Abstract

Aim To update the available evidence on the efficacy and safety of complementary medicines to assist in weight loss by conducting a systematic review and meta-analysis of herbal medicines for weight loss.Methods Four electronic databases (Medline, Embase, CINAHL and Web of Science) were searched from inception until August 2018. A total of 54 randomized placebo-controlled trials of healthy overweight or obese adults were identified. Meta-analyses were conducted for herbal medicines with >= 4 studies available. Weight differences of >= 2.5 kg were considered clinically significant.Results As a single agent, only Phaseolus vulgaris resulted in a statistically significant weight loss compared to placebo, although this was not considered clinically significant. No effect was seen for Camellia sinensis or Garcinia cambogia. Statistically, but not clinically, significant differences were observed for combination preparations containing C. sinensis, P. vulgaris or Ephedra sinica. Of the herbal medicines trialled in <= 3 randomized controlled trials, statistically and clinically significant weight loss compared to placebo was reported for Irvingia gabonensis, Cissus quadrangularis, and Sphaeranthus indicus combined with Garcinia mangostana, among others, but these findings should be interpreted cautiously because of the small number of studies, generally poor methodological quality, and poor reporting of the herbal medicine interventions. Most herbal medicines appeared safe for consumption over the short duration of the studies (commonly <= 12 weeks). Some warrant further investigation to determine effect size, dosage and long-term safety.Conclusion There is currently insufficient evidence to recommend any of the herbal medicines for weight loss included in the present review.


Authors

Maunder, Alison;  Bessell, Erica;  Lauche, Romy;  Adams, Jon;  Sainsbury, Amanda;  Fuller, Nicholas R.

Publons users who've claimed - I am an author

No Publons users have claimed this paper.

Contributors on Publons
  • 3 reviewers
  • pre-publication peer review (FINAL ROUND)
    Decision Letter
    2020/01/19

    19-Jan-2020

    Dear Ms. Bessell

    Thank you very much for submitting this revised manuscript. Following further peer review, I am pleased to tell you that it is now acceptable for publication in Diabetes, Obesity and Metabolism.

    The journal currently has an impact factor of 6.13 and is currently ranked 16th out of 145 journals in the Endocrine Category. Congratulations!

    This journal has recently begun a pilot of ‘transparent peer review’, which means that the anonymous peer reviewer comments and your point-by-point responses to those comments will be made accessible to readers when your paper is published online. Indeed the peer review trail will carry its own separate DOI number which allows the document to be cited. I am sure you will agree that making our peer review process and editorial decision-making, prior to final acceptance, more transparent will be welcomed by the wider scholarly community. Your support of this initiative is much appreciated.

    Your article cannot be published until the publisher has received the appropriate signed license agreement. Within the next few days, as the corresponding author, you will receive an email from Wiley’s Author Services system which will ask you to log in and complete the appropriate license.

    Diabetes, Obesity and Metabolism is now covered by Wiley’s Accepted Article service. This service ensures that accepted 'in press' manuscripts are published online very soon after acceptance, prior to copy-editing or typesetting. Accepted Articles appear in PDF-only format, without the accompanying full-text HTML. Each manuscript is assigned a Digital Object Identifier (DOI), which allows the article to be cited and tracked before it is allocated to an issue. After print publication, the DOI remains valid and can continue to be used to cite and access the article.

    Please note that it is the responsibility of the authors to check the accuracy of all numerical data included in their article at proof stage, as this will be the final opportunity to do so before publication.

    You will receive proofs as an Acrobat pdf file via email in approximately 3 weeks. Your email server must be able to accept attachments up to 4mb in size. Acrobat Reader will be required in order to read this file. This software can be downloaded (free of charge) from the following web site:

    http://www.adobe.com/products/acrobat/readstep2.html.

    All articles received for production can be tracked online by the corresponding author and you can access this facility at www.blackwellpublishing.com/bauthor.

    We thank you for supporting Diabetes, Obesity and Metabolism and we look forward to hearing from you in the future.

    Kind regards,

    Prof. Richard Donnelly
    Editor-in-Chief, Diabetes, Obesity and Metabolism

    P.S. We believe your images might be appropriate for use on the cover of the journal. This is an optional, premium service that aims to increase exposure and showcase your research through a different medium. The cost of this service is $1,800 for a front cover, which will be charged to you if your Cover Image is selected to be featured. If you would like to submit images from your paper, or an alternative original image related to the work, please email your suggestions to covers@wiley.com for consideration. Please see our Cover Image FAQ https://authorservices.wiley.com/author-resources/Journal-Authors/Promotion/journal-cover-image.html for details on Cover Image preparation and submission. Waivers and discounts are available, following the Wiley Open Access guidelines based on authors’ location: https://authorservices.wiley.com/open-science/open-access/for-authors/waivers-and-discounts.html.

    P.P.S. You can help your research get the attention it deserves! Check out Wiley’s Promotion Guide for best-practice recommendations for promoting your work at www.wileyauthors.com/maximize.

    Decision letter by
    Cite this decision letter
    Reviewer report
    2020/01/19

    Dear Authors,
    Thank you for your recent revisions, which have taken into account my previous comments. Overall, the manuscript has improved with these recent changes; I have no further comments.

    Reviewed by
    Cite this review
    Reviewer report
    2020/01/15

    The authors have addressed and responded to all the queries from reviewers and can be considered for publication in its current form.

    Cite this review
    Reviewer report
    2020/01/14

    The authors have answered the questions raised by the reviewer. I have no further comments.

    Reviewed by
    Cite this review
    Author Response
    2020/01/14

    Erica Bessell
    PhD Candidate
    The Boden Collaboration for Obesity, Nutrition, Exercise & Eating Disorders
    14th January 2020

    Dear Prof Donnelly,

    We would again like to re-submit the attached manuscript DOM-19-0849-RA.R1 entitled “Effectiveness of herbal medicines for weight loss: a systematic review and meta-analysis of randomised controlled trials”, to be considered for publication in Diabetes, Obesity and Metabolism.

    We have vastly shortened the length of this manuscript in an attempt to meet the journal requirement for reviews of 5000 words. We have reduced the manuscript from 8113 words to 6090 words. We feel it is not possible to reduce the manuscript further below its current length without compromising the clarity of this large but also very important piece of work which provides valuable insight into the field of herbal medicines. The reduction has been achieved through summarising sections of the results which are already detailed in the figures and supplementary tables, discussing the results more concisely in the discussion, and in some instances, rewording the text to eliminate unnecessary words. The sensitivity analyses have also been completely removed from the manuscript as we felt they did not add much value.

    We again thank the Reviewers for their constructive comments. We have addressed the Reviewers’ responses with a point-by-point reply to each comment including a description of changes made to the manuscript or provided a suitable rebuttal to any comment that was not addressed. The changes have been highlighted in the revised manuscript.

    We look forward to hearing back from you regarding this manuscript.

    Yours sincerely,

    Erica Bessell, Alison Maunder, Romy Lauche, Jon Adams, Amanda Sainsbury, & Nicholas R Fuller

    Responses to Reviewers' comments:

    Referee: 1

    Comments to the Author
    The authors have answered the question raised by the reviewer. I have no further comments.

    Referee: 2

    Comments to the Author
    An interesting study of Herbal medicines for weight loss. More detailed appraisal of quality has greatly improved the rationale of the findings however a little more work is needed.

    Most has to do with explaining the sensitivity analyses and the definition of clinical significance.

    Other suggested edits focus on minor language changes around study methodology. A description of the limitations of the methodology and the findings of the systematic review would add to the discussion, improve transparency and guide future research.

    Page 4

    1. Line 3 please insert ‘may’ – have fewer side effects (the systematic review did not compare the side effects of herbal medicines with conventional treatments and this statement is not accurate as it is). Also not consistent with lines 23-23 stating there is no evidence for safety.

    2. Line 31 without evaluation ‘of’ efficacy not ‘for’ efficacy

    Response: The above changes have been made at Page 3 Line 15 and Line 23.

    Page 5

    1. Line 37-44 The systematic review included randomised control trials investigating herbal medicines against placebo controls for improved weight loss (? in overweight individuals).

    Please note that ‘placebo controls’ have a different aim to ‘blinding’ in clinical trial design. The purpose of using placebo controls is to control the placebo effect, which is relevant when the aim of the study is to compare the difference between two groups and estimate the size of treatment effects or safety of an intervention. On the other hand the purpose of blinding is to control the expectations of study participants, clinicians’ administrators and researchers and prevent bias from subjective assessments of outcome measures – so the type of outcome measure and its subjectivity, is relevant in the appraisal of performance bias. For example participants or researchers expectations are unlikely to influence highly objective outcome measures (mortality for example) so blinding does not add any rigor to the trial.

    Suggest removing the word ‘blinding’ here, but keep it in the assessment of risk of bias.

    The ‘gold standard’ of any study is the one that answers the research question most accurately. The use of ‘gold standard’ is a value statement describing double blinded RCTs with placebo controls as being the highest value – but they are not necessarily the best for estimating the efficacy of herbal medicines for weight loss due to the implausibility of blinding participants to lifestyle (such as dietary energy intake and energy expenditure). These are very likely to confound the effects of the treatments. Suggest removing the reference to gold standard particularly when the study eligibility criteria included open label or unblinded studies.

    Response: The recommended changes have been incorporated and the text now reads as follows:
    ‘This type of trial reduces bias by controlling for participant differences through randomisation and the parallel-group design, and controls for the placebo effect through the use of a control group taking a placebo supplement.’ (Page 5, Line 2-4)

    Page 8

    1. Line 42-48 The estimation of clinical significance requires further articulation. The logic of RCTs comparing the efficacy of an intervention against placebo controls relies on the between group differences rather than changes in one group over time. A loss of 2.5kg before and after is not valid for the following reasons:
    2. Measuring change over time in one group removes the effect of randomisation and the control of participant characteristics (including variations in dietary intake and exercise)
    3. The length of time is not defined – 2.5 kg weight loss over two weeks has different clinical significance to 2 years (for example)
    4. Meta analyses on RevMan is calculated as the differences between groups at endpoint rather than within group differences over time.

