Phlebotonics for haemorrhoids

Abstract

Background

Haemorrhoids are variceal dilatations of the anal and perianal venous plexus and often develop secondary to the persistently elevated venous pressure within the haemorrhoidal plexus (Kumar 2005). Phlebotonics are a heterogenous class of drugs consisting of plant extracts (i.e. flavonoids) and synthetic compounds (i.e. calcium dobesilate). Although their precise mechanism of action has not been fully established, they are known to improve venous tone, stabilize capillary permeability and increase lymphatic drainage. They have been used to treat a variety of conditions including chronic venous insufficiency, lymphoedema and haemorrhoids.

Numerous trials assessing the effect of phlebotonics in treating the symptoms and signs of haemorrhoidal disease suggest that there is a potential benefit.

Objectives

The aim of this review was to investigate the efficacy of phlebotonics in alleviating the signs, symptoms and severity of haemorrhoidal disease and verify their effect post‐haemorrhoidectomy.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library 2011 issue 9 , MEDLINE (1950 to September 2011) and EMBASE (1974 to September 2011).

Selection criteria

Only randomised controlled trials evaluating the use of phlebotonics in treating haemorrhoidal disease were used. No cross‐over or cluster‐randomized trials were included for analysis and any trial which had a quasi‐random method of allocation was excluded.

Data collection and analysis

Two authors independently extracted the data and analysed the eligibility of the data for inclusion. Disagreements were resolved by meaningful discussion.

Main results

We considered twenty‐four studies for inclusion in the final analysis. Twenty of these studies (enrolling a total of 2344 participants) evaluated the use of phlebotonics versus a control intervention. One of these twenty studies evaluated the use of phlebotonics with a medical intervention and another study with rubber band ligation.

The remaining four studies included two which compared different forms of phlebotonics with each other, one study which evaluated phlebotonics with a medical intervention and one study which compared the use of phlebotonics with infrared photocoagulation. Eight studies were excluded for various reasons including poor methodological quality.

Phlebotonics demonstrated a statistically significant beneficial effect for the outcomes of  pruritus (OR 0.23; 95% CI 0.07 to 0.79) (P=0.02), bleeding (OR 0.12; 95% CI 0.04 to 0.37) (P=0.0002), bleeding post‐haemorrhoidectomy (OR 0.18; 95% 0.06 to 0.58)(P=0.004), discharge and leakage (OR 0.12; 95% CI 0.04 to 0.42) (P=0.0008) and overall symptom improvement (OR 15.99 95% CI 5.97 to 42.84) (P< 0.00001), in comparison with a control intervention. Although beneficial they did not show a statistically significant effect compared with a control intervention for pain (OR 0.11; 95% CI 0.01 to 1.11) (P=0.06), pain scores post‐haemorrhoidectomy (SMD ‐1.04; 95% CI ‐3.21 to 1.12 ) (P= 0.35) or post‐operative analgesic consumption (OR 0.54; 95% CI 0.30 to 0.99)(P=0.05).

Authors' conclusions

The evidence suggests that there is a potential benefit in using phlebotonics in treating haemorrhoidal disease as well as a benefit in alleviating post‐haemorrhoidectomy symptoms. Outcomes such as bleeding and overall symptom improvement show a statistically significant beneficial effect and there were few concerns regarding their overall safety from the evidence presented in the clinical trials.

However methodological limitations were encountered. In order to enhance our conclusion further, more robust clinical trials which take into account these limitations will need to be performed in the future.

Plain language summary

Phlebotonics for haemorrhoids

Haemorrhoids are among the most common benign anorectal pathologies which usually manifest with the common symptoms and signs of bleeding, pain, pruritus, swelling and discharge. The prevalence can vary from 4.4% in the general population to 36.4% in general practice. However their true prevalence will inevitably be underestimated due to the under‐reporting of these symptoms. Medical and conservative management with high‐fibre diets, stool softeners and laxatives are the preferred treatments for grade I‐II haemorrhoids whereas surgical procedures such as haemorrhoidectomy are reserved for the more severe forms of haemorrhoids.  Phlebotonics are a heterogeneous class of drugs used to treat haemorrhoidal disease in the less severe stages of first and second‐degree haemorrhoids, and during the thrombosis episodes. Although their true mechanism of action has not been well established, they are associated with strengthening of blood vessel walls, increasing venous tone, lymphatic drainage and normalizing capillary permeability. We considered twenty four studies for inclusion in this review. This review identified twenty randomised controlled trials enrolling a total of (2334) participants which compared an intervention using phlebotonics with a control intervention. Of these twenty studies, one study compared phlebotonics with a medical intervention and another with rubber band ligation. Of the remaining four trials, we identified two trials which compared phlebotonics with each other, one trial which compared phlebotonics with herbal therapy and one trial which compared phlebotonics with infrared photocoagulation. The trials obtained did not show any significant adverse events or side‐effects from the use of phlebotonics. The studies demonstrated a beneficial effect of phlebotonics in treating the symptoms and signs of haemorrhoidal disease as well as symptom relief post‐haemorrhoidectomy.

