Description of the condition
Menopause is a significant event in the lives of most women, as it marks the end of a woman's natural reproductive life. The perimenopausal and early postmenopausal years are typically characterised by falling levels of endogenous oestrogen, which can give rise to vasomotor symptoms that are severe and disruptive, particularly in the early and late menopausal transition and in early postmenopause, as categorised by the STRAW (STages of Reproductive Aging Workshop) criteria (Harlow 2012). These vasomotor symptoms include hot flushes (also known as 'hot flashes'), sweating and sleep disturbances.
Hot flushes are described as sudden feelings of heat in the face, neck and chest (WHO 1996). Hot flushes are frequently accompanied by skin flushing and perspiration, followed by a chill as core body temperature drops (Freedman 2001; Kronenberg 1990). Flushes vary in frequency, duration and severity and may be spontaneous and unpredictable (Freedman 1995). Hot flushes that occur during the night are typically referred to as night sweats. Flushes and night sweats are events of concern in themselves because they can disrupt sleep patterns and alter daily activities, which can lead to fatigue and decreased quality of life (Ayers 2013; NAMS 2004). Hot flushes are thought to result from both the brain's response to diminished hormones and hormonal fluctuations that occur during the menopausal transition, which leads to instability of thermoregulatory mechanisms (that regulate temperature) in the hypothalamus (Deecher 2007; Freedman 2001; Kronenberg 1987).
The prevalence of vasomotor symptoms varies with ethnicity. Flushes are less common among East Asian women (median 16%) than among American and European women (median 55%) (Freeman 2007). Up to 40% of Western women are affected severely enough to seek medical help (Freeman 2007; Gold 2006). An Australian prospective study with 13-year follow-up reported that the mean duration of troublesome vasomotor symptoms was 5.5 years (Col 2009). A study of more than 10,000 British women 54 to 65 years of age found that more than half (54%) were currently experiencing vasomotor symptoms (averaging 34 hot flushes or night sweats per week), which were problematic in 40% of cases and were fairly stable across the age range (Hunter 2012). Although hot flushes are reported as more prevalent and intense in the perimenopausal and early postmenopausal years, they continue to be important in up to 14.6% of women in their sixties and in 8.6% of women in their seventies (Roussouw 2007).
Description of the intervention
Most therapies designed to combat menopausal vasomotor symptoms aim to supplement levels of circulating oestrogen (Sikon 2004). The treatment of choice has traditionally been hormone therapy (HT), but, despite its effectiveness for symptom reduction, a marked and global decline has occurred in the prescription and use of HT because of concerns about long-term use, particularly worry about increased risk of chronic diseases (Bestul 2004; Haas 2004; Travers 2006). Although the combination of HT and unopposed oestrogen therapy was previously prescribed to prevent the onset of cardiovascular events as women grew older, a report of the Women's Health Initiative (WHI) trial, in 2002, indicated that the risks of this treatment outweighed the benefits (Roussouw 2002). Combined therapy was linked with increased risk of breast cancer, stroke, thromboembolism (blood clots), gallbladder disease and dementia. Unopposed oestrogen therapy increased the risk of stroke, thromboembolism and gallbladder disease, and other studies reported an increase in the incidence of breast cancer (Beral 2003). Data now available from 11 years of follow-up provided by WHI show that risks are influenced by the age of the woman, the time since menopause and whether the HT was combined or consisted of oestrogen only (NAMS 2012). Contraindications to HT include a family history or increased risk of cardiovascular disease, blood clotting disorders, venous thromboembolism or certain hormone-sensitive cancers (Anderson 2003; Grady 2000). Some women report adverse effects when taking HT (Bakken 2004; Bjorn 1999); potential side effects include breast tenderness, bloating and genital bleeding. Regulatory bodies around the world are now advocating that HT should be prescribed only in the smallest dose and for the shortest possible time (Europ Med Ag 2006; UK MHRA 2007).
Potential health risks associated with HT and further uncertainty surrounding actual benefits to be gained from it have caused many women to seek non-medical alternatives (Bair 2005; Newton 2002). 'Natural' therapies appear to be very popular among women; a survey of 866 women 45 to 65 years of age reported that 61% agreed or strongly agreed with the statement that natural approaches are better than hormone pills for menopausal symptoms (Newton 2002). In a national survey on women's use of complementary alternative medicine (CAM), more than 50% of CAM users indicated that such use was consistent with their beliefs, and 55% said that they wanted a natural approach to treatment (Chao 2006).
However, sufficient research on the risks and benefits of these approaches is lacking. A survey of women seen at a university clinic reported that 70% of women taking dietary supplements did not inform their doctors about their use, and only 4% had received information about such supplements from a healthcare provider (Mahady 2003). In a national survey, when women using CAM for menopausal symptoms consulted a doctor, their disclosure rate (of CAM) was much higher, with only 36% of women reporting that they did not disclose their self treatment with CAM to their doctors (Wade 2008).
Therapies based on phytoestrogens are among the most common alternatives to HT. Phytoestrogens are nonsteroidal plant compounds of diverse structure that are found in many fruits, vegetables and grains (Knight 1996; Thompson 1991). The most common types of phytoestrogens are coumestans, lignans and isoflavones. These compounds structurally resemble oestradiol (E2) and are shown to have weak oestrogenic activity (Makela 1994; Setchell 1998). When ingested in relatively large quantities, dietary phytoestrogens have been shown to have significant biological effects in several animal species (Adlercreutz 1995) and in humans (Wilcox 1990). In humans, they appear to have both oestrogenic and anti-oestrogenic effects, depending on the concentrations of circulating endogenous oestrogens and oestrogen receptors (Bolego 2003).
Isoflavones are among the most oestrogenically potent phytoestrogens; the major dietary isoflavones, genistein and daidzein, are found almost exclusively in legumes such as soy, chick peas, lentils and beans (Cassidy 1993). Urinary excretion of equol, a weak oestrogen, in humans eating soy-supplemented diets can greatly exceed the concentration of urinary endogenous oestrogens; this enhances the plausibility of human physiological health effects (Setchell 1984). Other classes of phytoestrogens—lignans and prenylated flavonoids—also have potent oestrogenic activity but are not as well studied (Adlercreutz 1987; Milligan 1999). Soy, a particularly abundant source of isoflavones, is a staple ingredient in the traditional Asian diet. It is postulated that high intake of soy among Asian women may account for lower rates of some menopausal symptoms in this group. Asian populations, such as those in Japan, Taiwan and Korea, are estimated to consume 20 to 150 mg per day of isoflavones, with a mean of about 40 mg from tofu (soy bean curd) and miso (soy bean paste). Soy includes such products as tofu, miso, aburage (fried thin tofu) and fermented or boiled soy beans. Further evidence that soy might be beneficial is suggested by a cohort study of Japanese women (Nagata 2001), which found a significant inverse association between frequency of flushes and higher levels of soy consumption. However, the findings of this study are contradicted by data from a cross-sectional study, which found that women who frequently consumed soy products were not less likely to report hot flushes or night sweats than women who never consumed soy products (Sievert 2007). Thus it is not clear whether frequent soy consumption explains the lower rate of hot flushes among different ethnic groups. Red clover (Trifolium pratense), another source of isoflavones, contains compounds that are metabolised to genistein and daidzein after consumption. The most studied red clover product is Promensil.
Potential adverse effects of phytoestrogens have included deficits in sexual behaviour in rats and impaired fertility in livestock (Bennetts 1946). No specific examples of toxicity among humans have been noted in countries in which soy is consumed regularly (Setchell 1997). It is generally considered difficult for humans to consume the quantity of isoflavones from natural soy foods needed to reach toxicological levels that induce pathological effects, as recorded in animals.
How the intervention might work
No clear explanation is known for how phytoestrogens might work in reducing hot flushes among perimenopausal and postmenopausal women.
It has been suggested that phytoestrogens act as selective oestrogen receptor modulators (SERMs), exerting anti-oestrogenic effects in the high-oestrogen environment of premenopause and oestrogenic effects in the low-oestrogen environment of postmenopause, where they act as weak agonists by stimulating oestrogen receptors (Seibel 2003). Phytoestrogens appear to show greater affinity for the oestrogen receptor beta (ERβ) than for the classical oestrogen receptor alpha (ERα). As a result, they preferentially express oestrogenic effects in the central nervous system, blood vessels, bone and skin without causing stimulation of the breast or uterus (Kuiper 1997). Thus, phytoestrogens may reduce vasomotor symptoms through their action on the vascular system without causing unwanted oestrogenic effects on other body systems.
Why it is important to do this review
Current use of phytoestrogen products among perimenopausal and postmenopausal women with vasomotor symptoms is high; an American cross-sectional analysis of more than 2,000 women (Study of Women's Health Across the Nation (SWAN)) reported that 11% of women with vasomotor symptoms used flaxseed products and 19% used soy products (Gold 2007). Several reviews have examined the efficacy of phytoestogen products in alleviating menopausal symptoms, but most have found no benefit or a very slight reduction in the frequency of daily hot flushes compared with placebo. Government agencies and healthcare organisations have also scrutinised the effects of phytoestrogens, particularly isoflavones (AFSSA 2005; Com Tox 2003). The North American Menopause Society (NAMS) position statement on the treatment of menopause-associated vasomotor symptoms suggests that women should consider isoflavone supplementation if their menopausal flushing does not respond to other interventions (NAMS 2004; NAMS 2011). However, NAMS acknowledges that the evidence base for this recommendation is poor.
