment of qualitative studies.19
bA score of ‘1’ indicates that the paper was judged to meet a given criterion. A score of ‘0’ indicates that it did not. Maximum score = 8.
Prevalence figures for herbal medicine use were calculated for each eligible study. Included studies were separated according to focus on adult or paediatric participants, and subdivided further by cancer site and type. Prevalence estimates (weighted to study sample size) and 95% confidence intervals (CIs) were calculated for each of these cancer types. Literature searches were completed in March 2009.
Results
The bibliographic database search identified 1653 citations, of which 365 were duplicates (Figure 1). Each of the 1288 unique references was screened for inclusion on the basis of title/abstract, with 1239 papers excluded as outside the eligibility criteria. The remaining 49 references were subject to full text evaluation, with 38 excluded at this stage, because the research had been conducted outside the UK, data on herbal medicine use were not separately reported, and neither prevalence of herbal medicine use nor the characteristics of users were reported. Eleven studies met the eligibility criteria.
Inclusion and exclusion of identified citations.
Inclusion and exclusion of identified citations.
Quality assessment
The quality of the included studies varied considerably, with scores ranging from two out of a possible eight20 to seven out of eight.21,22 Only five studies had a quality score of six or more.7,21–24 Across all studies, the quality criteria most poorly adhered to related to the representativeness of the study population, the reporting of prevalence, and the clarity of terminology describing herbal medicines and their use.
Characteristics of included studies
Scope
None of the included studies focused exclusively on herbal medicines. All but two covered the use of complementary and alternative therapies in general.25,26 These two studies examined the use of non-prescription medicines and vitamins, minerals and other supplements, respectively. Sample sizes ranged from n = 2520 to n = 714,22 with an average of 267 participants (Table 3). Three studies included only women with breast cancer22,26,27; one studied men with prostate cancer,23 and one focused on patients with pelvic cancers.21 Two studies obtained information from the parents of paediatric cancer patients,20,28 whilst the remaining six included adult patients with a range of cancers.
Characteristics of included studies
| Authors (year of publication) | Study setting; recruitment; sample size | Study design; response rate/ sampling frame | Type of cancer; time after diagnosis | Number using herbal medicines (%) | Reasons for herbal medicine use | Quality Assess- ment score (out of 8) | Comments |
|---|---|---|---|---|---|---|---|
| Catt et al. (2006)26 | UK; 6 oncology outpatient clinics; n = 208 | Standardized interviews; questionnaire as part of interview; convenience sample | Breast; >2 years post-diagnosis | 7 to 18 (3.4 to 8.7) | Not stated | 4 | Prevalence based on usage of Echinacea (n = 4), Black cohosh (n = 4), Chinese herbs (n = 3), Garlic (n = 7). Unclear if these are unique users or users of multiple herbal medicines, so prevalence calculated as a range |
| Corner et al. (2006)7 | South England; radiotherapy, chemotherapy and out-patient clinics at two centres; n = 304 | Postal questionnaire; narrative interviews; quota sampling | Any; >6 months post-diagnosis | 24 (7.9) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 6 | No individual herbs identified; prevalence calculated from figures on herbal medicine and TCM use; interviewees purposively sampled to reflect patterns of complementary and alternative therapies use in survey sample |