Summary of findings

for the main comparison.

Phlebotonic compared with control for haemorrhoidal disease
Patient or population: Patients with haemorrhoids Settings: University or general hospital setting Intervention: Phlebotonics Comparison: Placebo or control intervention
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Phlebotonic
Pain	527 per 1000	111 per 1000	OR 0.11 (0.01 to 1.11)	190 (2 studies)	+++O 1,2 moderate
Bleeding	253 per 1000	40 per 1000	OR 0.12 (0.04 to 0.37)	190 (2 studies)	+++O 4 moderate
Bleeding post‐haemorrhoidectomy	115 per 1000	25 per 1000	OR 0.18 (0.06 to 0.58)	314 (2 studies)	++++5 high
Pruritus	143 per 1000	33 per 1000	OR 0.23 (0.07 to 0.79)	176 (2 studies)	++++ 5 high
Discharge or leakage	266 per 1000	67 per 1000	OR 0.12 (0.04 to 0.42)	139 (2 studies)	+++O 6 moderate
Analgesic consumption (day 2) post‐operative	564 per 1000	436 per 1000	OR 0.54 (0.30 to 0.99)	202 (3 studies)	+++O 7 moderate
Overall symptom improvement (positive outcome)	566 per 1000	938 per 1000	OR 15.97 (5.97 to 42.84)	368 (5 studies)	++00 1,8 low
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; OR: Odds Ratio;
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. 1.     Serious inconsistency ; high statistical heterogeneity (I2>75%) . Heterogeneity may most likely be due to differences in formulations, doses and regimes of phlebotonics. 2.    No serious study limitations. Adequate blinding, however Cospite 1994 did not describe sequence generation or allocation concealment in adequate detail. 3.    No serious study limitations. Adequate methods of blinding, however methods of allocation concealment or sequence generation unclear for Mlakar 2005 and Ho 2000. 4.     No serious study limitations. Adequate blinding, however methods of sequence generation or  allocation concealment unclear for Misra 2000 or Ho 1995 . 5.    No serious study limitations. No statistical heterogeneity. Adequate methods of blinding, sequence generation and allocation concealment which were described in adequate detail. 6.    No serious inconsistency. Adequate methods of blinding, however methods of sequence generation and allocation concealment are unclear for  Cospite 1994. 7.    No serious limitations. Adequate methods of blinding, however allocation concealment unclear for Basile 2001, Basile 2002 and Colak 2003. 8.    No serious limitations. Adequate methods of blinding, however allocation concealment or sequence generation unclear for Cospite 1994, Mentes 2001, Thanapongsathorn 1992 and Wijayanegara 1992.

Background

Haemorrhoids are among the most common anorectal pathologies blamed for many anorectal complaints by both patients and medical practitioners alike. Although ambiguity often arises when defining haemorrhoids, for the purposes of this review haemorrhoids refer specifically to the pathological presence of circular venous type cushions present in and around the anal canal. They consist of a vascular circular‐like cushion and are found within the subepithelial space of the anal canal. Haemorrhoids consist of connective tissue cushions surrounded by arteriovenous communications between terminal branches of the superior rectal arteries and the inferior, superior and middle rectal veins (Thomson 1975). Due to their rich vascular supply, highly sensitive location and their tendency to engorge or prolapse, haemorrhoids are subsequently a common cause for many symptomatic anorectal complaints (Billingham 2004).

The classical position of haemorrhoids corresponds to three positions which are the right anterior, right posterior and the left lateral areas of the anal canal. Haemorrhoids may be external or internal. External haemorrhoids are covered with skin and internal haemorrhoids are covered with anal mucous membranes. The majority of symptoms and signs which patients present with, arise from internal haemorrhoids. The grading system used by Banov et al. to classify internal haemorrhoids is one of the most common grading systems used by clinicians and is used to guide therapeutic practice (Banov 1985). Internal haemorrhoids can be divided into four categories depending on the degree of prolapse (the protrusion of the haemorrhoid through the anus). First‐grade: protrude into the anal canal but do not prolapse; second‐grade: prolapse but spontaneously reduce; third‐grade: prolapse and require manual reduction; fourth‐grade: irreducible prolapse.