Thus, the aim of this review is to synthesise all available evidence on the efficacy, safety and acceptability of products containing phytoestrogens to assist women with vasomotor menopausal symptoms to reduce their symptoms by making good evidence-based treatment decisions.
To determine the efficacy, safety and acceptability of food products, extracts and dietary supplements containing high levels of phytoestrogens when compared with no treatment, placebo or hormone therapy for the amelioration of vasomotor menopausal symptoms (such as hot flushes and night sweats) in perimenopausal and postmenopausal women.
Criteria for considering studies for this review
Types of studies
All randomised controlled comparisons of food products, extracts or dietary supplements containing high levels of phytoestrogens (e.g. at least 30 mg/d of isoflavones) versus placebo, HT, no treatment or products containing low levels of phytoestrogens for the alleviation of vasomotor menopausal symptoms.
Types of participants
- Perimenopausal women, defined as women in the 45- to 55-year age range, who have menstruated within the past 12 months and are seeking treatment for menopausal vasomotor symptoms
- Postmenopausal women, defined as women who are older than 45 years of age, who have not menstruated for longer than 12 months and are seeking treatment for menopausal symptoms
Women experiencing spontaneous or surgical menopause (bilateral oophorectomy (removal of both ovaries)) were eligible. Trials were eligible only when most women had vasomotor symptoms.
Source of recruitment
Any healthcare setting or the community
- Intercurrent major disease
- Previous HT (hormone therapy) within one month of commencement of the study or an oestrogen implant within the past year
- Women with breast cancer or a history of breast cancer
- Women with no or inconsequential vasomotor symptoms at baseline
Types of interventions
All food products or dietary supplements containing high levels of phytoestrogens (> 30 mg/d of isoflavones, > 100 µg 8-prenylnaringenin or > 10,000 µg total lignans) versus placebo, hormone therapy, no treatment or food products with low levels of phytoestrogens given as perimenopausal or postmenopausal therapy for the alleviation of vasomotor menopausal symptoms for a period of at least 12 weeks. Studies in which phytoestrogens were combined with other therapies were excluded.
Types of outcome measures
- Change in vasomotor menopausal symptom scores (without distinction between types of vasomotor symptoms)
- Change in frequency of individual vasomotor symptoms or severity of individual vasomotor symptom scores (e.g. hot flushes and night sweats)
- Incidence of vasomotor symptoms (hot flushes and night sweats) after treatment
Studies were included if they measured vasomotor symptoms on a subscale of a compendium score, for example, Greene Score, Kupperman Index, Nordin Score, MacLennan Score or any other general menopausal symptom score that derives numerical results from a combination of vasomotor menopausal symptoms.
In addition, studies were included that measured individual vasomotor symptoms, for example, severity or frequency, or both, of hot flushes and night sweats (evaluated subjectively by participants, usually in daily diaries).
- Stimulation of the endometrium (endometrial thickness, rates of atrophic endometrium)
- Vaginal stimulation (pH, maturation value)
- Adverse events
- Acceptability of therapy (withdrawal due to adverse events or satisfaction rates)
Studies were included if they measured specific safety outcomes, such as measures of physiological oestrogenicity of the endometrium and vagina. Other possible safety outcomes could be measured that are related to the effects of oestrogen action on other tissue and organs, but these will be assessed in future reviews if evidence of a beneficial effect on symptoms is noted.
Search methods for identification of studies
The Trials Search Co-ordinator designed the search strategy for use with the electronic databases. The complete search strategies are listed in the Appendices of this review.
The Trials Search Co-ordinator of the Menstrual Disorders and Subfertility Group (MDSG) searched for all published and unpublished randomised controlled trials (RCTs) of phytoestrogens for vasomotor symptoms, with no language restriction, using the following electronic databases.
- MEDLINE (see Appendix 1).
- EMBASE (see Appendix 2).
- PsycInfo (see Appendix 3).
- AMED (see Appendix 4).
- Cochrane Central Register of Controlled Trials (see Appendix 5).
- MDSG Specialised Register of Controlled Trials (see Appendix 6).
The principal author of the review (AL) searched the following trial registers and websites.
- Trial registers for ongoing and registered trials: http://www.controlled-trials.com
- Citation indexes: http://scientific.thomson.com/products/sci
- Conference abstracts in the Web of Knowledge: http://www.wokinfo.com/
- LILACS database, for trials from the Portuguese- and Spanish-speaking world: http://bases.bireme.br/cgibin/wxislind.exe/iah/online/?IsisScript=iah/iah.xis&base=LILACS&lang=i&form=F
- Results from clinical trials of marketed pharmaceuticals: http://www.clinicalstudyresults.org
- PubMed: http://www.ncbi.nlm.nih.gov/pubmed/
- OpenSIGLE database: http://opensigle.inist.fr/
Searching other resources
The reference lists of retrieved potentially eligible studies and relevant reviews were also searched. Novogen, manufacturer of a standardised extract of phytoestrogens (Promensil), was contacted for details of unpublished trials.
Data collection and analysis
Selection of studies
Trials for inclusion in the review were selected at different times by two review authors (AL and FK, JM or JB) after the search strategy described previously was employed. First, titles and abstracts were scanned, and full-text copies of those that appeared relevant were retrieved to determine whether they met the inclusion criteria for the review. If necessary, authors of potential trials for inclusion were contacted to clarify study eligibility. Disagreements over selection were resolved by consensus. The selection process for the 2013 update has been documented on a flow chart (Figure 1).
|Figure 1. Study flow diagram.|
Data extraction and management
Data were extracted independently by at least two review authors (AL and FK, JM or JB), who used a specially designed data extraction form. Any discrepancies in data extraction were resolved by consensus. When necessary, additional information on trial methodology or original trial data were sought from the principal or corresponding author of any trials that met the eligibility criteria (see Acknowledgements for details of the authors who provided additional clarification of data beyond that reported in the publications).
Data extracted included details on study characteristics (participants, interventions and comparison groups) and outcome data. When necessary, missing data were imputed from data in other, similar trials or were calculated by using formulas suggested in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Assessment of risk of bias in included studies
All assessments of risk of bias were performed independently by at least two review authors (AL and FK, JM or JB), who used the Cochrane 'Risk of bias assessment tool' (Higgins 2011); results were compared. Any discrepancies were resolved by consensus. Criteria assessed included randomisation method, allocation concealment, blinding of participants and investigators, blinding of assessors, incomplete outcome data and selective outcome reporting. Summary assessments of risk of bias are presented in Figure 2 and Figure 3.
|Figure 2. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.|
|Figure 3. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
Risk of bias assessments have been incorporated into sensitivity analyses (see below).
Measures of treatment effect
When trials were combined in a meta-analysis, summary effect measures were calculated. For dichotomous data, the numbers of events in the intervention and control groups were used to calculate risk ratios (RRs), together with their 95% confidence intervals (CIs). For continuous data, the weighted mean difference (MD) between groups, together with 95% CIs, was calculated. For all other studies, findings in the individual study publications were reported in narrative format and compared.
Unit of analysis issues
The primary unit of analysis was per woman randomised. Only first-phase data from cross-over trials were analysed.
Dealing with missing data
When data were missing, attempts were made to obtain these data from the authors of relevant included studies. Clarifications of data and details from the publications were received from a number of study authors (see Acknowledgements).
Assessment of heterogeneity
We considered whether the clinical and methodological characteristics of included studies were sufficiently similar to warrant meta-analysis. When meta-analyses were performed, statistical heterogeneity was assessed by the Chi
Assessment of reporting biases
In view of the difficulty of detecting and correcting for publication bias and other reporting biases, the review authors have attempted to minimise their potential impact by ensuring a comprehensive search for eligible studies and by being alert to duplication of data. No funnel plot was generated, as most of the studies were synthesised narratively because of substantial heterogeneity.
A priori, it was decided that results from the included studies would be combined in meta-analysis only if similarities were noted in the baseline experience of hot flushes among participants: the composition, type and dosage of the phytoestrogen interventions; the duration of the studies; and the outcomes measured. Significant heterogeneity was seen in the isoflavone concentration of foods and extracts used in the trials that were considered to contain high levels of phytoestrogens. Because of this variation in isoflavone concentration and the variation in the general mix of constituents of each phytoestrogen intervention, pooling of different food products, tablets and extracts was not considered appropriate, and results were reported separately for each trial in table format (see Table 1 Table 2 and Table 3).
Data from five trials were combined in meta-analyses because the intervention was a standardised dose of Promensil (Baber 1999; Jeri 2002; Knight 1999; Tice 2003; van de Weijer 2002). It was planned that a fixed-effect model would be used to combine studies in the meta-analyses, although both fixed-effect and random-effects estimates were calculated initially and results compared.
In the forest plots, an increase in the risk of a particular binary outcome that may be beneficial (e.g. improvement in hot flush severity) or detrimental (e.g. proportion with adverse events) is displayed graphically in the meta-analyses to the right of the centre line, and a decrease in the risk of an outcome to the left of the centre line. Similarly, for continuous data, for some outcomes a higher value for an outcome may be considered beneficial (e.g. greater change in vasomotor symptom score) or detrimental (e.g. number of hot flushes per day), and interpretation will be guided by considering the graph labels that are reversed for benefit as opposed to detriment.
Subgroup analysis and investigation of heterogeneity
For most trials, when results were reported in tabular form, subgroup analysis was undertaken because of variation in the phytoestrogen interventions. Trials were grouped a priori according to the type of phytoestrogen given in the experimental arms. Subgroups included the following.