    Suggest changing to the following: ‘clinical significance was defined as a 2.5kg or more difference between herbal medicine and placebo controls at endpoint.’ – which is what was described further on in the text.

    Response: The suggested change has been made at Page 7 Line 27 to Page 8 Line 1.

    Page 8

    1. Line 54. Why was heterogeneity of studies limited to statistical diversity when it could be a consequence of clinical and methodological diversity also?

    Response: The section of the methods referred to in this recommendation describes the interpretation of the heterogeneity statistics produced in the meta-analysis. The heterogeneity statistics estimate the variance of the true effect size and relate this to the observed variance in the effect size. Factors contributing to this variance can include clinical and methodological diversity, as the Reviewer mentions. We have included comments on these types of diversity in studies throughout the manuscript.

    Page 9

    1. Line 5 A sentence or two explaining why the sensitivity analyses was limited to subgroups of studies with low risk of selection bias would improve transparency. Also please specify the type of selection bias – assessment of participants characteristics at baseline for example would be rational, particularly in the inclusion of participants with varied degrees of overweight at baseline. More detailed information about how selection bias was assessed as unclear, considering that no studies were found to be at high risk of bias due to inadequate (or insufficient reporting) and yet some studies did not report baseline characteristics or the differences between groups at baseline.

    Response: The sensitivity analyses have been removed from the manuscript to reduce the word count and because they were not seen to add much value to our results.
    To answer the Reviewer’s query, we used the risk of bias assessment tool from the Cochrane Collaboration to guide our assessments for each risk of bias domain. Selection bias relates to two criteria in this tool, which are random sequence generation and allocation concealment. Assessment of these two criteria relates to information reported in the methods. We assessed a lack of details regarding baseline characteristics as a form of ‘other bias’.

    Page 12

    1. Line 27. Please explain the meaning of less heterogeneity of studies investigating single interventions compared to complex interventions. What was it likely due to?

    Response: This sentence has been amended and now reads:
    ‘…combination preparations were more heterogeneous … potentially due to the variability in the herbal medicine formula and the greater representation of male participants.’ (Page 10 Line 18-19)

    1. Line 29 New paragraph for safety.

    2. Line 54 Suggest replace the word ‘meant’ with ‘increased’

    3. Line 59 Include attrition as a percentage (eg 23-45%)

    Response: These suggested changes have been made. The manuscript now reads ‘inadequate reporting in the studies increased risk of selection bias, performance bias and detection bias’ (Page 11 Line 9-10). The attrition rates for the studies has been added at Line 10-11.

    Page 13

    1. Line 6 Data of studies that did not report differences between groups at endpoint should be accounted for in the methods section (data analyses).

    Response: We have elaborated on this point in the methods section at Page 7 Line 16-18 as follows:
    ‘If these weight values were not reported, attempts were made to obtain the missing data from the RCT authors directly. If there was no further information provided, these studies were excluded from the meta-analysis and risk of ‘other bias’ was considered high.’

    1. Line 29 suggest delete ‘however’ and start a new sentence.

    Response: This has been amended on Page 11 Line 17.

    1. Line 50 Discuss the sensitivity analyses - how were the participants characteristics different between groups (ie how wasn’t the randomisation successful)

    Response: The sensitivity analyses have now been removed from the manuscript.

    Page 14

    1. Line 18 Suggest replace the word ‘conduct’ with ‘methodology’ throughout the results section and include % for attrition

    Response: The word ‘conduct’ has been replaced with ‘methodology’ throughout the manuscript. The % for attrition across the five studies with low risk of attrition bias has been added at Page 12 Line 11.

    1. Line 41 How were data incomplete?

    Response: The manuscript now details the nature of the incomplete data and reads as follows:
    ‘One study was excluded from the meta-analysis due to incomplete data. This study did not report final weights nor provide standard deviations for weight losses, however it did note no statistically significant difference…’. (Page 12 Line 16-18)

    Page 15

    1. Line 8 Suggest changing ‘substantial’ to ‘significant’ because it was.

    Response: This has been amended on Page 12 Line 23.

    1. Line 37 to 38 Please include attrition rates across studies.

    Response: The attrition rates for the studies have been added at Page 13 Line 14-15.

    1. Line 43 How was selection bias assessed if there were no measures comparing groups?

    Response: We used the risk of bias assessment tool from the Cochrane Collaboration to guide our assessments for each risk of bias domain. Selection bias relates to two criteria in this tool, which are random sequence generation and allocation concealment. Assessment of these two criteria relates to information reported in the methods. A lack of p-values for baseline or outcome comparisons (as reported on Page 13 Line 17) relates mostly to reporting bias (which assesses risk of selective reporting) and the domain ‘other bias’. In this case, the lack of p-values was a contributing factor towards assessing many of the studies on Ephedra sinica as high risk from ‘other bias’.

    Page 16

    1. Line 8 included not including

    2. Line 14 Suggest delete the word ‘data’ to ‘reported safety’.

    Response: The first suggested change is no longer applicable as it was part of the sensitivity analyses which have since been removed. We have changed ‘reported safety data’ to ‘reported on safety outcomes’ at Page 13 Line 22.

    1. Line 18-22 Was there a difference between groups? If there was no difference between group’s, it means that Ephedra has no more adverse effects than placebo. A comment would be welcome here. What was the nature of adverse effects found for Ephedra.

    Response: The text has been amended at Page 13 Line 25-26 as follows: ‘…with no difference between intervention and placebo (77% vs 76% of participants affected).’
    A comment on the nature of the adverse effects of Ephedra has also been added: ‘Symptoms included palpitations, hypertension and various gastrointestinal side-effects.’ (Page 14 Line 1-2)

    Page 17

    1. Line 33 delete ‘RCT’

    Response: This has been deleted at Page 15 Line 1.

    Page 18

    1. Line 3-5 Inadequate reporting of baseline characteristics – especially the differences between groups at baseline leads to inadequate reporting and high risk of selection bias rather than an unclear risk of selection bias. Selection bias is about the effectiveness of the randomisation in producing two groups without any significantly different characteristics. If they are the same at baseline but different at endpoint – the only logical explanation is the intervention. This is the logical basis of RCTs.

    Response: The Cochrane risk of bias assessment tool states random sequence generation and allocation concealment as the two criteria to consider when determining risk of selection bias, both of which relate to the ability of investigators or participants to influence group allocation. We assessed studies to have an unclear risk of selection bias when these criteria were not adequately reported. On the other hand, inadequate reporting of baseline characteristics contributed to our assessment of high risk from ‘other bias’, due to the risk of attributing a difference at endpoint to the herb when in fact the difference may also have existed at baseline.

    1. Line 56 Suggest remove ‘Well conducted’. Blinding is related to performance bias.

    Response: The suggested amendment has been made and the text now reads: ‘One study was rated at low risk of selection bias, attrition bias and other biases’. (Page 16 Line 10)

    Page 19

    1. Line 29 Please re-write this sentence.

    Response: This sentence has been changed to the following: ‘Eight single-herb products and nine combination preparations were investigated in only one RCT each’. (Page 16 Line 20-21)

    1. Line 35 please change the word ‘conduct’ to ‘methodology’

    Response: This change has been made throughout the manuscript.

    Page 20

    1. Line 5 Suggest change the word ‘Oriental’ to ‘traditional Chinese herbal medicine’

    Response: The word ‘oriental’ has been changed to ‘traditional herbal medicine’ on Page 17 Line 7-8 and Line 12. The word ‘Chinese’ was not included as the two products referred to are Iranian and Korean, respectively.

    1. Line 12 Please name the herbal medicines with significant effects for weight loss and those that do not.

    Response: These lists have now been added to the manuscript. To avoid listing all the herbal medicines twice, they are no longer listed in the paragraph above. The manuscript now reads:
    ‘Statistically significant weight differences favouring the intervention compared to placebo were reported in the RCTs investigating Aster spathulifolius Maxim (73), Evodia rutaecarpa (60), Garcinia atroviridis (74), Gynostemma pentaphyllum (75), Zingiber officinale Roscoe (76), Punica granatum (77), Dolichos biflorus and Piper betle (78), Scutellaria baicalensis and Platycodon grandiflorum (79), Imperata cylindrica (80), Lippia citriodora (81, 82) and the traditional herbal medicine Triphala (83). Four of these studies reported a weight difference compared to placebo which met the benchmark for clinical significance of -2.5 kg, ranging between -2.5 kg and -3.5 kg (73, 77, 78, 83). No effect on weight was found for Brassica rapa L. (84), Caralluma fimbriata (85), Glycine max (47), Citrus aurantium and Hypericum perforatum (86), Magnolia officinalis and Phellodendron amurense (87) and the traditional herbal medicine Taeeumjowi-tang (88).’ (Page 17 Line 3-12)

    1. Line 44 Suggest remove the word ‘furthermore’. Suggest ‘Results need to be interpreted with caution due to high risks of bias found in at least one domain for all studies. Nearly all studies were found to be at high risk of bias due to inadequate reporting, and insufficient calculation of sample sizes (which should be mentioned earlier in the methods section)

    Response: The manuscript has been amended and now reads as follows, at Page 18 Line 2-4:
    ‘The results need to be interpreted cautiously as nearly all RCTs were found to be at high risk of bias in at least one domain, notably due to insufficient detail of the herbal medicine and inadequate reporting of the trial design and methodology.’

    Page 21

    1. Line 6. Say ‘verification of herbal ingredients’ if this is what is meant by quality assessment.

    Response: This phrase was removed when reducing the length of the manuscript.

    1. Line 16 Adequate sample size should be discussed in the methods section and how this was assessed.

    Response: I believe the comment being referred to in the above recommendation is ‘too small to be clinically important’. This comment from the discussion was referring to the amount of weight loss, not the size of the studies. Additionally, in making the discussion more concise, this phrase has now been removed. We would like to note that we assessed whether the included studies reported on power calculations in the risk of bias analysis (contributing to ‘other bias’). In the conclusion, we state that ‘many of the included studies were small’; this is more a comment on the fact that most trials were small with the median sample size being 68 participants.