| Downer et al. (1994)24 | London, England; oncology and radiotherapy departments at two hospitals; n = 415 | Postal questionnaire; semi-structured interviews; 69% response | Any; >3 months post-diagnosis | 13 (3.1) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 6 | No individual herbs identified |
| Gami et al. (2003)21 | London, England; oncology follow-up clinic; n = 107 | Questionnaire; 89.2% response | Pelvic cancers; >1 year post-diagnosis | 3 (2.8) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 7 | No individual herbs specified; prevalence figure based on reported visits to herbalist |
| Hunter et al. (2004)27 | London, England; oncology clinic patient database at one hospital; n = 113 | Questionnaire (cross-sectional) and semi-structured interviews; 56.6% response | Breast; any stage post-diagnosis | 16 (14.2) | Not stated | 5 | No individual herbs specified; Participants were breast cancer patients aged 35–65, without relapse, prescribed adjuvant tamoxifen in past 5 years; 38.9% (n = 44) of survey respondents were interviewed |
| McCann and Newell (2006)20 | Leeds, England; specialist clinics at teaching hospitals; n = 25 | Questionnaire-based structured interview; quota sampling | Any; any stage post-diagnosis | 5 (20) | Not stated | 2 | No individual herb figures specified for cancer; 100 participants in total (n = 25 for each): healthy, cerebral palsy, inflammatory bowel disease, cancer |
| Molassiotis and Cubbin (2004)28 | UK; oncology unit patient records at one hospital; n = 49 | Questionnaire survey (cross-sectional); 51% response | Any; any stage post-diagnosis | 2 (4.1) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 3 | No individual herbs specified; questionnaires sent to parents of paediatric cancer patients |
| Rees et al. (2000)22 | Thames NHS region, England; Thames cancer registry data; n = 714 | Questionnaire survey, semi-structured interviews; 74% response | Breast; diagnosis between 1990–96 | 23 (3.2) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 7 | No individual herbs specified; Reported prevalence figure derivation unclear—paper discusses both use and visits to herbalists/complementary therapy practitioners in last 12 months and since cancer diagnosis; semi-structured interviews with 80 survey respondents, survey figures adjusted if survey and interview findings differed |
| Scott et al. (2005)29 | Three centres in Scotland and England; oncology outpatient clinics and hospice; n = 127 | Questionnaire survey; convenience sampling | Any; any stage post-diagnosis | 7 (5.5) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 4 | No individual herbs specified; reported prevalence of herbal use (amongst those using complementary and alternative therapies) varied: 14.3% (n = 5) used before diagnosis, 20% (n = 7) had used since diagnosis, 8.6% (n = 3) were using at time of survey |
| Werneke et al. (2004)25 | London, England; outpatient department at cancer centre; n = 382 | Questionnaire survey (cross-sectional); 63.6% response | Any; any stage post-diagnosis | 95 (24.9) | Improve health to fight cancer; boost immune system; reduce pain or hormonal disturbances; counter side-effects of conventional treatments | 5 | 5.3% of participants (n = 17) took only herbal medicines, 24.5% (n = 78) took both herbal medicines and other supplements. Prevalence calculated from total (n = 95); reasons for use based on herbal medicines and supplements |
| Wilkinson et al. (2008)23 | London, England; hospital patient administration system; n = 294 | Postal questionnaire (cross-sectional survey); 73% response | Prostate; any stage post-diagnosis | 13 (4.4) | Not stated | 6 | Prevalence figures quoted here based on usage of Saw Palmetto only; this was the only herb available for participants to indicate on a pre-coded list |
| Authors (year of publication) | Study setting; recruitment; sample size | Study design; response rate/ sampling frame | Type of cancer; time after diagnosis | Number using herbal medicines (%) | Reasons for herbal medicine use | Quality Assess- ment score (out of 8) | Comments |