- Trials in which the phytoestrogen given was in the form of dietary soy, such as flour, beverage or powder containing isoflavones.
- Trials in which the phytoestrogen given was in the form of a soy isoflavone extract.
- Trials in which the phytoestrogen given was in the form of a red clover extract.
- Trials in which the phytoestrogen given was in the form of a predominantly genistein extract.
- All other trials.
Statistical heterogeneity between the results of studies pooled in meta-analyses was examined by inspecting the scatter in the data points and the overlap in their confidence intervals and, more formally, by checking results of the Chi
Sensitivity analysis was conducted to compare differences among participants, interventions, outcomes and methodological quality of included studies.
- Comparison of trial results of all included studies with those studies at low risk of bias (with at least double blinding, adequate concealment, intention-to-treat analyses).
- Comparison of trial results of all included studies with those studies in which a power calculation was performed for sample size.
- Comparison of trial results of all included studies with those studies in which women were required to have at least five moderate to severe hot flushes per day before they were eligible to participate.
- Comparison of trial results of all included studies with those studies using more than 50 mg/d of isoflavones in the treatment group.
Overall summaries: 'Summary of results' tables
As few of the studies could be combined in meta-analyses, separate 'Summary of results' tables were generated to display efficacy, safety and acceptability outcomes for each trial ( Table 1, Table 2 and Table 3). Study results in these tables should be considered by referring to the quality of the individual study (Figure 3) to aid in interpretation of overall results.
Description of studies
Results of the search
For earlier versions of this review, 30 studies (2,730 participants) were included, 31 were excluded and 11 were awaiting classification (further details on the total number of potentially eligible trials are not available).
For the 2013 update of the review, the search retrieved 51 potentially eligible additional studies through inspection of titles and abstracts (Figure 1).
Of 51 potentially eligible studies in the 2013 update, a further 16 new studies met the inclusion, criteria together with two additional studies, which were later publications of studies already included in the review, with longer follow-up or additional results. A total of 43 RCTs (with 4,364 participants) is included in the review (Figure 1). Full details of the included studies are displayed in an additional table (Characteristics of included studies).
Study design and setting
A total of 38 studies used a parallel-group design, and the remaining five used a cross-over design. One cross-over trial had no washout period, and in the remaining four trials, the washout period ranged from seven days to one month. One cross-over trial was combined with parallel-group trials in forest plots; only data from the first phase of the trial before cross-over were used in these analyses.
Most participants were recruited solely from menopause clinics or through a mixture of advertisements and flyers placed in medical practices or in the community; the source of recruitment was not specified in 14 trials. Participants in these trials were experiencing vasomotor symptoms (hot flushes or night sweats) ranging from at least one flush per day to up to 15 flushes per day. Fifteen other trials were included in which vasomotor symptoms or scores on menopausal symptom indices were measured at baseline, although specification of the level of these symptoms was not a requirement for inclusion in the trial. Two of the trials measured the effects of treatment in subgroups (only those women with symptoms at baseline) of randomly assigned participants. One trial excluded women with severe menopausal symptoms who required medical treatment. Menopausal status was most often confirmed by follicle-stimulating hormone (FSH), luteinising hormone (LH) and plasma oestradiol measurements and/or by amenorrhoea ranging from two months to up to 10 years. Elderly women were not included; participants usually ranged in age from 40 to 65 years, although one trial included women up to 75 years of age. Because the minimum threshold of the last menstrual period ranged from two to 12 months or longer, many trials included a mix of perimenopausal and postmenopausal women. Three trials explicitly recruited perimenopausal women; women were required to have no more than one menstrual period during the three months before recruitment (ages ranged from 45 to 55 years) in one trial; in another, women were required to have had at least one period over the past 12 months (average time since last menstrual period was 16 weeks), and in another, women were 45 to 55 years of age and showed cycle irregularity over the previous 12 months or last menstruation at least three but no longer than 12 months previously. In most of the trials, women using HT, currently or recently, were excluded. Other exclusion criteria included women on a vegetarian diet or on a soy-rich diet, malignancy, comorbidities and taking medication that might interfere with assessment of vasomotor symptoms. It was not clear in most trials whether participants had a natural or surgical menopause, but nine trials specifically excluded women with a surgical menopause. Women in five trials were from Australia, seven trials were performed in Italy, eight in the USA, seven in Brazil and the remainder in Israel, Japan, Canada, Sweden, France, Ukraine, Belgium, Ecuador, Peru, Austria, Taiwan, the Netherlands, India, China and Iran.
Interventions used in the trials varied substantially.
Type and method of delivery of phytoestrogen
Trials were grouped into broad categories according to method of delivery and type of phytoestrogen.
- Thirteen trials assessed the effects of dietary substances in the form of flour, powder or beverages derived from soy isoflavones with varying amounts of phytoestrogen enrichment.
- Twelve trials assessed the effects of varying types of soy isoflavone extracts, usually in tablet form.
- Nine trials assessed the effects of red clover extracts (five of the nine used a standardised extract manufactured by Novogen under the brand name Promensil).
- Five trials assessed the effects of mainly genistein extracts on hot flushes.
- The remaining trials (n = 6) assessed other types of phytoestrogen supplements: Three trials investigated the effects of flaxseed dietary supplements (two of which had soy dietary supplement arms and were included in the first category and the other trial also assessed the effects of a flaxseed extract in addition to the flaxseed dietary supplement); one looked at two doses of a hop extract (Humulus luputus L.), one investigated the effects of a standardised natural S-(-)equol containing supplement (SE5-OH) (a metabolite of isoflavones) and another investigated the effects of an extract taken from the roots of Rheum rhaponticum (ERr 731) (which is considered a phytoestrogen supplement). The authors of this trial noted that ERr 731 has been used by perimeopausal and postmenopausal women in Germany since 1993.
Duration of the interventions provided was three months in most of the trials (or three months for the first phase of cross-over trials). Five trials had a duration of 16 weeks, nine trials had a duration of 24 weeks, one trial had a duration of 10 months, four trials had a duration of one year and three trials had a duration of two years.
The phytoestrogen interventions were mostly placebo controlled, although three open studies compared phytoestrogens with other types of control, either different diet with no phytoestrogens or calcium tablets. One other study included a blinded arm that compared flaxseed extract capsules with placebo capsules and another unblinded arm in which flaxseed dietary powder was used. Six placebo-controlled studies compared different doses of the phytoestrogen intervention, and two other placebo-controlled studies compared different types of phytoestrogens (e.g. comparison of a soy diet with a linseed diet or flaxseed muffins with soy muffins). Three studies compared phytoestrogens with HT and placebo, and another compared phytoestrogens solely with HT without a control group.
Most of the trials were pilot studies that did not use power calculations. The effect of the interventions provided in the included studies on total menopausal scores derived from general menopausal symptom questionnaires (such as that of Kupperman and Greene) was not an outcome of this review, although studies were included if they measured vasomotor symptoms on a subscale of a compendium score. Most included studies assessed the effectiveness of the intervention as the primary outcome, although effectiveness was measured in different ways (number of hot flushes per day after treatment, percentage decrease in frequency of hot flushes, severity score after treatment, proportion that reported any reduction in frequency). A few studies separately reported on the frequency and severity of night sweats. Frequency of hot flushes or night sweats was generally reported by participants themselves in a daily diary. Severity was recorded usually in the scales or subscales of general menopause symptom rating scales in different categories, but a few studies required that women record severity in prespecified categories in their daily diaries. Menopause symptom scales included Menopause Symptoms Questionnaire, Menopause Rating Scale, Kupperman Index, Greene Climacteric Scale, Menopause-Specific Quality of Life Questionnaire, Women's Health Questionnaire and the modified Climacteric Symptom Evaluation Checklist. These instruments commonly used a 4-point scale from 0 (no symptoms) to 3 (severe symptoms) to categorise severity, but a few scales used a larger number of categories.
Two studies specifically assessed the safety of the intervention (as measured by effects on endometrial stimulation) as the primary outcome, and 14 others assessed these measures as secondary outcomes. A few studies also assessed the effects of the intervention on the vaginal epithelium or on pH—each of which is a surrogate outcome that is a biological indicator of oestrogenic activity. Adverse events were reported in a few trials but generally were collected as spontaneous reports. Most trials provided details of withdrawals before the study was completed, and a few indicated whether these occurred because of adverse effects or because of problems with acceptability of the intervention.
Of 51 potentially eligible studies for the 2013 update, 28 were excluded because women were not symptomatic at baseline, the studies were not randomised, the duration of the study was less than 12 weeks, the interventions assessed were not included in the review, women had breast cancer, the intervention was a combination treatment, the study was a dose-finding study that did not include a control group or the interventions were not considered phytoestrogens. A further three studies, originally included in the review, were also excluded because the participants had minimal vasomotor symptoms at baseline (Dodin 2005; Duffy 2003; Woo 2003). A total of 60 studies have been excluded from the review (Characteristics of excluded studies). Five studies were considered potentially eligible in the 2013 update and are awaiting classification; a total of eight studies are now awaiting classification.
Risk of bias in included studies
Risk of bias in the included studies is summarised in chart format (Figure 2 and Figure 3). However, given that the results are mostly presented in narrative format in subgroups to reduce the variability of the intervention, the overall risk of bias of each study is also included in the 'Additional tables' summarising the results, so that the reader can judge the quality of the trial evidence for each subgroup separately.