    1. Line 50-54 Suggest delete ‘was seen when restricted to two small studies with low risk of ……’ as it repeats the results. Suggest replace with – ‘a larger weight-loss effect of Garcinia was found in the sensitivity analyses ……’

    Response: This suggestion is no longer applicable as the sensitivity analyses have been removed from the manuscript to reduce its length.

    Page 22

    1. Line 12 change ‘is’ to ‘was’

    Response: This has been changed on Page 18 Line 12.

    Page 23

    1. Line 14 Suggest insert … Very high risk of selective reporting and confirmation bias.

    Response: The sentence has been amended at Page 19 Line 5-9, and now reads:
    ‘However, these effects were observed in RCTs of only 8 to 10 weeks’ duration, and all of the two to three studies conducted on each of these herbal medicines were by the same research groups, which introduces a high risk of selective reporting and confirmation bias, so more evidence is needed before a definitive decision on their effect on weight loss can be reached.’

    Page 24

    1. Discussion of risk of bias is repetitive and should be reported in the first part of the results section. Suggest delete.

    Line 16 the sentence starting with … Only 22%of studies through to the sentence ending is not present, could be deleted due to repetition. The following sentence describing the clinical methodological and statistical heterogeneity could be moved to the beginning of the discussion following the risk of bias discussion.

    Response: The recommended section of text has been deleted from the discussion. Part of the section of text which followed has also been deleted to reduce the length of the manuscript. The rest has been moved to the first paragraph of the discussion (Page 18 Line 4-7).

    1. The discussion requires more information about the limitations of the systematic review, including the methods and the findings. For example, the limitations of not including papers in languages other than English and the limitations of including studies only comparing herbal medicines against placebo controls and not with pharmaceutical interventions for example. How generalisable are the findings?

    Response: We have elaborated on the paragraph regarding limitations of our systematic review at Page 19 Line 24 to Page 20 Line 6, as follows:
    ‘The findings of this review are only applicable in the contexts in which the studies were conducted and are limited by their quality and variability…
    Only English-language publications were included, which may restrict the representation of research from non-English speaking countries and limit the generalisability of the findings. Any non-published negative studies were naturally not identified; however, these studies would not alter the conclusions as no herbal medicines were found to be unequivocally beneficial for weight loss.’

    1. The conclusion could be more succinct and describe the quality of the evidence and summarise the findings.

    Response: The conclusion has been simplified and now reads as follows at Page 20 Line 9-18:
    ‘Many of the herbal medicines meta-analysed produced statistically significant weight loss compared to placebo, but these effects were below the level of weight loss that is considered clinically important. Some herbal medicines warrant further investigation in larger more rigorous studies to determine the effect size, dosage and long-term safety, notably Cissus quadrangularis, Sphaeranthus indicus and Irvingia gabonensis, and some products only represented by one study. Many of the included studies were small, of poor design and methodological quality, with inadequate reporting of the herbal medicine interventions. Future RCTs would benefit from trial registration, and ensuring the study is conducted and reported in a way that minimises bias and conforms with the CONSORT Statement for reporting of clinical trials (31). Currently there is insufficient evidence to recommend any of these herbal medicines for weight loss.’

    In the figures.

    1. An explanation of the Forest plots is not needed as most readers of JCEM already understand these including symbols etc.

    Response: The explanations of the Forest plots have been removed from each figure.

    Referee: 3

    Comments to the Author
    The authors have addressed most of the reviewers' comments with relevant responses in the revised submission.

    However, the authors did not respond to the query where it was suggested to rework on the search strategy and also screening of studies for inclusion in order to not miss any relevant study which meets inclusion criteria.

    The mention of Ebrahimzadeh Attari 2015 (41) study is an example which the authors missed to include (in the previous submission), but not the complete list of studies.

    The authors have to double-check the literature hits, if required, to make sure that they did not miss any relevant study for inclusion.

    Response: We are confident that our search and screening of studies was conducted thoroughly and was sufficient to identify all studies which met our specific inclusion criteria. To test this, we cross-checked the studies identified in our search against studies included in other similar reviews of herbal medicines to ensure we had captured the studies they included. Often times, our inclusion criteria were more strict than the other reviews.
    We would like to highlight that the Ebrahimzadeh Attari 2015 publication was identified in our search. We had included another publication in our review instead of the Ebrahimzadeh Attari 2015 publicaiton, as both publications reported on the same trial.

    Referee: 4

    Comments to the Author
    Dear Authors,

    Thank you for your revision, which has improved the document. I would, however, like to provide the following comments below, following my review of the latest version of the manuscript:-

    1. Page 2: Abstract. RESULTS. Please add…. As a single agent, only Phaseolus vulgaris resulted in… I feel that the addition of the words ‘As a single agent’ is important, because later on, the authors discuss about combination therapies that can produce statistically significant results.

    2. Page 2: Abstract. RESULTS. Of the herbal medicines trialled in < 3 randomised controlled trials…. Garcinia mangostana, among others, but these findings should be interpreted cautiously due to small number of studies, generally poor methodological quality….

    I suggest combining the 2 sentences above, because the latter sentence clearly only refers to the herbal medicines trialled in < 3 randomised controlled trials, and not related to meta-analysis for other single/combination agents.

    Response: These suggested changes to the abstract have been made at Line 10 and Line 16-17.

    1. Page 6. Top paragraph. It states that studies with ‘participants with specified medical co-morbidities’ were excluded. Please state exactly which specific medical comorbidities were excluded. Was it metabolic syndrome, dyslipidaemia, hypertension, pre-diabetes and diabetes? There is a lot of overlap of obesity with these conditions. If so, then potentially quite a few studies may have been excluded, as many of these obese/overweight patients will have these co-morbidities. Studying the effects of weight loss in this setting (with background co-morbidities) is more likely to provide ‘real world’ data. Studies which actually provide information on background co-morbidities may be publications that are more detailed (and should have been included in the analysis), because investigators actually do take time to obtain detailed medical history at baseline.

    Response: So that the results of this systematic review were as applicable as possible to the general population of people with overweight and obesity, we excluded studies which specifically recruited only participants with a particular comorbidity. The manuscript has been changed as follows:
    ‘…if the included participants were required to have a specific comorbidity. This did not preclude the possibility that the included participants may have had comorbidities, as is common in populations with overweight or obesity.’ (Page 5 Line 12-14).

    1. Page 8. The threshold for clinically significant weight loss is stated as -2.5 kg, and a specific reference is provided [12] - to support this statement. However, the reference provided is for a study conducted in Iran which showed that taking dietary supplements containing green tea, capsaicin and ginger resulted in a significant decrease in weight (-1.8 ± 1.5 vs. +0.4 ± 1.2 kg, respectively, p < 0.001). I could not find information from this study to support the point that -2.5kg weight loss reduced incidence of CV events and mortality. Please provide the correct reference.

    Response: We have checked the references throughout the manuscript relating to the threshold for clinically significant weight loss to ensure the reference is for the SCOUT trial (Caterson 2012). We believe the Reviewer may have been looking at the reference list for the supplementary material, where reference 12 is Taghizadeh 2017 (RCT on green tea). The correct reference from the reference list of the main manuscript is on Page 28 (note: now reference 11).

    1. Page 16. Irvingia gabonensis (African Mango). The authors have stated that the studies conducted in Cameroon, ‘were rated as high risk from other biases’. However, I feel that it would be very important to highlight large differences in patient baseline characteristics between active vs. placebo groups. For example, the baseline characteristics of the study conducted with 300 mg Irvingia gabonensis shows that the mean body weight in the placebo arm was 79.43 kg, while the mean body weight in the active arm was 105.10 kg! I feel that the authors need to make the readers aware of these discrepancies, which would have likely biased the overall results for this study.

    Response: These discrepancies have now been highlighted in the manuscript at Page 14 Line 12-15:
    ‘Notably, one study reported a difference in mean body weight at baseline of >25 kg between the intervention and placebo group (66), and another reported a difference of >1600 kJ in mean energy intake during the study between the intervention and placebo group (65).’

    1. Page 17. Cissus quadrangularis. It was my impression that the publication by Oben et al 2006 (The use of a Cissus quadrangularis formulation in the management of weight loss and metabolic syndrome) provided adequate information on baseline patient characteristics (see table in original publication), yet the authors state ‘inadequate reporting of baseline characteristics’. Please therefore justify this statement.

    Response: This has been clarified as the statement was referred to the publication by Oben 2008, which did not provide details of baseline characteristics stratified by treatment group. The text now reads: ‘…one study inadequately reported baseline characteristics (67).’ (Page 15 Line 15)



    Cite this author response
  • pre-publication peer review (ROUND 2)
    Decision Letter
    2019/11/20

    20-Nov-2019

    Dear Ms. Bessell

    Your revised paper has now been re-evaluated by one of the editors and two or more expert referees. Their comments to the authors are available at the bottom of this letter, alternately, you can log onto the manuscriptcentral website at http://mc.manuscriptcentral.com/dom to view their feedback.

    The revised version is far from acceptable. The reviewers have made various comments. In addition, the entire length of this manuscript needs to be drastically cut - to a maximum of 3,500 words for the main body of the text. This will require much of the detail to be moved from the main body to the Supplementary Appendix. Large tables also need to go into the Appendix.

    I am prepared to re-evaluate one further version. Unfortunately, if the next version does not meet with the referees' expectations & the above comments the manuscript will be rejected. Currently it is no where near being sufficiently competitive.

    The reviewers and editor would appreciate the opportunity to re-evaluate a revised manuscript which addresses the points raised. The revision should be accompanied by a separate document providing a point by point response to the reviewers’ comments and identifying where significant changes have been incorporated in the text.

    The revised manuscript will be re-assessed by the original reviewers and the handling editor. Only those manuscripts which are full competitive will be accepted for publication, therefore at this stage I cannot guarantee ultimate acceptance.

    Please use the link below to create and submit your revision online with no need to enter login details. Also, please ensure that you upload a WORD DOCUMENT and not a .pdf AND that you re-submit all figures and tables (whether they were revised or not).