|---|---|---|---|---|---|---|---|
| Catt et al. (2006)26 | UK; 6 oncology outpatient clinics; n = 208 | Standardized interviews; questionnaire as part of interview; convenience sample | Breast; >2 years post-diagnosis | 7 to 18 (3.4 to 8.7) | Not stated | 4 | Prevalence based on usage of Echinacea (n = 4), Black cohosh (n = 4), Chinese herbs (n = 3), Garlic (n = 7). Unclear if these are unique users or users of multiple herbal medicines, so prevalence calculated as a range |
| Corner et al. (2006)7 | South England; radiotherapy, chemotherapy and out-patient clinics at two centres; n = 304 | Postal questionnaire; narrative interviews; quota sampling | Any; >6 months post-diagnosis | 24 (7.9) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 6 | No individual herbs identified; prevalence calculated from figures on herbal medicine and TCM use; interviewees purposively sampled to reflect patterns of complementary and alternative therapies use in survey sample |
| Downer et al. (1994)24 | London, England; oncology and radiotherapy departments at two hospitals; n = 415 | Postal questionnaire; semi-structured interviews; 69% response | Any; >3 months post-diagnosis | 13 (3.1) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 6 | No individual herbs identified |
| Gami et al. (2003)21 | London, England; oncology follow-up clinic; n = 107 | Questionnaire; 89.2% response | Pelvic cancers; >1 year post-diagnosis | 3 (2.8) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 7 | No individual herbs specified; prevalence figure based on reported visits to herbalist |
| Hunter et al. (2004)27 | London, England; oncology clinic patient database at one hospital; n = 113 | Questionnaire (cross-sectional) and semi-structured interviews; 56.6% response | Breast; any stage post-diagnosis | 16 (14.2) | Not stated | 5 | No individual herbs specified; Participants were breast cancer patients aged 35–65, without relapse, prescribed adjuvant tamoxifen in past 5 years; 38.9% (n = 44) of survey respondents were interviewed |
| McCann and Newell (2006)20 | Leeds, England; specialist clinics at teaching hospitals; n = 25 | Questionnaire-based structured interview; quota sampling | Any; any stage post-diagnosis | 5 (20) | Not stated | 2 | No individual herb figures specified for cancer; 100 participants in total (n = 25 for each): healthy, cerebral palsy, inflammatory bowel disease, cancer |
| Molassiotis and Cubbin (2004)28 | UK; oncology unit patient records at one hospital; n = 49 | Questionnaire survey (cross-sectional); 51% response | Any; any stage post-diagnosis | 2 (4.1) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 3 | No individual herbs specified; questionnaires sent to parents of paediatric cancer patients |
| Rees et al. (2000)22 | Thames NHS region, England; Thames cancer registry data; n = 714 | Questionnaire survey, semi-structured interviews; 74% response | Breast; diagnosis between 1990–96 | 23 (3.2) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 7 | No individual herbs specified; Reported prevalence figure derivation unclear—paper discusses both use and visits to herbalists/complementary therapy practitioners in last 12 months and since cancer diagnosis; semi-structured interviews with 80 survey respondents, survey figures adjusted if survey and interview findings differed |
| Scott et al. (2005)29 | Three centres in Scotland and England; oncology outpatient clinics and hospice; n = 127 | Questionnaire survey; convenience sampling | Any; any stage post-diagnosis | 7 (5.5) | Reasons for use of herbal medicines could not be isolated from those stated for complementary and alternative therapies in general | 4 | No individual herbs specified; reported prevalence of herbal use (amongst those using complementary and alternative therapies) varied: 14.3% (n = 5) used before diagnosis, 20% (n = 7) had used since diagnosis, 8.6% (n = 3) were using at time of survey |