In all, 32 of the studies gave full descriptions of an adequate randomisation procedure and were considered at low risk of bias. The remaining 11 trials claimed that randomisation was the method of allocation, but the method was not described; these trials were considered at unclear risk of bias. Less than half of the studies (n = 19) reported methods to conceal allocation and were considered at low risk of bias; the remaining trials reported no details and were considered at unclear risk of bias.
Nearly all of the trials reported that treatments were blind to participants, investigators and outcome assessors, but the procedures used to ensure that this occurred were not always described. In many studies, the outcome assessors were the participants, as they evaluated their own experience of hot flushes through questionnaires. In four studies, blinding was not possible because the interventions were different types of diets or because phytoestrogens were compared with calcium (although for this latter study, lack of blinding was not likely to affect measurement of the primary outcome—endometrial stimulation).
Incomplete outcome data
For 20 studies, no dropouts or withdrawals were discussed, numbers were balanced between groups or missing data were imputed; these studies were considered at low risk of bias for incomplete outcome data. Eighteen studies were considered at high risk of bias; in these studies, dropouts and withdrawals ranged from 16% to 31%. Five studies were considered at unclear risk of bias, as the percentage of dropouts ranged from 10% to 15% and/or dropouts were unbalanced between randomly assigned groups.
Eighteen studies had low risk of bias, as all prespecified and potential outcomes were reported; seven had unclear risk of bias and 18 had high risk of bias because adverse events were not reported or because outcomes that had been prespecified were not reported.
Effects of interventions
Five of the included studies assessed the effects of Promensil, which is a standardised product, and their data were combined in a meta-analysis. Because of the heterogeneity of the phytoestrogen interventions provided in the other included studies (dose, composition, type), these data could not be pooled but were synthesised in narrative format and displayed in separate tables for efficacy, safety and acceptability outcomes (see Table 1, Table 2 and Table 3).
Primary outcome: efficacy
Of the 13 included studies that used some type of substance containing dietary soy and that had efficacy analyses of any kind, seven studies indicated that no significant differences in primary efficacy outcomes were noted between the soy intervention and control groups.
Of the remaining six studies, one study assessed vasomotor symptoms specified as "somatic" symptoms on the Menopause Rating Scale. The Carmigiani study reported that both women on hormone therapy and women taking dietary soy supplementation (90 mg isoflavone) had significantly improved somatic symptoms (hot flushes and muscle/joint problems) (46% and 50%, respectively) when compared with placebo (29%) (Carmigiani 2010). This study found a significant difference in the frequency of hot flushes. The Albertazzi study of 104 women compared soy powder containing 76 mg/d of isoflavones with casein powder over 12 weeks (Albertazzi 1998). Investigators reported a mean reduction of 1.6 flushes per day (95% CI -1.95 to -1.2) for participants consuming soy powder compared with placebo. This was also expressed as a 45% reduction in the number of hot flushes with soy powder compared with a 30% reduction with placebo powder. Two studies found that severity or intensity of hot flushes was significantly reduced by the intervention. Brezinski compared a phytoestrogen-enriched diet that was individualised for each participant by a dietician (exceeding the cutoff point of > 30 mg/d of isoflavones) versus a regular diet that avoided phytoestrogen-containing foods consumed by a control group (Brzezinski 1997). Hot flushes (rated in a menopause symptoms questionnaire) were reduced in severity in both arms of the study but to a significantly greater extent in the phytoestrogen diet group. This study was one of the few that was not blinded, and knowledge of treatment could have affected participants' assessments. In the Radhakrishnan study, a significantly higher proportion of women (84%) reported improvement in hot flush symptoms (severity) with soy protein when compared with placebo (60%), but no evidence was found of a significant difference in the hot flush score (mean hot flushes per day) after six months (Radhakrishnan 2009). Two studies reported other significant differences, but it is unclear whether the scores represented frequency or severity or a combination of the two. The Cheng study reported that women taking 60 mg isoflavones daily had a significantly lower hot flush score (57%) than those taking placebo, but details on what the score represented are not clear (both number of daily hot flushes and intensity were recorded) (Cheng 2007). The Hanachi study reported that soy milk significantly reduced hot flushes by 72% compared with control after three months, but no details were given of the actual values for each group (Hanachi 2008).
Secondary outcomes: safety
Of the six studies that assessed adverse events, five were negative (no significant differences between randomised groups) and one was positive. The positive study (Knight 2001) found that 75% of participants in the soy group had adverse events compared with 17% of the placebo group. Side effects included bloating, nausea, weight gain and concerns about bowel function.
In all three studies that assessed the effects of phytoestrogens on the endometrium, no evidence of a significant difference between groups was found.
Of four studies that assessed the effects of a soy diet on the vaginal maturation index, three found no evidence of a significant difference between phytoestogen and control groups, but one study reported that this index increased by 103% from baseline with a soy diet compared with a 6% increase with linseed and an 11% increase with placebo (Dalais 1998).
Secondary outcomes: acceptability
Of the four studies that assessed the acceptability of the phytoestrogen intervention compared with control, one study reported a difference in the rate of withdrawal due to adverse events (Knight 2001) (P value not reported). This small study reported that 25% of participants who consumed a beverage containing soy powder withdrew from the study because of dislike of the taste compared with 8% in the placebo group.
Dietary food supplements varied enormously in the type of product used in the trials, the formulation and the isoflavone content (42 mg/d to 134 mg/d). Sensitivity analysis was undertaken to attempt to explain differences in efficacy outcomes between the six positive studies and the seven negative studies. In particular, the difference between positive and negative trials was not explained by the level of isoflavones in the food product. Variability in trial results could have been caused by other factors for which no controls could be applied. Intestinal florae convert soy isoflavone to equol—a more potent oestrogenic isoflavone that is absorbed along with unconverted genistein and daidzein; this conversion is variable (Adlercreutz 1990) and may have influenced the heterogeneity of the results. The severity of hot flushes at baseline could also explain the differences. In the six positive studies, severity of hot flushes among participants was variable; two trials required that women have at least five or eight moderate to severe flushes per day, but in the other trials, hot flushes were mild or were unspecified.
Quality of the trials in this subgroup was variable; only one positive trial had low risk of bias. Of the six trials with positive findings, two trials had very high dropout rates (24% and 21%) and two trials were unblinded and were thus considered at high risk of bias.
Primary outcome: efficacy
Of the 12 studies that compared various types of soy extract in capsule or tablet form (11 vs placebo and one vs HT), nine studies (all vs placebo) reported significant differences in efficacy outcomes (frequency or severity). Five trials (Bicca 2004; Faure 2002; Khaodhiar 2008; Nahas 2007; Ye 2012) reported a reduction in the frequency of flushes (one also found a reduction in the frequency of night sweats); four trials found a reduction in severity of flushes as measured by the Kupperman vasomotor symptom score (Han 2002; Jou 2008; Nahas 2007) or by a subjective rating by participants on a scale of 1 to 3 (Upmalis 2000). This latter trial reported that severity of night sweats did not differ at the end of the study according to group. Not all of the positive studies described benefit from soy extracts; one trial found that women were significantly MORE likely to have hot flushes after isoflavone treatment (48.4%) than after placebo (31.7%), although this was a secondary outcome in the trial (Levis 2011). The trial that compared soy extract with oestrogen therapy (ET; Kaari 2006) reported no differences between them in the percentage of participants reporting any reduction in hot flushes (at six months, P = 0.74; Student's t test).
Secondary outcomes: safety
Of the eight studies that assessed safety outcomes, one assessed effects on endometrial stimulation, four on vaginal pH, five on endometrial thickness, six on vaginal maturation index and six on adverse events. The trial that compared soy extract with ET (unopposed oestrogen therapy) (Kaari 2006) reported significant improvement in vaginal pH and maturation index in the ET group. The soy extract group had a significantly thinner endometrium, less endometrial stimulation and fewer adverse events (all of which were genital bleeding in the ET group). One of the three other trials that compared soy extract with placebo found significantly greater improvement in vaginal pH in the soy group (Bicca 2004). One of the six studies that assessed adverse events reported that women taking soy extracts had a significant increase in rate of constipation and in fractures compared with women taking placebo (although this latter outcome was not considered to be related to treatment) (Levis 2011). For all other studies, no evidence was found of differences in endometrial thickness, vaginal maturation index or incidence of adverse events.
Secondary outcomes: acceptability
Two studies assessed the acceptability of the interventions as measured by withdrawal due to adverse events; no evidence of a difference between groups was found.
Sensitivity analyses exploring the effects of quality issues, levels of isoflavones in the active arm (ranging from 33 mg/d to 200 mg/d) or severity of flushes at baseline did not explain the differences in results. The five placebo-controlled trials that found a difference in flush frequency (Bicca 2004; Faure 2002; Khaodhiar 2008; Nahas 2007; Ye 2012) out of the nine that measured this outcome reported reduction ranging from 50% to 74% with soy extract compared with reduction ranging from 21% to 43% with placebo. In contradiction to these findings, one trial actually reported a greater proportion of hot flushes after treatment with soy extracts versus placebo. Six trials had a longer duration than the more usual 12 weeks, ranging from 16 weeks to two years. The trial that compared soy extract versus ET (Kaari 2006) found no difference in the percentage of participants reporting a reduction in hot flush frequency, but participants had only mild symptoms at baseline (55% and 72% had hot flushes at baseline in the soy and ET groups, respectively). Although no placebo group was included in the study, the authors concluded that their results suggest that the soy isoflavone extract at 120 mg/d was effective in relieving the frequency of hot flushes; however, the trial was considered at high risk of bias.