    PLEASE NOTE: This is a two-step process. After clicking on the link, you will be directed to a webpage to confirm.

    https://mc.manuscriptcentral.com/dom?URL_MASK=845b23bd99e745cbb778f8485c697e9d

    If you feel that your paper could benefit from English language polishing, you may wish to consider having your paper professionally edited for English language by a service such as Wiley’s at http://wileyeditingservices.com. Please note that while this service will greatly improve the readability of your paper, it does not guarantee acceptance of your paper by the journal.

    You have two months from today to provide your revision: if you need more time then please contact the Editorial Office at susan.lane@nottingham.ac.uk. Automatic reminders will be sent to you from the system.

    Kind regards,

    Prof. Richard Donnelly
    Editor-in-Chief, Diabetes, Obesity and Metabolism

    Referee: 1

    Comments to the Author
    The authors have answered the question raised by the reviewer. I have no further comments.

    Referee: 2

    Comments to the Author
    An interesting study of Herbal medicines for weight loss. More detailed appraisal of quality has greatly improved the rationale of the findings however a little more work is needed.

    Most has to do with explaining the sensitivity analyses and the definition of clinical significance.

    Other suggested edits focus on minor language changes around study methodology. A description of the limitations of the methodology and the findings of the systematic review would add to the discussion, improve transparency and guide future research.

    PLEASE SEE ATTACHED DOCUMENT

    Referee: 3

    Comments to the Author
    The authors have addressed most of the reviewers' comments with relevant responses in the revised submission.

    However, the authors did not respond to the query where it was suggested to rework on the search strategy and also screening of studies for inclusion in order to not miss any relevant study which meets inclusion criteria.

    The mention of Ebrahimzadeh Attari 2015 (41) study is an example which the authors missed to include (in the previous submission), but not the complete list of studies.

    The authors have to double-check the literature hits, if required, to make sure that they did not miss any relevant study for inclusion.

    Referee: 4

    Comments to the Author
    Dear Authors,

    Thank you for your revision, which has improved the document. I would, however, like to provide the following comments below, following my review of the latest version of the manuscript:-

    Page 2: Abstract. RESULTS. Please add…. As a single agent, only Phaseolus vulgaris resulted in… I feel that the addition of the words ‘As a single agent’ is important, because later on, the authors discuss about combination therapies that can produce statistically significant results.

    Page 2: Abstract. RESULTS. Of the herbal medicines trialled in < 3 randomised controlled trials….. Garcinia mangostana, among others, but these findings should be interpreted cautiously due to small number of studies, generally poor methodological quality….
    I suggest combining the 2 sentences above, because the latter sentence clearly only refers to the herbal medicines trialled in < 3 randomised controlled trials, and not related to meta-analysis for other single/combination agents.

    Page 6. Top paragraph. It states that studies with ‘participants with specified medical co-morbidities’ were excluded. Please state exactly which specific medical comorbidities were excluded. Was it metabolic syndrome, dyslipidaemia, hypertension, pre-diabetes and diabetes? There is a lot of overlap of obesity with these conditions. If so, then potentially quite a few studies may have been excluded, as many of these obese/overweight patients will have these co-morbidities. Studying the effects of weight loss in this setting (with background co-morbidities) is more likely to provide ‘real world’ data. Studies which actually provide information on background co-morbidities may be publications that are more detailed (and should have been included in the analysis), because investigators actually do take time to obtain detailed medical history at baseline.

    Page 8. The threshold for clinically significant weight loss is stated as -2.5 kg, and a specific reference is provided [12] - to support this statement. However, the reference provided is for a study conducted in Iran which showed that taking dietary supplements containing green tea, capsaicin and ginger resulted in a significant decrease in weight (-1.8 ± 1.5 vs. +0.4 ± 1.2 kg, respectively, p < 0.001). I could not find information from this study to support the point that -2.5kg weight loss reduced incidence of CV events and mortality. Please provide the correct reference.

    Page 16. Irvingia gabonensis (African Mango). The authors have stated that the studies conducted in Cameroon, ‘were rated as high risk from other biases’. However, I feel that it would be very important to highlight large differences in patient baseline characteristics between active vs. placebo groups. For example, the baseline characteristics of the study conducted with 300 mg Irvingia gabonensis shows that the mean body weight in the placebo arm was 79.43 kg, while the mean body weight in the active arm was 105.10 kg! I feel that the authors need to make the readers aware of these discrepancies, which would have likely biased the overall results for this study.

    Page 17. Cissus quadrangularis. It was my impression that the publication by Oben et al 2006 (The use of a Cissus quadrangularis formulation in the management of weight loss and metabolic syndrome) provided adequate information on baseline patient characteristics (see table in original publication), yet the authors state ‘inadequate reporting of baseline characteristics’. Please therefore justify this statement.

    Decision letter by
    Cite this decision letter
    Reviewer report
    2019/11/19

    Dear Authors,

    Thank you for your revision, which has improved the document. I would, however, like to provide the following comments below, following my review of the latest version of the manuscript:-

    Page 2: Abstract. RESULTS. Please add…. As a single agent, only Phaseolus vulgaris resulted in… I feel that the addition of the words ‘As a single agent’ is important, because later on, the authors discuss about combination therapies that can produce statistically significant results.

    Page 2: Abstract. RESULTS. Of the herbal medicines trialled in < 3 randomised controlled trials….. Garcinia mangostana, among others, but these findings should be interpreted cautiously due to small number of studies, generally poor methodological quality….
    I suggest combining the 2 sentences above, because the latter sentence clearly only refers to the herbal medicines trialled in < 3 randomised controlled trials, and not related to meta-analysis for other single/combination agents.

    Page 6. Top paragraph. It states that studies with ‘participants with specified medical co-morbidities’ were excluded. Please state exactly which specific medical comorbidities were excluded. Was it metabolic syndrome, dyslipidaemia, hypertension, pre-diabetes and diabetes? There is a lot of overlap of obesity with these conditions. If so, then potentially quite a few studies may have been excluded, as many of these obese/overweight patients will have these co-morbidities. Studying the effects of weight loss in this setting (with background co-morbidities) is more likely to provide ‘real world’ data. Studies which actually provide information on background co-morbidities may be publications that are more detailed (and should have been included in the analysis), because investigators actually do take time to obtain detailed medical history at baseline.

    Page 8. The threshold for clinically significant weight loss is stated as -2.5 kg, and a specific reference is provided [12] - to support this statement. However, the reference provided is for a study conducted in Iran which showed that taking dietary supplements containing green tea, capsaicin and ginger resulted in a significant decrease in weight (-1.8 ± 1.5 vs. +0.4 ± 1.2 kg, respectively, p < 0.001). I could not find information from this study to support the point that -2.5kg weight loss reduced incidence of CV events and mortality. Please provide the correct reference.

    Page 16. Irvingia gabonensis (African Mango). The authors have stated that the studies conducted in Cameroon, ‘were rated as high risk from other biases’. However, I feel that it would be very important to highlight large differences in patient baseline characteristics between active vs. placebo groups. For example, the baseline characteristics of the study conducted with 300 mg Irvingia gabonensis shows that the mean body weight in the placebo arm was 79.43 kg, while the mean body weight in the active arm was 105.10 kg! I feel that the authors need to make the readers aware of these discrepancies, which would have likely biased the overall results for this study.

    Page 17. Cissus quadrangularis. It was my impression that the publication by Oben et al 2006 (The use of a Cissus quadrangularis formulation in the management of weight loss and metabolic syndrome) provided adequate information on baseline patient characteristics (see table in original publication), yet the authors state ‘inadequate reporting of baseline characteristics’. Please therefore justify this statement.

    Reviewed by
    Cite this review
    Reviewer report
    2019/11/17

    The authors have addressed most of the reviewers' comments with relevant responses in the revised submission.

    However, the authors did not respond to the query where it was suggested to rework on the search strategy and also screening of studies for inclusion in order to not miss any relevant study which meets inclusion criteria.

    The mention of Ebrahimzadeh Attari 2015 (41) study is an example which the authors missed to include (in the previous submission), but not the complete list of studies.

    The authors have to double-check the literature hits, if required, to make sure that they did not miss any relevant study for inclusion.

    Cite this review
    Reviewer report
    2019/11/15

    An interesting study of Herbal medicines for weight loss. More detailed appraisal of quality has greatly improved the rationale of the findings however a little more work is needed.

    Most has to do with explaining the sensitivity analyses and the definition of clinical significance.

    Other suggested edits focus on minor language changes around study methodology. A description of the limitations of the methodology and the findings of the systematic review would add to the discussion, improve transparency and guide future research.

    PLEASE SEE ATTACHED DOCUMENT

    Reviewed by
    Cite this review
    Reviewer report
    2019/11/12

    The authors have answered the question raised by the reviewer. I have no further comments.

    Reviewed by
    Cite this review
    Author Response
    2019/11/11

    Erica Bessell
    PhD Candidate
    The Boden Collaboration for Obesity, Nutrition, Exercise & Eating Disorders
    11th November 2019

    Editor
    Richard Donnelly
    Diabetes, Obesity and Metabolism

    Dear Prof Donnelly,

    I would like to re-submit the attached manuscript DOM-19-0849-RA entitled “Effectiveness of herbal medicines for weight loss: a systematic review and meta-analysis of randomised controlled trials”, to be considered for publication in Diabetes, Obesity and Metabolism.

    We thank the Reviewers for their constructive comments which we believe have strengthened the reporting of the systematic review and meta-analysis. We have addressed the Reviewers’ responses with a point-by-point reply to each comment including a description of changes made to the manuscript or provided a suitable rebuttal to any comment that was not addressed. The changes have been highlighted in the revised manuscript.

    I look forward to hearing back from you regarding this manuscript.

    Yours sincerely,

    Erica Bessell

    Responses to Reviewers' comments:

    Referee: 1

    Comments to the Author
    Interesting study summarising the evidence of efficacy for herbal medicine against placebo for weight loss.