| Werneke et al. (2004)25 | London, England; outpatient department at cancer centre; n = 382 | Questionnaire survey (cross-sectional); 63.6% response | Any; any stage post-diagnosis | 95 (24.9) | Improve health to fight cancer; boost immune system; reduce pain or hormonal disturbances; counter side-effects of conventional treatments | 5 | 5.3% of participants (n = 17) took only herbal medicines, 24.5% (n = 78) took both herbal medicines and other supplements. Prevalence calculated from total (n = 95); reasons for use based on herbal medicines and supplements |
| Wilkinson et al. (2008)23 | London, England; hospital patient administration system; n = 294 | Postal questionnaire (cross-sectional survey); 73% response | Prostate; any stage post-diagnosis | 13 (4.4) | Not stated | 6 | Prevalence figures quoted here based on usage of Saw Palmetto only; this was the only herb available for participants to indicate on a pre-coded list |
Study participants were at various times since diagnosis; one study included patients described as currently receiving treatment,29 while another included women up to seven years from diagnosis.24 Four studies were conducted with patients who ranged from three months to two years from diagnosis20,21,24,26; one included patients up to more than five years after diagnosis23 and one covered women prescribed tamoxifen for breast cancer within the previous five years.27 Two did not specify time since diagnosis or treatment status,25,28 and only two indicated the disease stage and existence or otherwise of metastases.7,23
Study settings
All but one study22 were conducted with patients identified from outpatient or follow-up clinic records, five of them recruiting patients from a single centre.21,23,25,27,28 Only one eligible study included a centre outside England (Scotland)29 and seven were based in the South East of England.20,21,23–26,28 Three did not specify location more precisely than the country.26,28,29
Sampling and recruitment
A range of sampling and recruitment strategies was employed, although three studies20,25,29 outlined their strategies in insufficient detail to allow consideration of the appropriateness of their approach. In two cases, patients attending clinics were approached personally to participate in research by interview,21,26 five sampled patients from hospital records or cancer registries and posted questionnaires directly to eligible participants,22,23,25,26,28 whereas two identified potential participants from hospital records, screened for eligibility by post, then recruited from those consenting to participate.7,24
Data collection
Research methods varied, with studies using either surveys,20,21,23,26,28,29 or mixed methods research designs.20,22,24,27 Amongst those using surveys, postal questionnaires were the preferred data collection method7,22–24,28 in two cases accompanied by a smaller sample of interviews conducted to validate answers given to survey questions, or to allow qualitative exploration of reasons for using their chosen therapies.7,24 Studies also differed in the manner they elicited information about use of herbal medicines. Some used pre-coded lists of commonly used therapies referring variously to ‘herbal medicine(s)’7,22 ‘herbalism’,24 ‘herbal therapy or other plant extracts’.29 In other studies,23,26 it was unclear whether questionnaires used pre-coded lists, open-ended questions or, as in one case,25 a combination of these.
Prevalence of herbal medicine use among UK cancer patients