Severity scores were significantly different in four of the seven trials that measured this outcome; three trials used the Kupperman vasomotor scale (rating severity from 0 to 3; Han 2002; Jou 2005; Nahas 2004), and the other used a simple severity scale (with scores of 1 to 3 representing mild, moderate and severe symptoms) scored daily by participants (Upmalis 2000). Variability in the results of included trials was not explained by sensitivity analyses of quality and other aspects of the studies.
Red clover extracts
Nine trials assessed the effects of red clover extracts on outcomes. Five of these used Promensil, and data from these trials were included in meta-analyses. The other trials used MF11RCE (80-mg isoflavones), a red clover supplement with 40 mg isoflavones or a red clover extract with 120 mg isoflavones. We used data from the first phase of the Baber and Imhof cross-over trials.
Primary outcome: efficacy
Promensil: Five studies reported on the incidence of daily hot flushes after treatment with two different doses of Promensil (40 mg/d and 80 mg/d) (Baber 1999; Jeri 2002; Knight 1999; Tice 2003; van de Weijer 2002). No significant differences were reported between groups in the overall incidence of hot flushes (MD -0.93, 95% CI -1.95 to 0.10, I
|Figure 4. Forest plot of comparison: 1 Promensil versus placebo, outcome: 1.1 Incidence of hot flushes (number/d).|
Other red clover extracts (not included in the meta-analyses): One trial using a dose of 80 mg of red clover (Imhof 2006) found significant benefit for daily frequency of hot flushes and night sweats and for the mean percentage of decrease in these symptoms (P = 0.0001). Another study that assessed an unspecified red clover extract (same dose) reported benefits for hot flush and night sweat severity (as assessed by the Kupperman Index; Hidalgo 2005). After treatment, 15% of women taking 80 mg of red clover reported hot flushes compared with 98.1% of women taking placebo; values for night sweats were 30.2% and 92.5% for red clover and placebo, respectively (P < 0.05). The authors claimed that these values represented severity "as expressed as a percentage," but it is not clear what they meant. The other two studies assessing the efficacy of red clover (Del Giorno 2010; Geller 2009) found no difference between groups when treatment was given for 12 months.
Secondary outcomes: safety
Results are reported separately for Promensil and other red clover extracts.
One large trial of Promensil versus placebo assessed adverse events (Tice 2003). It reported no differences in the proportions of women who experienced any adverse event (RR 0.95, 95% CI 0.65 to 1.40). Also, no differences were found between groups in rates of specific adverse events, such as respiratory tract infection, headache, myalgia, nausea, arthralgia, diarrhoea and vaginal spotting. Two other studies (not included in the meta-analysis) also did not find differences between groups with respect to adverse events.
Three trials assessed the effects of treatment on endometrial thickness (Baber 1999; Geller 2009; Imhof 2006). One trial (included in the meta-analysis) found no difference in endometrial thickness after 12 weeks of treatment with Promensil. The other two trials reported different findings: One reported a significant decrease of 15% in endometrial thickness in women treated with red clover compared with zero change in women treated with placebo (SD of change not given, P < 0.001; Imhof 2006), and the other found no evidence of a significant difference between groups (Geller 2009).
One study that assessed an unspecified red clover extract reported significant changes in all vaginal cytology indexes (karyopyknotic index, cornification index, maturation index) when compared with placebo (P < 0.05) (Hidalgo 2005).
Secondary outcomes: acceptability
No trials assessed the acceptability of treatment.
Variability in study results appeared to be explained in part by trial quality. One small Peruvian study (n = 30) of poor quality (Jeri 2002) reported highly significant effects for both frequency and severity but gave no details on randomisation method, allocation concealment or baseline comparability. Inclusion of this study in the meta-analyses, combined with findings of other, larger trials of better quality, caused highly significant heterogeneity, and a random-effects model was chosen for presentation of results. Exclusion of this small trial of poor quality from the meta-analyses reduced heterogeneity and the summary effect estimate, suggesting that benefit derived from a dose of 40 mg of Promensil per day was no longer significant. Another study (van de Weijer 2002) reported a significant benefit of Promensil (at a dose of two tablets per day) but did not provide an indication of the variability around the estimate so could not be included in the meta-analysis. A large trial of good quality (n = 252) (Tice 2003) that compared two types of red clover extract—Promensil (two tablets per day) and Rimostil—versus placebo reported no significant change in the frequency of hot flushes between groups and no significant difference in the change in vasomotor score over the period of the study. One of these studies also compared a higher dose of Promensil (160 mg/d) with placebo, but substitution of these values in the meta-analysis did not alter the results.
Four studies compared other types of red clover extract (Del Giorno 2010; Geller 2009; Hidalgo 2005; Imhof 2006), but findings were inconclusive regarding efficacy. Two of these studies found benefit for 80 mg of red clover, but two others reported no evidence of significant differences associated with a dose of 40 mg or 120 mg of red clover at the end of 12 months of treatment. Variation in the findings was explained to some extent by differing quality of the trials. Results from larger studies of better quality appeared to conflict with those from smaller studies of poorer quality.
Five studies assessed the effects of predominantly genistein extracts on outcomes (Crisafulli 2004; D'Anna 2007; Evans 2011; Ferrari 2009; Sammartino 2003). Genistein doses ranged from 30 mg to 60 mg per day. Duration of treatment with genistein ranged from 12 weeks to two years.
Primary outcome: efficacy
All four studies assessing primary efficacy—two with unclear risk of bias and two with low risk of bias—reported that genistein significantly improved the frequency of hot flushes when compared with placebo (Crisafulli 2004; D'Anna 2007; Evans 2011; Ferrari 2009). One study also found that the duration of hot flushes was reduced compared with placebo (although this is not an outcome of this review). The Crisafulli study also compared genistein with continuous hormone therapy; it reported a 24% mean reduction in daily hot flushes with genistein compared with placebo (P < 0.05) and a 30% reduction in daily hot flushes with HT compared with genistein (P < 0.05). The other three studies reported a mean percentage reduction in daily hot flushes from baseline ranging from 41% to 61% with genistein in comparison with a mean reduction ranging from 7% to 29% with placebo. Two studies of 12 weeks' duration found no evidence that the severity or intensity of hot flushes differed between genistein and placebo groups (Evans 2011; Ferrari 2009), but a study of longer duration reported that hot flush severity declined significantly when compared with placebo over two years (D'Anna 2007).
Secondary outcomes: safety
Four studies did not find a significant difference between groups in endometrial thickness after genistein treatment, and one study found no evidence of a significant difference in the vaginal maturation value.
Two studies found no evidence of a significant difference in adverse events between randomly assigned groups. In one study, most of these adverse events were gastrointestinal. No severe adverse events were experienced by participants.
Secondary outcomes: acceptability
The proportion of participants who were satisfied with treatment was similar in both groups in one study (79% with genistein compared with 69% with placebo; Heger 2006).
The four studies had low or unclear risk of bias. All studies evaluating efficacy consistently reported significant reductions in the frequency of hot flushes at the end of treatment ranging from 12 weeks to two years. Individual trial characteristics were generally fairly similar. Doses of genistein ranged from 56 mg to 60 mg of genistein per day, with one trial using a 30-mg dose. Women generally had at least four hot flushes per day, and two studies reported average numbers of eight and nine hot flushes per day. Reductions in the number of hot flushes with genistein ranged from 24% to 56% against placebo. Placebo response, when reported, was variable; the two-year study reported a reduction from baseline of 7.2%, but two shorter studies of 12 weeks' duration reported reductions of 27% and 29%. Results of the effects of genistein on the severity of hot flushes were more mixed and could not be explained by trial characteristics or quality.
Three studies compared flaxseed dietary supplement or flaxseed extract versus placebo or control (Colli 2012; Dalais 1998; Lewis 2006); one compared two strengths of hop extract with placebo (Heyerick 2006), one compared a natural supplement containing S-(-)equol, a daidzein metabolite, with placebo in women who were equol nonproducers (Aso 2012) and one compared an extract of Rheum rhaponticum (ERr 731) with placebo in perimenopausal women (Heger 2006). Duration of treatment was 12 weeks in most of the studies, but one trial had a duration of 16 weeks and another had a duration of 24 weeks.
Primary outcome: efficacy
All six trials assessed efficacy outcomes. Two trials found no evidence of a difference between groups in frequency or intensity of hot flushes after flaxseed treatment (Dalais 1998; Lewis 2006), and both Lewis and Colli did not find evidence of a time by interaction effect of hot flush severity or intensity between groups. The study evaluating hop extracts did not find evidence of a significant difference in the Kupperman Index hot flush severity score, although a non-significant trend favoured both hop extract doses (Heyerick 2006). The study comparing S-Equol with placebo found that S-Equol was associated with a significantly larger decrease in hot flush frequency from baseline (62.8%) compared with placebo (23.6%) in women with three or more hot flushes per day and a significant improvement in hot flush severity with S-Equol (61%) compared with placebo (45%) (Aso 2012). Results of the trial were assessed in postmenopausal women with at least one hot flush per day and low rates of equol excretion, as this was considered to reduce the possibility of confounding by background isoflavone intake. Another trial investigating the effects of an extract from the roots of Rheum rhaponticum (ERr 731) in perimenopausal women in the Ukraine reported that both the frequency and the severity of moderate to severe hot flushes and sweats (Menopause Rating Scale II and Menoqol vasomotor score) were significantly reduced with ERr 731 compared with placebo (mean decrease of 2.5 points compared with mean decrease of 1.2 points) at week 12 (Heger 2006). Effects were assessed blindly by both participants and investigators.