This medical problem is one of the most common gastrointestinal conditions with a prevalence varying from 4.4% in the general population to 36.4% in general practice (Abramowitz 2001). Haemorrhoids occur equally in both men and women. The prevalence is higher in the Caucasian population with a greater incidence in higher socioeconomic groups and a peak incidence between the ages of 45‐65 years (Johanson 1990).

The symptoms derived from structural changes of the normal anatomic padding constitute symptomatic haemorrhoids known as "piles". The majority of symptoms are derived from enlarged internal haemorrhoids and are generally associated with chronic straining either due to constipation, diarrhoea or prolonged periods trying to defecate. It is also common during pregnancy and childbirth (Beck 1998). Common symptoms and signs are rectal bleeding, prolapse, pruritus (itching), pain and less frequently soiling or mucoid discharge. Rectal bleeding (the most common presenting sign) is classically bright red as there is a high blood oxygen content within the arteriovenous anastomosis. This bleeding is usually described as bright red spotting on the toilet tissue or dripping in the toilet bowl and normally occurs at the end of defecation, separately from the stool. External haemorrhoids may be asymptomatic or associated with discomfort or with acute extreme pain in the event of a local thrombosis (formation of a clot). It should be noted that despite the common prevalence of haemorrhoids it is important to exclude other gastrointestinal conditions such as colonic carcinoma particularly in the elderly which can often present with rectal bleeding.

Medical and conservative management in the primary care setting are the mainstay of treatment for grade one internal and non‐thrombosed external haemorrhoids. This consists of high‐fibre diets, adequate fluid intake, stool softeners, topical and systemic analgesics, proper anal hygiene and in some cases a short course of topical steroid cream. High‐fibre diets in particular, help reduce the severity and duration of symptoms (Moesgaard 1982). Grade two haemorrhoids are generally treated with non‐surgical procedures such as rubber‐band ligation (Johanson 1992), injection sclerotherapy or cryosurgery. Surgical haemorrhoidectomy is generally reserved for those with grade three, grade four or grade two haemorrhoids with failed non‐surgical management. Currently, it is estimated that approximately 5‐10% of patients with haemorrhoids will eventually require treatment with haemorrhoidectomy while the rest can be managed conservatively. It should be noted that surgical management does pose significant risks such as post‐operative urinary retention which can occur in approximately 20 percent of patients (Bleday 1992) and therefore conservative management where appropriate, using drugs such as phlebotonics or lifestyle changes is ideal.

Phlebotonics are a heterogeneous class of drugs mainly used to treat chronic venous insufficiency (the inability of veins to transport blood towards the heart) (Martinez 2005) and haemorrhoids (Quijano 2005). They are generally used in the less severe stages of haemorrhoids, first and second‐grade haemorrhoids and during the thrombosis episodes. Although their true mechanism of action has not been well established, they are generally associated with strengthening of blood vessel walls, increasing venous tone, lymphatic drainage and normalising capillary permeability (Beck 2011). Most of these drugs are natural products extracted from plants (flavonoids, sapsonides, etc), but synthetic products are also used (calcium dobesilate i.e.). The true benefits and efficacy of these drugs are still not clear. The main adverse effects of flavonoids are gastrointestinal symptoms, however some case‐control studies have shown an associated risk of agranulocytosis (severe reduction of white blood cells which in some cases can be life threatening) for calcium dobesilate (Ibanez 2000).There have been numerous examples in the literature as well as randomised controlled trials where classes of phlebotonics have been evaluated for their effectiveness in the treatment of haemorrhoids. So far no rigorous evaluation has been performed assessing the effectiveness and safety of phlebotonics in the treatment of haemorrhoids or their benefit post‐haemorrhoidectomy. For this reason it was a necessity to perform a review.

Objectives

The main objective of this review was to determine the efficacy and safety of phlebotonics in improving the symptoms and signs of haemorrhoids as well as to determine their effect post‐haemorrhoidectomy.