    There are some basic requirements that should be addressed including the use of inaccurate terms and inaccurate reporting of the quality of studies.

    More information about the assessment of risk of bias is needed, and limitations of the study would improve transparency and rigour.

    Page 3

    1. Lines 18-23 Evidence for lifestyle intervention is low quality due to high risk of attrition bias and loss of benefits of randomisation. Describing this would improve the rationale for synthesising the evidence for herbal medicine for weight loss.

    Response: Risk of bias in lifestyle interventions is now mentioned as follows:
    ‘While research shows increased physical activity and decreased energy intake are key components of the treatment of obesity, lifestyle intervention studies are subject to a high risk of bias from lack of allocation concealment and attrition bias, which reduces the benefits of randomisation (5).’

    1. Line 37 more detail about the side effects of pharmaceuticals including the nature and prevalence to explain the context of herbal alternatives.

    Response: This detail has been added as follows:
    ‘For example, a common anti-obesity medication, Orlistat, works by preventing the absorption of dietary fat from the intestine, resulting not only in weight loss, but also gastrointestinal side-effects such as abdominal pain, faecal urgency and unwanted oily faecal discharge (14). The severity of these symptoms is proportional to the amount of dietary fat consumed. A more recently approved anti-obesity medication, Liraglutide, has relatively fewer side-effects, but up to 14% of those taking it still report side-effects, mainly nausea and vomiting (13).’

    Page 4

    1. Line 33 this is a Systematic Review of RCTs, not a critical review.

    2. Line 35 suggest delete ‘contemporary’

    Response: ‘A critical review’ has been amended to ‘a systematic review’. The word ‘contemporary’ has been deleted.

    Page 5

    1. Line 54 -57 More detail about how the consistency between treatment arms was assessed for additional lifestyle interventions such as dietary and exercise interventions. Inconsistency between groups may explain the study results, independent to the herbal medicine (versus placebo controls).

    Response: This has been reworded to read:
    ‘Studies were eligible for inclusion regardless of any additional diet or lifestyle interventions used, as long as the same diet or lifestyle intervention was used in both the herbal intervention and placebo groups. Studies that used a different regimen for diet and lifestyle between the intervention and placebo groups were excluded.’

    Method section

    Quite thorough. More information needed about control groups and about the assessment of study quality.

    1. More detail about only including studies investigating herbal medicine against placebo controls is needed. What was the rationale? How were placebo controls assessed as being true placebo?

    Response: The following has been added to the methods section:
    ‘This particular type of trial is the gold-standard method for determining the effect of the herbal medicine, as it reduces bias by controlling for participant differences through randomisation and the parallel-group design, and controls for the placebo effect through the use of a blinding and having a control group taking a placebo supplement.’
    In terms of the placebo controls used in each study, this information is listed in Supplemental Table S4. The following has also been added to the ‘Study selection’ subsection of the methods: ‘Information on the type of herbal medicine formulation and placebo used was also extracted for each study.’

    1. The method for assessing and determining risk of bias is inadequate. If study authors have not reported against standard criteria (method of randomisation, blinding, analyses, attrition, reporting, outcomes etc), risk of bias is assessed as being high, rather than unclear. This requires revision.

    Response: We have followed the Cochrane Handbook guidelines for risk of bias in our assessments in this review and thus we respectfully disagree with the above point. The Cochrane Handbook notes in Section 8.3.2 (Reporting versus conduct) that ‘it can be tempting to resort to assessing the adequacy of reporting’ rather than ‘the risk of bias in the actual design and conduct of the study’. In an attempt to avoid this, many studies were rated as ‘unclear risk of bias’ for some criteria due to inadequate reporting. We have added mention of this in the discussion as follows:
    ‘…inadequate reporting of allocation concealment and blinding of participants, study personnel, and outcome assessments was common.’
    We have also elaborated on our assessment of risk of bias using the Cochrane Handbook in the methods as follows:
    ‘Studies were considered to have a low risk of bias for a criterion if the report detailed a methodology which was adequate and did not introduce potential bias, as per Table 8.5.d of the Cochrane Handbook (30). Likewise, studies were considered to have a high risk of bias for a criterion if the report detailed a methodology which is known to be inadequate in removing potential bias. Studies were rated as having an unclear risk of bias if there was no information or insufficient information provided to judge the criterion otherwise.’

    1. Line 32 44 Studies administrating a weight loss diet, including a low calorie or meal replacement shakes or exercise sessions or advise for exercise are investigating herbal medicine against placebo in addition to weight-loss lifestyle interventions. Uneven participation between test and control groups for any of these could confound the results of the RCT. More detail about how these risks of bias were assessed is needed.

    Response: The types of problems covered by the ‘other bias’ domain have been elaborated in the methods section to include this as follows:
    ‘Other biases include other major flaws in the trial design or conduct or reporting, conflicts of interest, or differing levels of adherence to the added lifestyle interventions between the intervention and placebo groups (if reported).’

    Results section

    Page 9 comments here should be reported for all interventions in the results section

    1. Line 9 Please insert an overall statement about the quality of the evidence at the start of the results (applies to all interventions).

    Response: The section on risk of bias within studies has been moved to the second paragraph position in the results section. A summary of the quality of studies for each herbal medicine has been included in the first paragraph of each herbal medicine section. An example for C. sinensis follows:
    ‘There was inadequate reporting of trial design and conduct across the studies, which largely prevented assessment of risk of selection bias, performance bias, and detection bias. Six studies were at low risk of attrition bias, while four studies were rated as having high risk. Seven studies were rated as high risk from other biases for reasons including lack of trial registration, potential conflict of interest with one or two authors working for the company owning the herbal medicine, and trials including a run-in period to remove participants with low compliance.’

    1. Line 50 suggest the word ‘investigated’ or ‘examined’ rather than ‘used’ as RCTs assess efficacy not utility and it is misleading. Please change this throughout the results section.

    Response: This has been amended throughout the results section to read ‘investigated’ or ‘examined’ instead of ‘used’.

    1. Include a summary of the quality of studies for each herbal intervention.

    Response: This has been amended as per point number 9.

    1. Line 57 to 59 Averages and medians of study participants in RCTs in meta-analyses are of little value without an estimate of the number needed to treat and of little relevance unless the meta-analyses included insufficient participants. One the other hand, the total number in the meta-analyses is relevant and important. Suggest reporting the total number of study participants in the meta-analyses and/or RCTs (if only one) and remove medians and average numbers of participants in RCTs. (Apply to reporting on all interventions).

    Response: This has been amended throughout the results section. Mean and median number of participants have been removed and only total number of participants for all studies remains. For example, ‘Camellia sinensis was investigated in 12 studies (n = 1179)’.

    Page 10

    1. Line 31 Include the number of participants in the meta-analyses (12 studies, n=xxxx)

    Response: The number of participants in each meta-analysis has been added throughout the results.

    1. More discussion about the risks of bias is needed including the types of placebo controls.

    Response: This has been addressed as per point number 9.

    Page 11

    1. Line 18 ‘investigated’ not ‘used’.

    2. Line 22-25 total number not average and median.

    3. Line 52 total not average and median

    Response: These have been amended as per point numbers 10 and 12.

    1. Line 53-55 Suggest replace ‘when only studies were assessed to have low risk of bias’ with ‘the sensitivity analyses’ which should be explained in the methods section.

    Response: ‘Studies with low risk of selection bias’ has been replaced with ‘the sensitivity analysis’. The method has been amended to clarify the use of sensitivity analyses and now reads:
    ‘Sensitivity analyses were conducted to test the robustness of each meta-analysis. The sensitivity analyses only included those studies which were assessed to have a low risk of selection bias (random sequence generation and allocation concealment).’

    Page 12

    1. Line 10 suggest replace ‘highly’ with ‘significant’ (heterogeneity)

    Response: The text has been amended to read ‘…were observed to have significant heterogeneity’.

    1. Line 25 Please describe adverse effects in more detail – eg unpleasant gastro-intestinal effects or side-effects.

    Response: This now reads ‘one reported increased unpleasant gastrointestinal effects of the intervention compared to placebo’.

    1. Line 31 Investigated or examined not used.

    2. Line 42 -44 Remove average and median participants and insert total for meta-analyses.

    Response: These have been amended as per point numbers 10, 12, and 13.

    Page 13

    1. Line 10 suggest delete ‘robust’ as there is no qualification of this descriptor.

    Response: The word ‘robust’ has been removed and the text now reads ‘Only one study remained in the sensitivity analysis, which reported a change in weight of…’.

    1. Lines 22-35 this paragraph could be more concise with more detail about the nature and duration of GI adverse effects.

    Response: More detail regarding the adverse effects has been included to this paragraph, and the remainder of the paragraph has been amended to be more concise. It now reads as follows:
    ‘All seven studies that investigated Phaseolus vulgaris reported on treatment-related adverse effects, and three studies additionally reported on adverse events, which were minimal and comparable between the intervention and placebo groups (45, 58, 59). Five studies reported no treatment-related adverse effects (54, 57-60), and the remaining two studies reported increased gastrointestinal effects, such as discomfort and flatulence, in the intervention compared to the placebo group (45, 61). The flatulence was reported to resolve within one to three days (61). In one study, a participant from the intervention group withdrew due to constipation (61).’

    Page 14

    1. Line 3 Suggest change ‘the duration of ‘interventions’ to the duration of ‘studies’. Perhaps a simplified inclusion to the previous sentence – ‘over 8 to 26 weeks’ because the time that a treatment was administered is part of dosage.

    Response: This change has been made throughout the results. The text now reads either ‘The duration of studies…’ or this information has been added to the end of the sentence regarding dosage as, for example, ‘administered over 8 to 17 weeks…’.

    1. Line 10 suggest changing ‘(n=4)’ to ‘four studies’ to prevent misunderstanding about the number of participants and the number of studies.

    Response: This has been amended to ‘Meta-analysis of four studies (n = 382) of Ephedra sinica as part of combination-herb preparations…’.