Estimates of the prevalence of herbal medicine use in eligible studies ranged from 3.124 to 21.8% among adults,25 and from 4.128 to 20%20 in the paediatric studies. Studies varied in both the definition and specification of herbal medicines. In eight studies, prevalence figures were supplied for ‘herbal medicines’ in general.7,20–22,24,27–29 In the remaining three, specific herbs were named, either via open questions or pre-coded lists of herbs.23,25,26 Two papers asked about visits to a herbal practitioner, which were made by 2.821 and 2.4%29 of participants, respectively (Table 4).
Prevalence of herbal medicine use in included studies
| Studies | Sample size | Number using herbal medicines since diagnosis | Prevalence of herbal medicine use (%) | Weighted prevalencea (95% CI) | |
|---|---|---|---|---|---|
| Adult | |||||
| Pre-2000 | Downer et al. (1994)24 | 415 | 13 | 3.1 | 3.1 (1.7 to 5.3) |
| Mixed tumour sites | Werneke et al. (2004)25 | 382 | 95 | 24.9 | |
| Scott et al. (2005)29 | 127 | 7 | 5.5 | ||
| Corner et al. (2006)7 | 304 | 24 | 7.9 | 15.5 (13.1 to 18.2) | |
| Breast | Rees et al. (2000)22 | 714 | 23 | 3.2 | |
| Hunter et al. (2004)27 | 113 | 16 | 14.2 | ||
| Catt et al. (2006)b,26 | 208 | 7 to 18 | 3.4 to 8.7 | 4.4 to 5.5 (3.3 to 7.1) | |
| Prostate | Wilkinson et al. (2008)c,23 | 294 | 13 | 4.4 | 4.4 (2.4 to 7.4) |
| Visits to herbalist | Gami et al. (2003)21 | 107 | 3 | 2.8 | 2.8 (0.6 to 8.0) |
| Paediatric | Molassiotis and Cubbin (2004)28 | 49 | 2 | 4.1 | |
| McCann and Newell (2006)20 | 25 | 5 | 20 | 9.5 (3.9 to 18.5) | |
| Studies | Sample size | Number using herbal medicines since diagnosis | Prevalence of herbal medicine use (%) | Weighted prevalencea (95% CI) | |
|---|---|---|---|---|---|
| Adult | |||||
| Pre-2000 | Downer et al. (1994)24 | 415 | 13 | 3.1 | 3.1 (1.7 to 5.3) |
| Mixed tumour sites | Werneke et al. (2004)25 | 382 | 95 | 24.9 | |
| Scott et al. (2005)29 | 127 | 7 | 5.5 | ||
| Corner et al. (2006)7 | 304 | 24 | 7.9 | 15.5 (13.1 to 18.2) | |
| Breast | Rees et al. (2000)22 | 714 | 23 | 3.2 | |
| Hunter et al. (2004)27 | 113 | 16 | 14.2 | ||
| Catt et al. (2006)b,26 | 208 | 7 to 18 | 3.4 to 8.7 | 4.4 to 5.5 (3.3 to 7.1) | |
| Prostate | Wilkinson et al. (2008)c,23 | 294 | 13 | 4.4 | 4.4 (2.4 to 7.4) |
| Visits to herbalist | Gami et al. (2003)21 | 107 | 3 | 2.8 | 2.8 (0.6 to 8.0) |
| Paediatric | Molassiotis and Cubbin (2004)28 | 49 | 2 | 4.1 | |
| McCann and Newell (2006)20 | 25 | 5 | 20 | 9.5 (3.9 to 18.5) | |
aWeighted to study sample size: studies with larger samples contribute more to the weighted prevalence estimate than those with smaller sample sizes.
bUnclear whether stated herbal medicine usage was based on unique responses or several study participants each using multiple herbal medicines.
cPrevalence calculations based on stated usage of Saw Palmetto; only herbal medicine option available on a pre-coded list.
Weighted prevalence figures for herbal medicine use were calculated on the basis of cancer type. Studies focusing on mixed tumour sites7,25,29 had a higher prevalence than those that recruited only breast cancer patients22,26,27 (16.8%; 95% CI 14.2 to 19.7 vs. 4.4 to 5.5%; 95% CI 3.3 to 7.1). Weighted prevalence for the paediatric studies was 9.5% (95% CI 3.9 to 18.5).20,28
In the studies identifying specific herbs,23,25,26 15 different herbal medicines were named, most within a single study25 (Table 5).
Prevalence of named herbs in included studies
| Studies | Sample size | Named herbs mentioned | Number of people using | % using | |
|---|---|---|---|---|---|
| Adult | |||||
| Pre-2000 Mixed tumour sites | Downer et al. (1994)24 | 415 | – | – | |
| Werneke et al. (2004)25 | 382 | Echinacea | 35 | 9.2 | |
| Evening primrose oil | 33 | 8.6 | |||
| Gingko biloba | 16 | 4.2 | |||
| Milk thistle | 11 | 2.9 | |||
| Essiac | 10 | 2.6 | |||
| Chinese remedies (exc. green tea) | 7 | 1.8 | |||
| Garlic | 7 | 1.8 | |||
| St John's wort | 6 | 1.6 | |||
| Arnica | 5 | 1.3 | |||
| Valerian | 5 | 1.3 | |||