Secondary outcomes: safety
Three trials assessed endometrial thickness after six months or one year and reported no significant difference between groups (Colli 2012; Crisafulli 2004; Heger 2006). Also no evidence of a change in vaginal maturation index was noted with flaxseed compared with placebo or control (Colli 2012; Dalais 1998). In one trial, women ingesting flaxseed meal withdrew from treatment in greater numbers than those taking flaxseed extract or placebo because of gastrointestinal complaints (Colli 2012), but these differences were not tested statistically.
Secondary outcomes: acceptability
One trial (Heger 2006) reported that 63% of women taking ERr 731 were satisfied with their treatment compared with 32% of women taking placebo (no P value reported).
Studies using similar interventions in this subgroup were too few for sensitivity analyses to be undertaken.
Summary of main results
This review has assessed the effectiveness, safety and acceptability of foods, supplements or extracts containing phytoestrogens when compared with placebo, no treatment and HT in randomised studies completed by the end of July 2013. It has been able to pool only the studies that used Promensil in meta-analyses because of the heterogeneity of the other phytoestrogen interventions.
Primary outcome: efficacy
Of the 13 included studies that used some type of substance containing dietary soy and had efficacy analyses of any kind, seven studies indicated that no significant differences were seen between the soy intervention group and the control group in terms of primary efficacy outcomes. In studies that reported significant findings, interventions included phytoestrogen-enriched diets, soy milk, fruit drinks with isoflavones and soy powders. Only one trial had low risk of bias, and participants varied in the severity of their flushes at baseline. Sensitivity analyses could not explain the variable results. Thus, the findings from these trials must be considered only tentative, as variability and significant bias influencing the findings cannot be excluded.
Overall, no evidence suggested that a diet with high levels of soy phytoestrogens had a positive effect on hot flush frequency or severity.
Of the 11 trials that compared soy extracts with placebo, nine had some positive results and two were negative. Five of nine studies found significant improvement in hot flush frequency with soy extract, but one found that soy extract was associated with more hot flushes than were seen with placebo. Four of seven studies found that hot flush severity was significantly reduced with soy extract, but most of these studies were at high risk of bias. One other study at high risk of bias found no difference in the effect of soy extract or hormone therapy on hot flush symptoms (as measured by the Kupperman Index).
Given the variability in the interventions, the severity of hot flushes at baseline and the potential for risk of bias, no overall conclusive evidence showed that soy extracts had a positive effect on hot flush frequency or severity.
Red clover extracts
Five studies assessed the effects of Promensil, and four studies assessed the effects of other red clover extracts. Findings were inconclusive and could largely be explained by risk of bias. The two larger studies at low risk of bias found no evidence of benefit with red clover extracts.
Overall, no evidence suggested that red clover extracts had a positive effect on hot flush frequency or severity.
All four studies found consistent benefit for hot flush frequency with doses of genistein ranging from 30 to 60 mg per day in women with moderate to severe hot flushes, although benefits for hot flush severity were more mixed. Although benefits were found with genistein, they were significantly less than those associated with continuous hormone therapy in one study. These positive findings should be considered tentative as, in two of the four studies, effects on hot flushes were secondary outcomes, and in one study, measurements were made in a subgroup from the total study population, which may have introduced bias (Schulz 2005).
Overall, genistein extracts appeared to significantly reduce the numbers of hot flushes experienced by symptomatic postmenopausal women but to a lesser extent than hormone therapy.
Among six trials that assessed the effects of other types of phytoestrogens, no evidence of an efficacy difference was noted between flaxseed or linseed diets or extracts and placebo or control in three trials (two of which also contained soy diet arms) or in one trial that assessed a phytoestrogen preparation derived from hops in two different doses in women who had two to five daily flushes. In one trial, ERr 731 was associated with a significant reduction in hot flush and sweating severity symptoms (Menopause Rating Scale II) and in Menoqol vasomotor score compared with placebo. Another trial reported that hot flush frequency and severity were significantly reduced with the S-Equol supplement (a daidzein metabolite) when compared with placebo.
Overall, although benefits were reported from single trials investigating a phytoestrogen extract from the rhubarb plant (ERr 731) and an equol supplement (SE5-OH), data were insufficient to permit determination of whether any other type of phytoestrogen product had significant effects on vasomotor symptoms.
Data on oestrogenic effects on the endometrium and the vagina and rates of adverse effects were collected in only a few trials and were considered together rather than in subgroups.
Phytoestrogen products do not appear to have an oestrogen agonistic effect on the endometrium when given for up to one year, in contrast to hormone replacement therapy. No evidence was derived from two studies in the review to suggest that phytoestrogens promote a proliferative endometrium (Balk 2002; Kaari 2006). Most studies reported no difference in endometrial thickness between phytoestrogens and placebo. One study actually found a significant reduction in endometrial thickness from baseline (Imhof 2006). The lack of an oestrogenic effect on the endometrium was further supported by a study comparing phytoestrogens with HT; endometrial thickness significantly changed from baseline with HT, but no change was seen with phytoestrogens (Kaari 2006).
Evidence of the effects of phytoestrogens on the vaginal maturation index and on pH is mixed. Five placebo-controlled trials found no evidence of a stimulatory effect, but two other studies described a positive oestrogenic effect of increasing cellular mitotic activity, as evidenced by improvement in maturation indices—one with a soy diet versus a wheat diet and the other with a red clover extract (Dalais 1998; Hidalgo 2005). In two studies with a hormone therapy comparator, vaginal cytology values with HT were significantly different from those obtained with soy (Carmigiani 2010; Kaari 2006). Other studies not included in this review, wherein participants were asymptomatic or had breast cancer, have also produced conflicting data. Three studies found evidence of improvement in maturation values (Baird 1995; Chiechi 2003; Uesugi 2004), but three others have not confirmed these results (Duncan 1999; Manonai 2006; Nikander 2005). Characteristics of the individual trials provided no clues that could explain these mixed results. Similarly, one unpublished study described improvement in vaginal pH with a soy extract when compared with placebo (Bicca 2004), and another, much larger study did not find evidence of a difference (Upmalis 2000). In the study that compared soy extract versus HT, vaginal pH improved significantly more with HT than with soy (Kaari 2006). It is hoped that a Cochrane systematic review will be prepared to specifically assess the effects of phytoestrogens on urogenital menopausal symptoms to provide further clarification.
Only three of 17 trials found a significant difference in adverse event rates (Colli 2012; Knight 2001; Levis 2011). The phytoestrogen supplement in the first small trial was give in the form of a powder, and the authors included data on the dislike of the taste of soy powder in the total incidence of adverse events. In this case, the adverse event rate was linked to the type of product used in the soy diet and was more appropriate as a measure of acceptability. Data on the total incidence of adverse effects, with dislike of taste excluded, were not available. The Levis trial, which investigated the effects of soy on bone loss and on vasomotor symptoms, reported that fracture rates were significantly higher in the group of women consuming soy, but this is likely to be a chance event, as all fractures were associated with a traumatic event rather than with osteoporosis. Women ingesting flaxseed meal as a dietary supplement in another trial were more likely to withdraw from treatment because of gastrointestinal complaints when compared with those given flaxseed extract or placebo (Colli 2012). In a trial that compared soy extract with combined HT (Kaari 2006), women in the latter arm were more likely to experience genital bleeding, which is a common symptom of HT in perimenopausal women. This symptom was not experienced by women who took phytoestrogen supplements. Adverse events were most often collected incidentally during the trials.
Few trials specifically assessed this outcome in spite of the high dropout rate in many of the studies. No evidence was found of a difference in acceptability of any of the phytoestrogen products used when compared with placebo, except for one trial; 63% of those taking ERr 731 were satisfied with their treatment compared with 38% taking placebo.
In summary, generally no conclusive evidence showed a benefit of phytoestrogen-enriched or -derived products for menopausal vasomotor symptoms, with the exception of products containing a minimum of 30 mg per day of genistein, which have been evaluated for up to two years in four studies. Also, no evidence indicated that products derived from phytoestrogens were associated with stimulation of the endometrium and/or vagina of women with vasomotor symptoms. No evidence suggested an increase in adverse events, and limited data suggest that phytoestrogen supplements were well tolerated.
Overall completeness and applicability of evidence
The review included 43 studies, but variation in components and duration of the intervention and in types of participants and severity of their hot flushes, as well as potential variation in metabolism and absorption among individuals and a high placebo response rate, generally resulted in inconclusive findings. Although some trials reported a significant beneficial effect of phytoestrogen treatment on symptoms, strong evidence was found of a placebo effect, with improvements in frequency ranging from -1% to -59%—similar to the placebo effect found in the Cochrane review of hormone replacement for vasomotor symptoms (MacLennan 1999). When some of the included studies reported significant differences between groups, it was not possible to tease out which of the many variables that differed between included studies might have explained the results.
Studies varied according to the total amount or 'dose' of isoflavone given in the active treatment arm. The rationale for a role for isoflavones is supported by epidemiological evidence from a community-based study, which found that the incidence of hot flushes was inversely related to the quantity of soy foods consumed and the daily intake of isoflavones (Nagata 2001). Studies of Japanese women claim a typical daily consumption of 20 to 54 mg of isoflavones (Nagata 2001; Somekawa 2001). Most of the studies in this review provided treatments with at least 50 mg per day of isoflavones; some used more than 100 mg of isoflavones per day. Examination of the pattern of results within each subgroup did not indicate that trials were more likely to be positive if they used higher doses of isoflavones. Comparison of total isoflavone levels in treatments may not be useful, as the isoflavone profiles of different supplements and extracts differ considerably. Broad groupings in this review into 'types' of phytoestrogens provided no clues as to the best way that phytoestrogens can be delivered for therapeutic effect.