Methods

Criteria for considering studies for this review

Types of studies

All published and unpublished parallel group double‐blinded randomised controlled trials (RCTs) were eligible for inclusion in this review. Randomised controlled trials published as abstracts or as letters were also included in the primary analysis. However, in order to explore the potential bias resulting from their inclusion, a secondary sensitivity analysis was also conducted. Quasi‐random studies were excluded.

Types of participants

1) Patients of both genders and all ages receiving phlebotonics for first‐, second‐, third‐ and fourth‐grade haemorrhoids, symptomatic or asymptomatic, will be eligible as well as patients who received phlebotonics post‐haemorrhoidectomy.

2) Pregnant women or women with haemorrhoids after delivery will be eligible.

3) Patients of both genders and all ages receiving phlebotonics with thrombosed haemorrhoids will be eligible.

Types of interventions

We intended to obtain studies involving the use of phlebotonics administered orally or topically at any dose in the following comparisons. However we envisaged that the majority of the comparisons would involve phlebotonics and control interventions.

‐Phlebotonic versus phlebotonic, comparing different classes of phlebotonics (e.g flavonoids vs non‐flavonoids),

‐Phlebotonic versus placebo,

‐Phlebotonic versus conservative management (lifestyle intervention, high fibre diet),

‐Phlebotonic versus medical intervention,

‐Phlebotonic versus surgical intervention,

‐Lifestyle changes (fibre) or topical treatment plus phlebotonics versus lifestyle changes (fibre) or topical treatment (fibre) plus placebo,

‐Lifestyle changes (fibre) or topical treatment plus phlebotonics versus lifestyle changes(fibre) or topical treatment,

‐Non‐operative or surgical treatment plus phlebotonics versus non‐operative or surgical treatment plus placebo,

‐Non‐operative or surgical treatment plus phlebotonics versus non‐operative or surgical treatment.

Phlebotonics are a class of drugs which will include:

1) Natural products: 
 ‐Bioflavonoids: Quercetin, rutin, hesperidin, rutosides (troxerutin, buckwheat herb extract, ruscus aculeatus), diosmin, hidrosmin, ginko biloba; saponosides: escin (horse chestnut seed extract), 
 2) Synthetic products: calcium dobesilate, naftazone, aminaftone, chromocarbe, 
 3) Others: iquinosa, flunarizine, sulfomucopoly‐saccharide.

Types of outcome measures

The main outcome measures assessed were related to symptom control, relief and relapse rates of symptomatic haemorrhoids. Attempts at contacting authors were made if data from studies was ambiguous or insufficient.

The outcome measures were categorized into primary and secondary.

1) Primary outcome measures included the most frequent symptoms of haemorrhoids. These were;

i) Measurements of bleeding,

ii) Measurements of pain,

iii) Measurements of pruritis.

2) Secondary outcome measures included the following;

i) Analgesic consumption,

ii) Discharge or leakage,

iii) Overall symptom improvement,

iv) Compliance,

v) Any adverse events (side‐effects)

Search methods for identification of studies

The search strategy advocated by Cochrane colorectal cancer group (CCCG) was adopted. The databases used were primarily the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE and EMBASE. The databases CINAHL, AMED, the British Nursing Index (BNI) and Health Business Elite were also used but resulted in no additional search results being obtained. The search strategies used when searching the MEDLINE and EMBASE databases are outlined below;

MEDLINE

1 hemorrhoid*

2 exp HEMORRHOID/

3 haemorrhoid*

4 hemorhoid*

5 hemorroid*

6 haemorroid*

7 haemorhoid*

8 1 OR 2 OR 3 OR 4 OR 5 OR 6 OR 7

9 phlebotonic*

10 flavonoid*

11 bioflavonoid*

12 escin OR quercetin OR rutin OR hesperidin OR rutoside* OR diosmin OR hidrosmin OR saponosides

13 calcium AND dobesilate

14 diosmin OR naftazone OR aminaftone OR chromocarbe OR daflon OR naphtha OR venotoni* OR hydroxyethylrutoside dobesilate OR naftazone.