| Kava kava | 3 | 0.8 | |||
| Siberian ginseng | 3 | 0.8 | |||
| Passion flower | 2 | 0.5 | |||
| Aloe vera | 2 | 0.5 | |||
| Indian remedies inc. | 2 | 0.5 | |||
| Turmeric & ginger | |||||
| Panax ginseng | 2 | 0.5 | |||
| Wild yam | 2 | 0.5 | |||
| Golden seal | 1 | 0.3 | |||
| Mistletoe | 1 | 0.3 | |||
| Slippery elm | 1 | 0.3 | |||
| Scott et al. (2005)29 | 127 | 7 (herbal therapy or other plant extracts) | – | ||
| 3 (herbalist consulted, not clear if subsumed in the 7 above) | |||||
| Corner et al. (2006)7 | 304 | 22 (herbal medicine) | – | ||
| 2 (Chinese herbal medicine) | |||||
| Breast | Rees et al. (2000)22 | 714 | 23 | – | |
| Hunter et al. (2004)27 | 113 | 16 | |||
| Catt et al. (2006)26 | 208 | Garlic | 7 | 3.4 | |
| Echinacea | 4 | 1.9 | |||
| Black cohosh | 4 | 1.9 | |||
| Chinese herbs | 3 | 1.4 | |||
| Prostate | Wilkinson et al. (2008)23 | 294 | Saw palmetto | 13 | 4.4 |
| Visits to herbalist | Gami et al. (2003)21 | 107 | 3 | – | |
| Paediatric | |||||
| Molassiotis and Cubbin (2004)28 | 49 | 2 | – | ||
| McCann and Newell (2006)20 | 25 | 5 | – | ||
| Studies | Sample size | Named herbs mentioned | Number of people using | % using | |
|---|---|---|---|---|---|
| Adult | |||||
| Pre-2000 Mixed tumour sites | Downer et al. (1994)24 | 415 | – | – | |
| Werneke et al. (2004)25 | 382 | Echinacea | 35 | 9.2 | |
| Evening primrose oil | 33 | 8.6 | |||
| Gingko biloba | 16 | 4.2 | |||
| Milk thistle | 11 | 2.9 | |||
| Essiac | 10 | 2.6 | |||
| Chinese remedies (exc. green tea) | 7 | 1.8 | |||
| Garlic | 7 | 1.8 | |||
| St John's wort | 6 | 1.6 | |||
| Arnica | 5 | 1.3 | |||
| Valerian | 5 | 1.3 | |||
| Kava kava | 3 | 0.8 | |||
| Siberian ginseng | 3 | 0.8 | |||
| Passion flower | 2 | 0.5 | |||
| Aloe vera | 2 | 0.5 | |||
| Indian remedies inc. | 2 | 0.5 | |||
| Turmeric & ginger | |||||
| Panax ginseng | 2 | 0.5 | |||
| Wild yam | 2 | 0.5 | |||
| Golden seal | 1 | 0.3 | |||
| Mistletoe | 1 | 0.3 | |||
| Slippery elm | 1 | 0.3 | |||
| Scott et al. (2005)29 | 127 | 7 (herbal therapy or other plant extracts) | – | ||
| 3 (herbalist consulted, not clear if subsumed in the 7 above) | |||||
| Corner et al. (2006)7 | 304 | 22 (herbal medicine) | – | ||
| 2 (Chinese herbal medicine) | |||||
| Breast | Rees et al. (2000)22 | 714 | 23 | – | |
| Hunter et al. (2004)27 | 113 | 16 | |||
| Catt et al. (2006)26 | 208 | Garlic | 7 | 3.4 | |
| Echinacea | 4 | 1.9 | |||
| Black cohosh | 4 | 1.9 | |||
| Chinese herbs | 3 | 1.4 | |||
| Prostate | Wilkinson et al. (2008)23 | 294 | Saw palmetto | 13 | 4.4 |
| Visits to herbalist | Gami et al. (2003)21 | 107 | 3 | – | |
| Paediatric | |||||
| Molassiotis and Cubbin (2004)28 | 49 | 2 | – | ||
| McCann and Newell (2006)20 | 25 | 5 | – | ||
Herbal medicine users
Characteristics
As in studies amongst the general population,2,3 some studies included here found that cancer patients using complementary and alternative therapies as broadly defined were more likely to be younger, female, middle class and highly educated7,24,26 but none examined the socio-demographic characteristics of herbal medicine users separately.
Motives, expectations, knowledge and information
Only one study25 examined the motives and expectations of people using herbal medicines and nutritional or vitamin supplements (rather than complementary therapies in general), but did not differentiate between the three. This study reported that patients gave non-specific reasons for use, such as improving health or to fight cancer. Almost a third of those taking herbal medicines, nutritional or vitamin supplements were not sure about the purpose of the substance they were taking. No studies examined participants’ information sources regarding herbal medicines, although two23,28 reported friends and family as important sources of information on complementary and alternative therapies generally.