    1. Line 14-16 The SMD sentence belongs in the method section rather than the results.

    Response: This sentence has been moved to the method and now reads:
    ‘SMD was reported in the meta-analyses when the results presented in the included studies were in varying units of weight.’

    Page 15

    1. Line 44 More detail about which biochemical markers please.

    Response: This has been amended to:
    ‘One single-herb trial additionally reported that blood cell count, markers of liver and kidney function, and vital signs remained within the normal range in both the active and placebo groups’.

    Page 16

    1. Line 16 Report the quality of studies

    Response: This has been addressed as per point number 9.

    1. Line 31 Suggest change the word ‘relate’ to ‘attribute the effects to…’

    Response: This has been amended from ‘relate this effect’ to ‘attribute the effects’.

    Page 17

    1. Suggest changing subheading to ‘Evidence from single RCTs for herbal medicine complex preparations’

    Response: The subheading has been amended to ‘Evidence from single RCTs for herbal medicine preparations’. As eight of the 17 studies discussed in this section are of single-herb preparations, we did not include the word ‘complex’ in the subheading.

    Page 18

    1. Line 12 to 16 – Suggest move this to the second paragraph position of the results section. The second sentence describes the method of reporting and should be in the methods section not the results.

    Response: This paragraph on risk of bias within studies has been moved to the second paragraph position in the results section. Figure numbers have been updated accordingly. The second sentence of the paragraph has been moved to the method and now reads:
    ‘The judgements about each risk of bias domain are presented as percentages across all trials, and the judgements about each risk of bias domain for each trial are included as supplemental material.’

    1. Please note if studies have not reported allocation concealment, they are usually judged to be at high risk of bias not unclear risk. Rigorous results depend on the integrity of randomisation to control confounding factors on the results. The authors determination (judgement) of unclear risks of bias requires much more clarification because most authors would determine these to be high risk rather than unclear and what appears to be an underestimation of the low quality of the evidence. This is highlighted with the statement that ‘low risk of selection bias was found for 22% of studies’ which could be more transparently stated as ‘78% of studies were found to be at a high-risk bias’.

    Response: As per point number 7, we respectfully disagree with the Reviewer’s recommendation that RCTs with inadequate reporting of allocation concealment should be judged at high risk of bias, rather than unclear risk. We have clarified the judgements of unclear risk of bias in the manuscript, as per point number 8. We would also like to highlight that ‘unclear risk of bias’ is not interpreted as better than ‘high risk of bias’, because in any sensitivity analyses conducted, they are both excluded. Thus, we have not made changes to the text mentioned in the above point.

    Page 18

    Discussion

    1. Suggest starting with Paragraph: Page 21 line 31 (answer the research question) next section should summarise the quality of evidence.

    2. Suggest sentences that repeat results are removed to retain readers interest.

    Response: The suggested paragraph, which answers the research question and summarises the quality of the evidence, has been moved to the beginning of the discussion. Repetition of the results within the discussion has also been removed.

    1. More about the new information of this Study. A comparative assessment of adverse effects of herbal verses pharmaceuticals would add to the impact of this SR.

    Response: A paragraph addressing this has been added to the discussion as follows:
    ‘This systematic review is the first in 18 years to comprehensively review the literature and provide meta-analyses of the available evidence from randomised, placebo-controlled trials of herbal medicines for weight loss. Across the 54 studies, a wide range of countries and cultures were represented, improving the generalisability of the findings. However, despite the introduction in 2006 of the Elaborated CONSORT Statement extension for herbal medicine interventions (32), many studies were still not adequately reporting on trial design and conduct, or on the composition of the herbal medicines being investigated.
    The results presented here do indicate that herbal medicines may have less side-effects than the pharmaceutical treatments available, and potentially dangerous products are eventually banned, such as the case of Ephedra sinica. It should be noted that rigorous collection and reporting of safety data as required in Phase III trials for pharmaceuticals is not a requirement prior to the marketing of herbal medicines. In terms of the studies included in this systematic review, adverse effects were often sparsely reported and sometimes not reported at all.’

    1. Line 46 include the total number of participants

    Response: The total number of participants across the 54 studies (n = 4331) has been added.

    1. A section about the limitations of the systematic review is needed for transparency and to show that the authors are aware of the limitations of their conclusions.

    Response: A paragraph on limitations has been added to the end of the discussion as follows:
    ‘These findings are limited by the quality and variability of the studies included. Only 22% of studies were rated as having a low risk of bias in two or more categories of the bias assessment tool, and only one study was rated as low risk of bias in four categories. This is compared to 39% of studies rated as having a high risk of bias in two or more categories. Attrition bias was high in a third of the studies, and inadequate reporting of allocation concealment and blinding of participants, study personnel, and outcome assessments was common. The low quality of the studies in terms of risk of bias would serve to decrease the confidence in the conclusions reached if they were favourable, however as most herbs were not found to have a beneficial effect on weight, the risk of attributing a positive effect when none exists is not present. For most herbal medicines, the included studies were of a range of durations, dosages and formulations. This was in part accounted for in the use of random effects models for the meta-analyses, but many of the studies in this review were not meta-analysed. In addition to this, the review included all RCTs of herbal medicines for weight loss, regardless of the additional lifestyle interventions delivered, providing that the same lifestyle intervention was used in both the intervention and placebo groups. These factors may have contributed to the heterogeneity observed in the effect sizes. Additionally, the search strategy employed in this review used language restrictions (English only) to identify published studies. Thus, those published in other languages were excluded, and any non-published negative studies were naturally not identified.’

    1. Recommendations for future research are needed to contextualise this work.

    Response: The following text is contained within the conclusion, including an additional sentence and an original sentence regarding future research:
    ‘Future RCTs would benefit from trial registration, and ensuring the study is conducted and reported in a way that seeks to minimise bias and conforms with the CONSORT statement for reporting of clinical trials (32).’
    ‘Some of these herbal medicines warrant further investigation in larger more rigorous studies to determine the size of the effect, dosage and long-term safety, namely combination products including Irvingia gabonensis, Ilex paraguariensis, Cissus quadrangularis, Sphaeranthus indicus, and some products which are only represented by one study.’

    Conclusion

    1. More emphasis on the quality of studies and the limitations of the evidence base is needed.

    Response: The quality of studies and limitations of the evidence have now been extensively discussed in the final few paragraphs of the discussion, as per point number 38. For completeness, the conclusion has also been elaborated as follows:
    ‘There is a paucity of RCTs of herbal medicines for weight loss and although some are larger studies, many are small in sample size, of poor design and methodological quality, with inadequate reporting of the herbal medicine interventions. These factors could easily contribute to the heterogeneity observed for some herbal medicines. The variability of the included studies in terms of herbal medicine formulation, dosages and study duration, and the varied inclusion of additional lifestyle interventions may have further contributed to a lack of consistent effect observed for some herbs. Future studies would benefit from trial registration, and ensuring the study is conducted and reported in a way that seeks to minimise bias and conforms with the CONSORT Statement for reporting of clinical trials (32).’

    Referee: 2

    Comments to the Author
    The manuscript by Alison Maunder et al is a systematic review on the effectiveness of herbal medicines for weight loss. The weight reduction effect of RCTs on several different herbs including Camellia sinensis and Garcinia cambogia were meta-analysed respectively.

    The weight reduction effect of natural herbs deserves more attention and research. The review summarizes relevant studies and finds that there is currently insufficient evidence to recommend any of these herbal medicines for the treatment of excess weight.

    Here, I have only one comment. It would be better, if the authors focused on just one herb, such as Camellia sinensis (green tea) and performed in-depth analysis and discussion with its chemical ingredients and known pharmacological effects.

    Response: The aim of our review was to identify and assess the efficacy and safety of any herbal medicines that have been investigated in randomised, parallel group, placebo-controlled trials (the gold standard for this type of research). It is noted in the manuscript that it has been 18 years since a systematic review of this nature has been published, with the expectation of many new herbal medicines having been investigated since. In light of this, the authors feel the original approach is still warranted, in order to synthesise new evidence for herbal medicines which have long been reported to assist with weight loss, and to identify new herbal medicines which may as yet not have substantial evidence behind them. This way, our review incorporates all evidence of a certain quality (RCTs) for the use of any and all herbal medicines for weight loss.

    Referee: 3

    Comments to the Author
    Thank you for the opportunity to review this paper. Though it is not a totally unique topic of discussion, it is well written, and the authors have conducted a thorough meta-analysis, and have provided additional insights into the use of herbal supplements for weight loss.

    Points to raise:-

    1) Page 10 - there is mention of a benchmark of weight loss threshold which is deemed clinically significant i.e. -2.5 kg. Where does this benchmark come from? I believe more further discussion of why this threshold is applied, is very important, especially since drug regulators (e.g. EMA) tend to apply a benchmark of 5% placebo-adjusted body weight loss (BWL) and 10% absolute weight loss, as clinically significant for pharmaceutical weight loss products.

    Therefore, please justify the rationale for selecting a threshold of -2.5kg as 'clinically significant' for herbal products.

    Response: The rationale for selecting -2.5kg as a benchmark for clinical significance is explained in the method. The benchmark of 5% weight loss is a somewhat arbitrary value used in clinical trials to separate weight losers from weight maintainers, and to assess the difference in outcomes and markers of health between these two groups. The smaller goal of 2.5kg was found to be the lowest amount of weight loss to still result in improved outcomes, and thus was adopted for this review. The method reads as follows:
    ‘Clinical relevance of weight loss estimates was defined by a difference of at least 2.5 kg when comparing the change of weight from baseline to the end of study between the intervention group and placebo. This amount of weight loss was selected as it was found to be sufficient to reduce the incidence of cardiovascular events and mortality in a recent large-scale clinical trial (12).’

    2) Page 2 - In the abstract, there is also mention of this benchmark of clinical significance - I would recommend adding information on the specific threshold i.e. - 2.5 kg in the abstract itself, as some readers may think that a higher threshold e.g. -5% placebo-adjusted BWL, is applied.

    3) Page 2 - Results. I think it is particularly important to mention therapies in which there were no statistically significant differences e.g. Camellia sinensis as a single herb.