15 9 OR 10 OR 11 OR 12 OR 13 OR 14

16 8 AND 15

The Cochrane Controlled Trials Register and Embase

1 hemorrhoid*

2 exp HEMORRHOID/

3 haemorrhoid*

4 hemorhoid*

5 hemorroid*

6 haemorroid*

7 haemoroid*

8 hemoroid*

9 1 OR 2 OR 3 OR 4 OR 5 OR 6 OR 7 OR 8

10 phlebotonic*

11 flavonoid*

12 exp FLAVONOID/

13 bioflavonoid*

14 quercetin OR rutin OR hesperidin OR rutoside* OR diosmin OR hidrosmin OR saponosides OR escin

15 calcium AND dobesilate

16 diosmin OR daflon OR naphtha OR venotoni* OR dobesilate OR naftazone OR aminaftone OR chromocarbe OR hydroxyethylrutoside

17 10 OR 11 OR 12 OR 13 OR 14 OR 15 OR 16

18 9 AND 17

These electronic databases were explored vigorously using the broad search strategies such as those described above to identify all the randomised controlled trials available evaluating the efficacy of phlebotonics for haemorrhoids. All searches were run from the earliest date available (1950 for MEDLINE, 1974 for EMBASE, 1981 for CINAHL and 1985 for AMED). A final search of MEDLINE, EMBASE, AMED and CINAHL was undertaken three months before publication. Free Internet search engines such as Google Scholar, Intute and Turning Research Into Practice (TRIP) databases were also used but revealed no useful study data. All journals were included and retrieved into the final analysis. Non‐English language journals were retrieved and were subsequently translated.

In order to minimize publication bias, a grey literature search was performed. The Healthcare Management Information Consortium (HMIC) of the Department of Health Library (UK), the King's Fund health charity database, OpenSIGLE and the National Technical Information Service (NTIS) was used to search for all relevant grey literature such as conference abstracts.

Attempts at contacting pharmaceutical companies and editors of general medical, general practice, gastroenterology and general surgery journals were made for information on any papers on phlebotonics and haemorrhoids undergoing peer‐review.

It is acknowledged that some completed trials are never published. In order to minimize bias, it was a necessity to search for unpublished trials. Ongoing trials were searched mainly from the following databases;

• Meta Register of Controlled Trials (mRCT); http://www.controlled‐trials.com

• US NIH register; http://clinicaltrials.gov

• Register of the Center for Clinical Trials and Evidence‐Based Healthcare; http://trialscentral.org

• The Association of the British Pharmaceutical Industry (ABPI); Trials database ‐ www.cmrinteract.com/clintrial/

Other national and international trial registers were searched for, if appropriate. Particular pharmaceutical or biotech industry trials registers were searched for and the Clinical Trial Results web site (www.clinicaltrialresults.org/) were searched for various slide presentations reporting the results of relevant clinical trials related to phlebotonics and haemorrhoids.

Data collection and analysis

1) Selection of Studies:

Two reviewers (NP and DL) were involved in the selection of studies. Once all the papers were collected, one reviewer determined the eligibility of the studies and excluded papers from the initial searches unrelated to haemorrhoids in humans. This decision was based on account of the title, abstract and content. The second reviewer independently scrutinized the studies excluded from this selection process. Disagreements were addressed by meaningful discussion and where necessary the remaining reviewers were contacted for further assistance.

Afterwards the full text of potentially relevant reports were retrieved in order so that they could be examined by the two reviewers (for compliance with the pre‐stated eligibility criteria). These two reviewers examined the full text reports independently, disagreements were addressed and the remaining reviewers were contacted if a resolution could not be found. Where appropriate, it was necessary to liaise with investigators to request further information such as incomplete results data.

All potentially relevant trial reports were scrutinized for multiple publications of the same data. It was necessary to compare criteria such as similar author names, locations, settings and similar interventions in order to avoid duplicate publications.

Those trials searched for but classified as not eligible are displayed in a table together with the reason for their exclusion.

2) Data Collection:

Data was collected and recorded onto specially designed forms by two reviewers. These data collection forms were piloted and tested using a representative sample of collected studies and were modified pending further changes necessary. Data collection involved two reviewers and was proof‐read by the remaining reviewers. Disagreements arose when collecting data, and the remaining reviewers were consulted where necessary. Attempts at contacting study authors where data was insufficient were also made.

The specialized data collection forms were paper forms. Data extracted contained key characteristics of each trial (Higgins 2009) with the following subheadings;

i) Eligibility; whether this trial was eligible and reason for exclusion,

ii) Study methods; study design, total study duration, sequence generation, allocation sequence concealment, blinding and bias,

iii) Participants; age, sex, nationality total number and degree of haemorrhoids,

iv) Interventions; the specific intervention which took place and their intervention detail,

v) Outcomes; the outcome definition and the unit of interest. The details and prevalence of individual symptoms before and after treatment (making note of any improvements), haemorrhoidal symptom scores, global assessments of symptoms and where possible patient satisfaction scores post treatment. Where measurement scales were used it was noted whether or not they were standard scales and whether they were validated,

vi) Results; Number of participants. For each outcome of interest ‐ sample size, missing participants, summary data for each intervention group (e.g. 2×2 table for dichotomous data; means and SDs for continuous data),

vii) Miscellaneous; Key conclusions of the study authors ‐ miscellaneous comments from the study authors and references to other relevant studies.