In addition to the heterogeneity of interventions used in the included studies, good evidence of variability was noted in the metabolism and absorption of isoflavones by individuals, which can lead to variations in serum concentrations of parent isoflavones and their metabolites (Rowland 2003; Wiseman 2004). It has been claimed that only 20% to 30% of the general population in the United States possess gut microflorae that convert the isoflavone daidzein to the more oestrogenic dihydroxy isoflavan equol (Setchell 2002) in comparison with 50% to 60% of Asians (NAMS 2011). It has been suggested that 'equol producers' make up a distinct subpopulation that may be associated with the greatest benefit from soy isoflavones for relief of hot flushes, and they may explain anecdotal reports by many women of phytoestrogen effectiveness in relieving hot flushes. The studies included in this review generally did not control or stratify for this added potential source of variation in response to treatment with phytoestrogens, although investigators in one trial stated that isoflavone supplementation improved symptoms only in women with the ability to produce equol. Another recent trial found a significant benefit associated with use of a supplement of S-(-)equol in equol-nonproducing postmenopausal Japanese women. Further research should consider the role of equol production.
In spite of variation in doses, duration and components of phytoestrogen products, extracts containing at least 30 mg of genistein, a type of isoflavone, appeared to reduce both frequency and severity of hot flushes in women, but less so than hormone therapy. A possible rationale is the suggestion that genistein is the most potent isoflavone with regard to receptor binding and transactivation (Muthyala 2004). No evidence of safety concerns in the short term was found. Further research is needed to confirm the efficacy of high doses of genistein in menopausal women.
Quality of the evidence
The quality of the trials was variable. Although most studies blinded both participants and assessors, less than half reported adequate allocation concealment to protect against selection bias, and high rates of attrition and potential selective outcome reporting in half of the trials suggest that they were at high risk of attrition and reporting bias. Substantial variation in the included studies precluded combining findings in forest plots, and this has limited any definitive conclusions.
Four trials have suggested benefit for the reduction of hot flush frequency and severity from high doses of genistein, but these findings should be considered tentative because of methodological shortcomings in some of the trials. The effect of genistein on menopausal vasomotor symptoms requires verification.
Potential biases in the review process
This review has the following limitations. Because we were unable to pool most of the studies and in many cases had no access to the original data, we accepted the statistical methods used in each study and reported study results in tabular form. Many studies did not use an appropriate statistical method to measure changes in frequency or severity over time; endpoint analysis may have obscured different patterns of response. Some trials used unvalidated menopausal symptom questionnaires to assess severity, and not all scales were similar. For example, the Greene Vasomotor Scale includes hot flushes and night sweats, whereas the vasomotor scale used by Kotospoulos includes hot flushes, lightheadedness and headache.
A second weakness of the review is the potential for publication bias. The search for relevant studies was very comprehensive, and attempts were made to access the grey literature. However, several of the studies waiting assessment are positive studies in abstract form, and their inclusion may paint a different picture. Publication bias usually operates in a differential manner, leading to a higher probability of publication of studies that indicate positive results.
Also, some potential for bias may be associated with the need for the review authors to determine which interventions would be considered as phytoestrogens, given the wide variety of sources and doses of phytoestrogenic compounds.
Agreements and disagreements with other studies or reviews
One review has suggested that phytoestrogen products might be more effective in women with more severe flushes at baseline (Huntley 2004). This hypothesis has been supported by another study, which proposed that the effectiveness of phytoestrogens for hot flush relief is seen only in those with five or more hot flushes per day (Messina 2003). Another systematic review and meta-analysis (Howes 2006) also came to the conclusion that women more severely affected by vasomotor symptoms derived a small benefit from isoflavone supplementation and suggested a cut point of around four flushes per day. However, the authors of that review pooled highly variable studies, causing significant statistical heterogeneity, which affects the credibility of effect estimates. These suggestions have not been confirmed by the sensitivity analyses performed in this review, which compared the overall pattern of results with studies that required women to have at least five hot flushes per day. In addition, the North American Menopause Society (NAMS) position paper on isoflavones, which used strict criteria for evaluation of isoflavones, concluded that no linear dose-response relationship was observed (NAMS 2011), and a recent systematic review (Taku 2012) concluded that the effect of baseline frequency of hot flushes is unclear. When included trials show huge variation in characteristics of participants, types of intervention and outcomes measured, it is important to consider the quality of the trials. One of the largest trials (Tice 2003) was a study of good quality with high compliance, a low dropout rate and good generalisability (including a broad cross section of the population). It required that women have at least 35 hot flushes per week to participate, and investigators found no evidence of benefit for Promensil or Rimostil.
The efficacy findings of this review are broadly in accord with those of most systematic reviews assessing the effects of phytoestrogens on menopausal symptoms published over the past decade (Bolanos 2010; Coon 2007; Geller 2005; Glazier 2001; Haimov-Kochman 2005; Howes 2006; Huntley 2004; Jacobs 2009; Krebs 2004; Low Dog 2005; Nedrow 2006; Villaseca 2012; Williamson-H 2006), and this review has included several more recently published studies. In general, most reviews have concluded that phytoestrogen supplementation has no effect or a very mild effect on vasomotor symptoms. However, NAMS, using strict criteria for selection of relevant trials, concluded that soy-based isoflavones are modestly effective in relieving menopausal symptoms, and that their use in women with distressing symptoms is 'reasonable' (NAMS 2011). A recent systematic review reported that soy isoflavone supplements derived by extraction or chemical synthesis were significantly more effective than placebo in reducing the frequency and severity of hot flushes (Taku 2012). This review included trials with a shorter duration of treatment (six weeks) and women with previous breast cancer; these types of studies were excluded from this review. The Taku review authors also acknowledged substantial heterogeneity in their findings and have noted that further research is warranted to clarify the influence of additional factors, such as dose, isoflavone form, baseline hot flush frequency and duration of treatment.
The Taku review has also confirmed the findings of Williamson-H 2006, which focused specifically on soy isoflavone extracts and stratified studies according to the amount of genistein included in the extract. Findings of this review suggest that supplements that provide at least 15 mg of genistein per day are effective, whereas those providing less genistein are not, and this hypothesis has been supported by the findings of this review. However, this hypothesis has not been supported by the longitudinal Study of Women's Health Across the Nation, which assessed the association between vasomotor symptoms and ethnicity during the menopausal transition in 3,198 women (Gold 2006). Investigators in this study reported that genistein intake in their sample was not related to vasomotor symptoms, and that it did not account for the reduced symptom reporting that they found among Asian women after adjusting for covariates. This hypothesis needs to be further investigated in randomised trials for inclusion in future updates of this review.
The safety of phytoestrogen products has been investigated in another systematic review and assessed in separate subgroups: gynecological or urinary, gastrointestinal, musculoskeletal, neurological or sensory and nonspecific (Tempfer 2009). The review authors concluded that phytoestrogen supplements had a safe side effect profile, along with moderately elevated rates of gastrointestinal side effects. Use of phytoestrogens was not associated with increased risk of endometrial or breast cancer. Most studies in the review were of limited duration. These findings support the conclusion of this review that phytoestrogen-enriched products appear to be safe when used for up to two years. A study of women without hot flushes at baseline (not eligible for inclusion in this review) (Unfer 2004) reported that long-term treatment (up to five years) with soy (150 mg/d isoflavones) was associated with increased occurrence of simple endometrial hyperplasia compared with placebo. This finding has not been confirmed by other studies, but the long-term endometrial safety of high doses of phytoestrogen supplements has not been fully established.
Implications for practice
No conclusive evidence shows that phytoestrogen supplements effectively reduce the frequency or severity of hot flushes and night sweats in perimenopausal or postmenopausal women, although benefits reported with concentrates of genistein may be beneficial and should be further investigated. Many of the included studies were of poor quality, and results were inconsistent, providing no guidance on which type of product is likely to be more beneficial. Women need to be reassured that these symptoms usually abate over time. When therapy is desired or required, the use of unspecified phytoestrogen supplements is not based on good quality evidence of benefit, although it is possible that high doses of genistein may offer relief. No evidence shows harmful side effects in the short term resulting from the use of these supplements.
Implications for research
More research is required to test the following hypotheses.
In addition, to enhance comparability, future trials should be based on phytoestrogen products that are well characterised; they should provide strict monitoring of participants throughout the trial, should be well powered and of adequate duration and should use validated measurements of outcome.
We would like to acknowledge the work of the Trial Search Co-ordinators of MDSG—Mrs Haiyan Feng and Mrs Marian Showell—in the design and implementation of comprehensive search strategies to locate trials for the review. We would also like to thank the Managing Editors—Mrs Jane Clarke and Mrs Helen Nagels—for their assistance in preparing the review for publication.