    Response: The abstract has been amended to include mention of the benchmark for clinical significance and of therapies for which no statistically significant differences were observed. The following has been added to the Methods subsection of the abstract: ‘Weight differences of ≥2.5 kg were considered clinically significant.’
    The Results subsection of the abstract now reads as follows:
    ‘Of those meta-analysed, only Phaseolus vulgaris resulted in a statistically significant weight loss compared to placebo, though not considered clinically significant. No effect was seen for Camellia sinensis or Garcinia cambogia. Statistically, but not clinically, significant differences were observed for combination preparations containing Camellia sinensis, Garcinia cambogia, Phaseolus vulgaris, and/or Ephedra sinica. Of the herbal medicines trialled in ≤3 randomised controlled trials, statistically and clinically significant weight loss compared to placebo was reported for Irvingia gabonensis, Cissus quadrangularis, and Sphaeranthus indicus combined with Garcinia mangostana, among others. These findings…’

    Referee: 4

    Comments to the Author
    Manuscript ID: DOM-19-0849-RA
    Journal: Diabetes, Obesity and Metabolism
    Title: Effectiveness of herbal medicines for weight loss: a systematic review and meta-analysis of randomised controlled trials

    This systematic review and meta-analysis evaluated the efficacy and safety of herbal medicines for treating excess weight in healthy adults with overweight or obesity. The authors have included a total of 54 randomised placebo-controlled trials, and concluded that the current clinical evidence from randomised placebo-controlled trials is not sufficient enough for recommending these herbal medicines for reducing excess body weight in healthy adults with overweight or obesity.

    The authors of this study before its publication require considering the following changes in the manuscript.

    1. The authors, in eligibility criteria, mentioned that “only randomised placebo-controlled trials (RCTs) were eligible for inclusion” however, it is not clear whether the authors included RCTs of parallel-group or cross-over or both.

    2. The risk of bias tool (reference 30) which was used in this study seems to be for parallel-group RCT, but no clear details are reported by the authors.

    Response: This review included only randomised, parallel-group, placebo-controlled trials. This has been clarified in the manuscript as requested. The details of the risk of bias tool have also been extended to note this, and now reads:
    ‘…using the standard risk of bias assessment tool for parallel-group trials as recommended by the Cochrane Collaboration (30).’

    1. Authors need to justify the use of random effects models in the meta-analysis.

    Response: The use of random effects models has been justified as follows:
    ‘Random effects models were used as the true treatment effect was predicted to vary between studies, due to varied treatment dosages and durations, and differences in age and sex of participants.’

    1. Authors, in Garcinia cambogia and Phaseolus vulgaris, mentioned that “one study each was excluded from the meta-analysis due to incomplete data (reference 53 and 59);” did authors check the availability of clinical study report (CSR) or searched the trial if available in the clinical trial registry like CT.gov.

    Response: Trial registration was checked by the authors and attempts were made to contact the listed corresponding authors to obtain any missing data. In the case of these two references mentioned above, no response was received regarding the missing data, and neither study had been registered on a clinical trials database. Thus, they had to be excluded from the meta-analyses.

    1. Irvingia gabonensis: Authors mentioned that “as only three studies were identified, a meta-analysis could not be conducted.” However, in general, at least two (or more than two) studies are required for conducting meta-analysis, though the complete requirement depends on the power of the statistical tests in meta-analysis.

    Response: In this systematic review and meta-analysis, a decision was made to only conduct meta-analyses for herbal medicines with ≥4 studies identified. This decision is mentioned in the abstract and methods. This decision was made firstly, to be pragmatic, as doing many more meta-analyses for the herbs investigated in only two studies would not have provided much extra information; and secondly, to achieve some meaningful samples sizes and allow for the use of sensitivity and subgroup analyses. To remove the ambiguity, mention of ‘not conducting a meta-analysis as there were less than four studies’ has been removed from the manuscript. This includes one sentence removed from the results, and one sentence from the discussion replaced with the following:
    ‘However, only two to three studies conducted by the same research groups have been completed for each of these herbal medicines, so more evidence is needed before a definitive decision on their effect on weight loss can be reached.’

    1. In Herb medicine preparations with one RCT only, authors reported that “seventeen studies used herbal medicines that were utilised in only one RCT each,” however, for example, authors did not include another placebo-controlled RCT in Zingiber officinale which reported the outcome weight loss. Considering this, the authors need to rework on the search strategy and also screening of studies for inclusion.

    Reference: Ebrahimzadeh Attari V, Asghari Jafarabadi M, Zemestani M, Ostadrahimi A. Effect of Zingiber officinale supplementation on obesity management with respect to the uncoupling protein 1‐3826A> G and ß3‐adrenergic receptor Trp64Arg polymorphism. Phytotherapy Research. 2015 Jul;29(7):1032-9 (https://www.ncbi.nlm.nih.gov/pubmed/25899896).

    Response: We thank the Reviewer for bringing this to our attention. This paper was identified in our search strategy, along with a second published report of the same RCT which was included in our review (Ebrahimzadeh Attari, 2016). A mistake meant that the data was not included (although it was extracted). We have now added this data.

    1. Authors are requested to provide the statements (judgment) for risk of bias for all 54 included studies as supplementary document.

    Response: The authors’ judgements for each risk of bias domain for all 54 included studies are provided in the supplemental material (Supplemental Figure S1).

    1. The study concluded that the current clinical evidence from 54 RCTs is not sufficient enough to recommend the herbal medicines (listed in the systematic review) for reducing the body weight in healthy adults with overweight or obesity. However, this conclusion should be read with caution as 44% of the included studies asked the participants to maintain their usual dietary intake, the participants from 43% of the studies provided with dietary advice or meal plans for general health and/or weight loss, or were provided with a normo-caloric diet through food deliveries (4 studies). The remaining studies provided participants with low-calorie, meal replacement shakes. In terms of physical activity, 70% of studies requested participants to maintain their usual levels, 20% of studies provided advice for exercise, and 9% of studies held exercise sessions for participants to attend.

    As exercise and diet are contributing factors for weight loss in addition to supplementation of herbal medicines, based on this heterogeneity of included placebo-controlled RCTs, the conclusion of the study should be more robust rather than stating a general conclusion for the overall of body of evidence.

    Response: The discussion and conclusion has been amended to highlight the limitations as per point numbers 38 and 40 from Reviewer 1.



    Cite this author response
  • pre-publication peer review (ROUND 1)
    Decision Letter
    2019/09/20

    20-Sep-2019

    Dear Ms. Bessell

    Thank you for your patience, this manuscript has now been evaluated by one of the editors and two or more expert referees. The comments from reviewers are available at the bottom of this letter, alternately, you can log onto the manuscriptcentral website to view their feedback.

    The reviewers have also given the editor a separate assessment of the quality, originality and likely impact of your submission. These comments and scores, together with the editor’s judgment, are used to prioritise manuscripts for publication. The process is highly competitive, DOM has an impact factor of 6.13 and is currently ranked 16th out of 145 journals in the Endocrine Category.

    You will see that the reviewers have raised a number of points which require clarification and amendment, and suggested ways in which the manuscript might be improved. These comments are intended to be constructive and helpful.

    In addition, the handling editor has made the following comments to summarise the peer review:
    This needs major revision to improve the clarity, in particular the inclusion-exclusion criteria for the meta-analysis need to be more clearly described.

    The reviewers and editor would appreciate the opportunity to re-evaluate a revised manuscript which addresses the points raised. The revision should be accompanied by a separate document providing a point by point response to the reviewers’ comments and identifying where significant changes have been incorporated in the text.

    The revised manuscript will be re-assessed by the original reviewers and the handling editor. Only those manuscripts which are full competitive will be accepted for publication, therefore at this stage I cannot guarantee ultimate acceptance.

    Please use the link below to create and submit your revision online with no need to enter login details. Also, please ensure that you upload a WORD DOCUMENT and not a .pdf AND that you re-submit all figures and tables (whether they were revised or not).

    PLEASE NOTE: This is a two-step process. After clicking on the link, you will be directed to a webpage to confirm.

    https://mc.manuscriptcentral.com/dom?URL_MASK=47b1162a248749e38e1156b5f5fec187

    If you feel that your paper could benefit from English language polishing, you may wish to consider having your paper professionally edited for English language by a service such as Wiley’s at http://wileyeditingservices.com. Please note that while this service will greatly improve the readability of your paper, it does not guarantee acceptance of your paper by the journal.

    You have two months from today to provide your revision: if you need more time then please contact the Editorial Office at susan.lane@nottingham.ac.uk. Automatic reminders will be sent to you from the system.

    Kind regards,

    Prof. Richard Donnelly
    Editor-in-Chief, Diabetes, Obesity and Metabolism

    Referee: 1

    Comments to the Author
    Interesting study summarising the evidence of efficacy for herbal medicine against placebo for weight loss.

    There are some basic requirements that should be addressed including the use of inaccurate terms and inaccurate reporting of the quality of studies.

    More information about the assessment of risk of bias is needed, and limitations of the study would improve transparency and rigour.

    ALSO SEE ATTACHED DOCUMENT FOR DETAILED COMMENTS

    Referee: 2

    Comments to the Author
    The manuscript by Alison Maunder et al is a systematic review on the effectiveness of herbal medicines for weight loss. The weight reducion effect of RCTs on several different herbs including Camellia sinensis and Garcinia cambogia were meta-analyzed respectively.

    The weight reduction effect of natural herbs deserves more attention and research. The review summarizes relevant studies and finds that there is currently insufficient evidence to recommend any of these herbal medicines for the treatment of excess weight.

    Here, I have only one comment. It would be better, if the authors focused on just one herb, such as Camellia sinensis (green tea) and performed in-depth analysis and discussion with its chemical ingredients and known pharmacological effects.

    Referee: 3

    Comments to the Author
    Thank you for the opportunity to review this paper. Though it is not a totally unique topic of discussion, it is well written, and the authors have conducted a thorough meta-analysis, and have provided additional insights into the use of herbal supplements for weight loss.