Discussion
Prevalence estimation
Estimates of herbal medicine use ranged from 3.1 to 24.9%, with 9 of 11 included studies indicating a prevalence of less than 10%. This is much lower than population-based studies of herbal medicine purchasing rates among the general public (i.e. not restricted to cancer patients). Thomas2 found that 19.8% (95% CI 18.3 to 21.3) of people in twelve English health authorities (n = 2669) had bought over-the-counter (OTC) herbal medicines in the previous twelve months, and 31.4% (95% CI 29.6 to 33.2) had ever bought them. More recently, Ipsos MORI (n = 2032) found that 26% (95% CI 24.0 to 28.0) had bought OTC herbal medicines in the past two years, and 35% (95% CI 33.0 to 37.0) had ever bought them.3 Whilst the ‘purchase’ of herbal medicines is not synonymous with their subsequent ‘use’, it may be assumed that the majority who buy herbal medicines go on to use them. All but one25 of the studies cited here suggest that rates of purchase (and probable use) may be lower for cancer patients than for the general population.
Several factors could account for estimates of herbal medicine use among cancer patients being lower than for the general population. It has been suggested that when people are diagnosed with cancer, they stop using herbal medicines.7 In contrast, another study has shown that while the use of complementary and alternative therapies in general decreased after cancer diagnosis, this was not the case in relation to herbal medicines (3.9% before diagnosis, 5.5% after).29
Second, cancer patients may not disclose herbal medicine use. People have been found to be reluctant to admit use of complementary and alternative therapies to healthcare professionals.8,11 Most studies included in this review were conducted in association with clinics involved in patient treatment, which may have affected people's willingness to participate if they had used complementary and alternative therapies, or to admit to their use. Nevertheless, the study indicating the highest prevalence (24.9%) appears to have been conducted in a dedicated cancer treatment hospital,25 whilst participants in a study reporting one of the lowest prevalence figures (3.2%) were recruited through cancer registry sampling and participated via self-completed postal questionnaires.22 It may be that healthy populations are more prepared to acknowledge herbal self-medication than cancer patients, irrespective of the source of the questioning.
Finally, question framing may influence response. Open questions about herbal medicine use may underestimate true prevalence, as some respondents may not consider the supplements or preparations that they are taking to be herbal medicines. Closed questions using pre-coded lists must include all potential herbal medicines, or they may lead to prevalence being underestimated. Response bias (i.e. people using complementary and alternative therapies being more likely to participate in a survey) may mean prevalence is overstated.30 The only study in this review which directly asked participants about herbal medicine use showed the highest prevalence, and the figure closest to studies conducted in the general population.25 The variation in prevalence estimates appears to be as much due to heterogeneous approaches to sampling, questionnaire design and data collection as to patient characteristics. The study of prostatic cancer23 reported only the use of saw palmetto (Serenoa repens). However, studies in the USA31 and Australia32 have indicated a wider range of herbal medicines used by men with prostate cancer, including ginseng (Panax ginseng), cat's claw (Uncaria tomentosa), garlic (Allium sativum) and gingko (Ginkgo biloba). Thus, the estimate of the prevalence of herbal medicine use among patients with prostate cancer relates to a single herb only and may underestimate true usage.
Specific herbs and their prevalence
Information about specific herbs taken by cancer patients, and the frequency of use remains unclear. Of the named herbs, the most commonly mentioned were Echinacea, followed by evening primrose oil (Oenothera biennis), Gingko biloba and milk thistle (Silybum marianum).25 This is consistent with other studies among the general population. One study conducted in northwest England33 listed evening primrose oil, Gingko biloba, St John's wort (Hypericum perforatum) and garlic (Allium sativum) as the most popular; preoperative patients used garlic, ginseng (Panax ginseng), Gingko biloba and St John's wort.34
Generalizability
Most eligible studies involved single clinics and often included small numbers of participants. A variety of recruitment and data collection methods were used and details of the sampling frame were rarely reported. The heterogeneity of the included studies in terms of sample size, study design, cancer type and time since diagnosis, and the terminology used to define herbal medicines means that the overall picture of herbal medicine use amongst cancer patients remains unclear. These factors confound attempts to generalize from the available literature to cancer patients in the UK as a whole, and internationally.