The following authors of included studies provided additional clarification of data and quality issues: Rod Baber, Marilene Bica, Amnon Brzezinski, Fabien Dalais, Peter Chedraui, Hope Ricciotti, Peter Heger, Francesco Squadrito and Mal Evans.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Appendix 1. MEDLINE search strategy
1 exp perimenopause/ or exp postmenopause/ (17644)
2 postmenopaus$.ti,ab,sh. (41580)
3 menopaus$.ti,ab,sh. (43511)
4 exp Climacteric/ or exp Hot Flashes/ or exp Menopause/ (46568)
5 hot flash$.ti,ab. (1560)
6 hot flush$.ti,ab. (1646)
7 climacteric.ti,ab. (3594)
8 (vagina$ adj3 atroph$).ti,ab,sh. (586)
9 (vagina$ adj3 dry$).ti,ab. (645)
10 endometri$.ti,ab. (62890)
11 or/1-10 (133776)
12 Phytoestrogens/ (2493)
13 phytoestrogen$.ti,ab. (2979)
14 Soy Foods/ (813)
15 soy$.ti,ab. (35598)
16 exp isoflavones/ or coumestrol/ or genistein/ or pterocarpans/ or rotenone/ (13154)
17 linseed.mp. or Flax/ (2037)
18 isoflavon$.ti,ab. (5734)
19 red clover.ti,ab. (715)
20 daidzein.ti,ab. (2323)
21 promensil.ti,ab. (8)
22 or/12-21 (50378)
23 11 and 22 (2078)
24 randomized controlled trial.pt. (337111)
25 controlled clinical trial.pt. (85195)
26 randomized.ab. (252126)
27 placebo.tw. (143456)
28 clinical trials as topic.sh. (162463)
29 randomly.ab. (184559)
30 trial.ti. (108477)
31 (crossover or cross-over or cross over).tw. (54671)
32 or/24-31 (825664)
33 exp animals/ not humans.sh. (3782187)
34 32 not 33 (761678)
35 23 and 34 (686)
36 2012$.ed. (638938)
37 35 and 36 (43)
Appendix 2. EMBASE search strategy
1 exp "menopause and climacterium"/ or climacterium/ or early menopause/ or menopause/ or postmenopause/ (76520)
2 postmenopaus$.ti,ab. (47787)
3 menopaus$.ti,ab. (43440)
4 climacter$.ti,ab. (4494)
5 exp Hot Flush/ (9977)
6 hot flush$.ti,ab. (2262)
7 hot flash$.ti,ab. (2066)
8 (vagina$ adj3 atroph$).ti,ab,sh. (777)
9 (vagina$ adj3 dry$).ti,ab,sh. (1100)
10 endometri$.ti,ab. (75307)
11 or/1-10 (176874)
12 exp PHYTOESTROGEN/ (4474)
13 phytoestrogen$.ti,ab. (3724)
14 plant estrogen$.ti,ab. (69)
15 Daidzein/ or Isoflavone Derivative/ or Genistein/ or Soybean Oil/ or Soybean/ or Isoflavone/ or Soybean Protein/ or soy$.mp. (57048)
16 COUMESTROL DERIVATIVE/ or COUMESTROL/ (662)
17 coumestrol.ti,ab. (404)
18 pterocarpan$.ti,ab,sh. (274)
19 rotenone$.ti,ab,sh. (5229)
20 linseed.ti,ab,sh. (1785)
21 red clover.ti,ab,sh. (947)
22 daidzein.ti,ab,sh. (4250)
23 promensil.ti,ab,sh. (14)
24 or/12-23 (66860)
25 11 and 24 (3465)
26 Clinical Trial/ (871161)
27 Randomized Controlled Trial/ (328650)
28 exp randomization/ (59349)
29 Single Blind Procedure/ (16360)
30 Double Blind Procedure/ (110736)
31 Crossover Procedure/ (34922)
32 Placebo/ (204401)
33 Randomi?ed controlled trial$.tw. (78497)
34 Rct.tw. (9931)
35 random allocation.tw. (1183)
36 randomly allocated.tw. (17608)
37 allocated randomly.tw. (1829)
38 (allocated adj2 random).tw. (711)
39 Single blind$.tw. (12516)
40 Double blind$.tw. (130405)
41 ((treble or triple) adj blind$).tw. (280)
42 placebo$.tw. (178872)
43 prospective study/ (213060)
44 or/26-43 (1272542)
45 case study/ (16909)
46 case report.tw. (230790)
47 abstract report/ or letter/ (843460)
48 or/45-47 (1086430)
49 44 not 48 (1237229)
50 25 and 49 (1414)
51 limit 50 to yr="2012 -Current" (45)
Appendix 3. PsycINFO search strategy
1 exp menopause/ (2681)
2 postmenopaus$.tw. (1757)
3 menopaus$.tw. (3373)
4 perimenopaus$.tw. (440)
5 climacteric.tw. (392)
6 hot flash$.tw. (268)
7 hot flush$.tw. (152)
8 (vagina$ adj3 atroph$).tw. (26)
9 (vagina$ adj3 dry$).tw. (99)
10 or/1-9 (4865)
11 Phytoestrogens.tw. (58)
12 soy.tw. (217)
13 linseed.tw. (8)
14 isoflavon$.tw. (96)
15 red clover.tw. (6)
16 daidzein.tw. (28)
17 promensil.tw. (0)
18 or/11-17 (312)
19 10 and 18 (60)
20 limit 19 to yr="2012 -Current" (5)
Appendix 4. AMED search strategy
1 exp climacteric/ or exp menopause/ or exp postmenopause/ (499)
2 postmenopaus$.tw. (374)
3 menopaus$.tw. (657)
4 exp Climacteric/ (499)
5 hot flash$.tw. (47)
6 hot flush$.tw. (33)
7 climacteric.tw. (49)
8 (vagina$ adj3 atroph$).tw. (2)
9 (vagina$ adj3 dry$).tw. (10)
10 or/1-9 (890)
11 exp Phytoestrogens/ (54)
12 phytoestrogen$.tw. (132)
13 Soy Foods/ (12)
14 soy$.tw. (226)
15 exp isoflavones/ (96)
16 linseed.tw. (7)
17 isoflavon$.tw. (278)
18 red clover.tw. (14)
19 daidzein.tw. (53)
20 promensil.tw. (0)
21 or/11-20 (514)
22 10 and 21 (68)
23 limit 22 to yr="2012 -Current" (1)
Appendix 5. CENTRAL search strategy
1 exp perimenopause/ or exp postmenopause/ (3249)
2 postmenopaus$.ti,ab,sh. (7655)
3 menopaus$.ti,ab,sh. (4117)
4 exp Climacteric/ or exp Hot Flashes/ or exp Menopause/ (5212)
5 hot flash$.ti,ab. (325)
6 hot flush$.ti,ab. (570)
7 climacteric.ti,ab. (573)
8 (vagina$ adj3 atroph$).ti,ab,sh. (113)
9 (vagina$ adj3 dry$).ti,ab. (116)
10 endometri$.ti,ab. (3122)
11 or/1-10 (13074)
12 Phytoestrogens/ (148)
13 phytoestrogen$.ti,ab. (171)
14 Soy Foods/ (67)
15 soy$.ti,ab. (1349)
16 exp isoflavones/ or coumestrol/ or genistein/ or pterocarpans/ or rotenone/ (468)
17 linseed.mp. or Flax/ (111)
18 isoflavon$.ti,ab. (514)
19 red clover.ti,ab. (35)
20 daidzein.ti,ab. (139)
21 promensil.ti,ab. (7)
22 or/12-21 (1716)
23 11 and 22 (516)
24 limit 23 to yr="2007 -Current" (174)
Appendix 6. MDSG search strategy
Keywords CONTAINS "menopausal"or"*Menopause"or"perimenopause"or"perimenopausal"or "Postmenopausal"or "postmenopause"or"climacteric "or "vasomotor"or"hot flashes"or "hot flushes"or"vaginal atrophy"or "vaginal dryness"or Title CONTAINS "menopausal"or"*Menopause"or"perimenopause"or"perimenopausal"or "Postmenopausal"or "postmenopause"or"climacteric "or "vasomotor"or"hot flashes"or "hot flushes"or"vaginal atrophy"or "vaginal dryness"
Keywords CONTAINS "Phyto-Female complex" or"phytoestrogen"or "phytoestrogens" or"phytosterols"or "soy" or"soy-protein diet" or"soy-nut diet" or"soybean"or "soybeans"or "soyfem preparation" or"soymilk" or"isoflavones" or"isoflavonoids" or"red clover"or "daidzein"or "promensil" or Title CONTAINS "Phyto-Female complex" or"phytoestrogen"or "phytoestrogens" or"phytosterols"or "soy" or"soy-protein diet" or"soy-nut diet" or"soybean"or "soybeans"or "soyfem preparation" or"soymilk" or"isoflavones" or"isoflavonoids" or"red clover"or "daidzein"or "promensil"
Last assessed as up-to-date: 30 July 2013.
Protocol first published: Issue 1, 1999
Review first published: Issue 4, 2007
Contributions of authors
Anne Lethaby registered the title; undertook searches, selection of studies, data extraction, quality assessment and data entry; and wrote the review.
Julie Brown undertook searches, selection of studies, data extraction, contact with authors and quality assessment and commented on the final version of the review.
Jane Marjoribanks undertook selection of studies, data extraction, quality assessment and preparation of tables and commented on the final version of the review.
Fredi Kronenberg undertook selection of studies, data extraction and quality assessment and commented on both the protocol and the final review.
Helen Roberts provided clinical input and commented on the final version of the review.
John Eden commented on the final version of the review.
Declarations of interest
Anne Lethaby provided advice and suggestions to the author of the unpublished Brazilian study (Bicca 2004) that has been included in this review. She is included as an author of that unpublished paper.
Sources of support
- Department of Obstetrics and Gynaecology, University of Auckland, New Zealand.
- No sources of support supplied
Medical Subject Headings (MeSH)
MeSH check words
* Indicates the major publication for the study