    Points to raise:-

    Page 10 - there is mention of a benchmark of weight loss threshold which is deemed clinically significant i.e. -2.5 kg. Where does this benchmark come from? I believe more further discussion of why this threshold is applied, is very important, especially since drug regulators (e.g. EMA) tend to apply a benchmark of 5% placebo-adjusted body weight loss (BWL) and 10% absolute weight loss, as clinically significant for pharmaceutical weight loss products.

    Therefore, please justify the rationale for selecting a threshold of -2.5kg as 'clinically significant' for herbal products.

    2) Page 2 - In the abstract, there is also mention of this benchmark of clinical significance - I would recommend adding information on the specific threshold i.e. - 2.5 kg in the abstract itself, as some readers may think that a higher threshold e.g. -5% placebo-adjusted BWL, is applied.

    3) Page 2 - Results. I think it is particularly important to mention therapies in which there were no statistically significant differences e.g. Camellia sinensis as a single herb.

    Referee: 4

    Comments to the Author
    Manuscript ID: DOM-19-0849-RA
    Journal: Diabetes, Obesity and Metabolism
    Title: Effectiveness of herbal medicines for weight loss: a systematic review and meta-analysis of randomised controlled trials

    This systematic review and meta-analysis evaluated the efficacy and safety of herbal medicines for treating excess weight in healthy adults with overweight or obesity. The authors have included a total of 54 randomised placebo-controlled trials, and concluded that the current clinical evidence from randomised placebo-controlled trials is not sufficient enough for recommending these herbal medicines for reducing excess body weight in healthy adults with overweight or obesity.

    The authors of this study before its publication require considering the following changes in the manuscript.

     The authors, in eligibility criteria, mentioned that “only randomised placebo-controlled trials (RCTs) were eligible for inclusion” however, it is not clear whether the authors included RCTs of parallel-group or cross-over or both.

     The risk of bias tool (reference 30) which was used in this study seems to be for parallel-group RCT, but no clear details are reported by the authors.

     Authors need to justify the use of random effects models in the meta-analysis.

     Authors, in Garcinia cambogia and Phaseolus vulgaris, mentioned that “one study each was excluded from the meta-analysis due to incomplete data (reference 53 and 59);” did authors check the availability of clinical study report (CSR) or searched the trial if available in the clinical trial registry like CT.gov.

     Irvingia gabonensis: Authors mentioned that “as only three studies were identified, a meta-analysis could not be conducted.” However, in general, at least two (or more than two) studies are required for conducting meta-analysis, though the complete requirement depends on the power of the statistical tests in meta-analysis.

     In Herb medicine preparations with one RCT only, authors reported that “seventeen studies used herbal medicines that were utilised in only one RCT each,” however, for example, authors did not include another placebo-controlled RCT in Zingiber officinale which reported the outcome weight loss. Considering this, the authors need to rework on the search strategy and also screening of studies for inclusion.

    Reference: Ebrahimzadeh Attari V, Asghari Jafarabadi M, Zemestani M, Ostadrahimi A. Effect of Zingiber officinale supplementation on obesity management with respect to the uncoupling protein 1‐3826A> G and ß3‐adrenergic receptor Trp64Arg polymorphism. Phytotherapy Research. 2015 Jul;29(7):1032-9 (https://www.ncbi.nlm.nih.gov/pubmed/25899896).

     Authors are requested to provide the statements (judgment) for risk of bias for all 54 included studies as supplementary document.

     The study concluded that the current clinical evidence from 54 RCTs is not sufficient enough to recommend the herbal medicines (listed in the systematic review) for reducing the body weight in healthy adults with overweight or obesity. However, this conclusion should be read with caution as 44% of the included studies asked the participants to maintain their usual dietary intake, the participants from 43% of the studies provided with dietary advice or meal plans for general health and/or weight loss, or were provided with a normo-caloric diet through food deliveries (4 studies).

    The remaining studies provided participants with low-calorie, meal replacement shakes. In terms of physical activity, 70% of studies requested participants to maintain their usual levels, 20% of studies provided advice for exercise, and 9% of studies held exercise sessions for participants to attend.

    As exercise and diet are contributing factors for weight loss in addition to supplementation of herbal medicines, based on this heterogeneity of included placebo-controlled RCTs, the conclusion of the study should be more robust rather than stating a general conclusion for the overall of body of evidence.

    Decision letter by
    Cite this decision letter
    Reviewer report
    2019/09/18

    Manuscript ID: DOM-19-0849-RA
    Journal: Diabetes, Obesity and Metabolism
    Title: Effectiveness of herbal medicines for weight loss: a systematic review and meta-analysis of randomised controlled trials

    This systematic review and meta-analysis evaluated the efficacy and safety of herbal medicines for treating excess weight in healthy adults with overweight or obesity. The authors have included a total of 54 randomised placebo-controlled trials, and concluded that the current clinical evidence from randomised placebo-controlled trials is not sufficient enough for recommending these herbal medicines for reducing excess body weight in healthy adults with overweight or obesity.

    The authors of this study before its publication require considering the following changes in the manuscript.

     The authors, in eligibility criteria, mentioned that “only randomised placebo-controlled trials (RCTs) were eligible for inclusion” however, it is not clear whether the authors included RCTs of parallel-group or cross-over or both.

     The risk of bias tool (reference 30) which was used in this study seems to be for parallel-group RCT, but no clear details are reported by the authors.

     Authors need to justify the use of random effects models in the meta-analysis.

     Authors, in Garcinia cambogia and Phaseolus vulgaris, mentioned that “one study each was excluded from the meta-analysis due to incomplete data (reference 53 and 59);” did authors check the availability of clinical study report (CSR) or searched the trial if available in the clinical trial registry like CT.gov.

     Irvingia gabonensis: Authors mentioned that “as only three studies were identified, a meta-analysis could not be conducted.” However, in general, at least two (or more than two) studies are required for conducting meta-analysis, though the complete requirement depends on the power of the statistical tests in meta-analysis.

     In Herb medicine preparations with one RCT only, authors reported that “seventeen studies used herbal medicines that were utilised in only one RCT each,” however, for example, authors did not include another placebo-controlled RCT in Zingiber officinale which reported the outcome weight loss. Considering this, the authors need to rework on the search strategy and also screening of studies for inclusion.

    Reference: Ebrahimzadeh Attari V, Asghari Jafarabadi M, Zemestani M, Ostadrahimi A. Effect of Zingiber officinale supplementation on obesity management with respect to the uncoupling protein 1‐3826A> G and ß3‐adrenergic receptor Trp64Arg polymorphism. Phytotherapy Research. 2015 Jul;29(7):1032-9 (https://www.ncbi.nlm.nih.gov/pubmed/25899896).

     Authors are requested to provide the statements (judgment) for risk of bias for all 54 included studies as supplementary document.

     The study concluded that the current clinical evidence from 54 RCTs is not sufficient enough to recommend the herbal medicines (listed in the systematic review) for reducing the body weight in healthy adults with overweight or obesity. However, this conclusion should be read with caution as 44% of the included studies asked the participants to maintain their usual dietary intake, the participants from 43% of the studies provided with dietary advice or meal plans for general health and/or weight loss, or were provided with a normo-caloric diet through food deliveries (4 studies).

    The remaining studies provided participants with low-calorie, meal replacement shakes. In terms of physical activity, 70% of studies requested participants to maintain their usual levels, 20% of studies provided advice for exercise, and 9% of studies held exercise sessions for participants to attend.

    As exercise and diet are contributing factors for weight loss in addition to supplementation of herbal medicines, based on this heterogeneity of included placebo-controlled RCTs, the conclusion of the study should be more robust rather than stating a general conclusion for the overall of body of evidence.

    Cite this review
    Reviewer report
    2019/09/16

    Thank you for the opportunity to review this paper. Though it is not a totally unique topic of discussion, it is well written, and the authors have conducted a thorough meta-analysis, and have provided additional insights into the use of herbal supplements for weight loss.

    Points to raise:-

    Page 10 - there is mention of a benchmark of weight loss threshold which is deemed clinically significant i.e. -2.5 kg. Where does this benchmark come from? I believe more further discussion of why this threshold is applied, is very important, especially since drug regulators (e.g. EMA) tend to apply a benchmark of 5% placebo-adjusted body weight loss (BWL) and 10% absolute weight loss, as clinically significant for pharmaceutical weight loss products.

    Therefore, please justify the rationale for selecting a threshold of -2.5kg as 'clinically significant' for herbal products.

    2) Page 2 - In the abstract, there is also mention of this benchmark of clinical significance - I would recommend adding information on the specific threshold i.e. - 2.5 kg in the abstract itself, as some readers may think that a higher threshold e.g. -5% placebo-adjusted BWL, is applied.

    3) Page 2 - Results. I think it is particularly important to mention therapies in which there were no statistically significant differences e.g. Camellia sinensis as a single herb.

    Reviewed by
    Cite this review
    Reviewer report
    2019/09/08

    The manuscript by Alison Maunder et al is a systematic review on the effectiveness of herbal medicines for weight loss. The weight reducion effect of RCTs on several different herbs including Camellia sinensis and Garcinia cambogia were meta-analyzed respectively.

    The weight reduction effect of natural herbs deserves more attention and research. The review summarizes relevant studies and finds that there is currently insufficient evidence to recommend any of these herbal medicines for the treatment of excess weight.

    Here, I have only one comment. It would be better, if the authors focused on just one herb, such as Camellia sinensis (green tea) and performed in-depth analysis and discussion with its chemical ingredients and known pharmacological effects.

    Reviewed by
    Cite this review
    Reviewer report
    2019/08/29

    Interesting study summarising the evidence of efficacy for herbal medicine against placebo for weight loss.

    There are some basic requirements that should be addressed including the use of inaccurate terms and inaccurate reporting of the quality of studies.

    More information about the assessment of risk of bias is needed, and limitations of the study would improve transparency and rigour.

    ALSO SEE ATTACHED DOCUMENT FOR DETAILED COMMENTS

    Reviewed by
    Cite this review
All peer review content displayed here is covered by a Creative Commons CC BY 4.0 license.