Terminology and differentiation
Inconsistent use of terminology and definitions in the field of complementary and alternative therapies2,8 hampers attempts to assess prevalence and interpret the significance of research findings. There is no agreed set of definitions; amongst the studies included in this review, the terms therapy and medicine were often used interchangeably.22,23 Substances taken orally were variously described as medicines,25 remedies or treatments,26 and categories such as ‘oils’ failed to distinguish between herbal and fish oils.26 Establishing a robust and well-defined terminology in relation to all aspects of complementary and alternative medicines/therapies would be an important step towards improving the validity and comparability of studies in the field.
Attitudes, motivations, expectations and knowledge
Declared motivations for using complementary and alternative therapies amongst cancer patients participating in studies cited in this review include the hope of a cure, remission or preventing disease spread22–24; reducing treatment side effects7,22,28,29; boosting the immune system7,25; reducing stress/aiding relaxation;7,27 and improving quality of life.23 None of the included studies attempted to differentiate between users and non-users, or to disaggregate participants according to therapy type. Furthermore, people may use herbal medicines for reasons unrelated to cancer (e.g. co-morbidity) but none of the included studies reported on this. Little attention has been paid to people's understanding of, or knowledge about, the complementary and alternative therapies they use. For some therapies, some imprecision may be an integral part of the therapeutic effect, but where there is potential risk, it is crucial that accurate information is made available.
Limitations of this review
This review was restricted to studies investigating herbal medicine use amongst cancer patients in the UK, as use is known to vary between cultures and countries, with many herbs named in the context of traditional herbal medicine not well known within the western tradition.35 The use of herbal medicines by cancer patients has also been reported as more widespread in some other countries than in the UK. For example, among breast cancer patients in China, 86.4% used Chinese herbal medicines, almost all of them as supplementary to conventional treatments.36 Due to differing cultural/usage traditions and national regulatory environments for herbal medicines, the findings of this review have limited international generalizability. Furthermore, although weighted prevalence estimates for herbal medicine use amongst cancer patients were calculated, the heterogeneity of eligible studies, and the varied sampling frames used to collect data, mean that these estimates must be interpreted with caution.
Conclusions
There were 289 107 new diagnoses of cancer in 200537; at the weighted best estimate of prevalence of herbal medicine use across the studies included in this review (7.6 to 8.0%; 95% CI 6.6 to 9.1), this would equate to between 19 081 and 26 308 newly diagnosed patients each year taking herbal medicines. Given that the prevalence of cancer is much greater than the incidence, it is likely that a substantial number of people are taking herbal medicines. With such a high number of potential users, a better evidence base for understanding all aspects of herbal medicine use by cancer patients is clearly needed. We found that despite the considerable literature on the use of complementary and alternative therapies, there was little information on the use of herbal medicines specifically. Studies that did include them were often of poor quality. Data interpretation was further complicated by the lack of consistent terminology. Meaningful comparisons within and across populations, and the generation of pooled prevalence estimates, would become feasible if consistent terminology was adopted.
It is generally believed that many complementary therapies, including herbal medicines, have little potential for harm, but this is not true of all herbal medicines. Furthermore, some that are normally safe may adversely affect cancer treatments. Given the potential risks, there is a need to better understand which herbal medicines patients with cancer use, what they know and believe about them, and what they need to know to use them safely.38 With this knowledge, information resources can be created for healthcare professionals and patients that will enable patients to participate safely in their own care. This systematic review has confirmed the lack of available evidence to address these important issues. Well-designed research is still needed.
Funding
This work was supported by Macmillan Cancer Support.
Conflict of interest: None declared.