3) Assessment of bias and study quality:

The studies selected were assessed for whether the study correctly avoids bias. Bias specifically refers to a systematic error or deviation from the truth in results or inferences (Higgins 2009). Methodological quality however refers to the extent to which the authors have carried out their work to the highest possible standard (Juni 2001). It was recommended by the Cochrane collaboration that the level of bias was to be assessed rather than the study quality (Higgins 2009). In order to assess the risk of bias a two part domain‐based tool was used. A pre‐designed form was created with seven specific domains. This addressed 'sequence generation', 'allocation concealment', 'blinding', 'incomplete outcome data', 'selective outcome reporting' and 'other issues and sources of bias'. Within each entry of the domain, Part One involved a description of what happened in that study. Part Two involved a judgement relating to the risk of bias for that entry. The answers 'yes', 'unclear' or 'no' were allocated for each of part two. These corresponded to 'low risk of bias', 'unclear risk of bias' or 'no risk of bias', respectively. This assessment for each study was performed by two reviewers (NP and AC) with the remaining reviewers participating if disagreements arose. Pre‐set criteria for allocating an answer such as 'yes, no or unclear' were already defined (Higgins 2009). If the data reported in each trial was insufficient, the judgement allocated was ‘Unclear' risk of bias.

Using RevMan5, a 'Risk of bias graph' was generated illustrating the proportion of studies with each judgement (Yes, No or Unclear). A 'Risk of bias summary figure' was then generated to draw conclusions about the overall risk of bias.

A comprehensive description of the assessment of the risk of bias tool is shown below;

Assessment of Risk of Bias

Assessments of risk of bias were performed using the Cochrane collaboration risk of bias tool. The following outline is shown below:

i) Sequence Generation; assessed as yes, no or unclear:

Yes; When the allocation described to generate the sequence was described in sufficient detail,

Unclear; When it was not or incompletely described,

No; When the sequence was not generated or a quasi‐random method of allocation was described.

ii) Allocation concealment; assessed as yes, no or unclear:

Yes; When the study described sufficient concealment of the allocation,

Unclear; When it was not described or incompletely described,

No; When it was clear that the allocation of participants to the trial had been broken.

iii) Blinding of participants; assessed as yes, no or unclear:

Yes; When blinding took place either as single or double blinded,

Unclear; When it was not clearly described,

No; When it was an unblinded study.

iv) Incomplete data outcome; assessed as yes, no or unclear:

Yes; When the study described the completeness of data for each outcome stated including attrition and exclusion and these did not include a significant number of exclusions,

Unclear; When it was not or incompletely described,

No; When there were significant exclusions.

v) Free of selective outcome reporting; assessed as yes, no or unclear:

Yes; When the studies primary and secondary outcomes of interest have been reported,

Unclear; Insufficient information to report if all the outcome measures have been reported,

No; When missing outcomes are not reported.

vi) Other sources of bias; assessed as yes, no or unclear:

Other sources of bias which we considered were funding agencies especially pharmaceutical. Papers which did not state the source of funding were deemed as 'Unclear'.

Other sources of bias that we considered were baseline imbalance.

4) Addressing reporting bias

In order to detect the degree of reporting bias, a funnel plot was constructed detecting the intervention effects against the study size and precision. It is acknowledged that if reporting bias does occur, the funnel plot will have an asymmetrical appearance with the degree of asymmetry resembling the degree of bias. Funnel plots of effect estimates against their standard error of the intervention effects were plotted using RevMan 5.

When interpreting a funnel plot, asymmetry was acknowledged, as a variety of other factors could be contributing other than reporting bias (Egger 1997). A visual assessment of funnel plot asymmetry was performed, however for an objective determination statistical tests were considered. The approach used by Egger 1997 for dichotomous data corresponds to a linear regression of the log odds ratio on its standard error, weighted by the inverse of the variance of the log odds ratio. If continuous data were present, a simple statistical test involving the linear regression of the intervention effect estimates on their standard errors was considered.

5) Presenting results and Summary of Findings

In order to present results in a systematic and concise format a variety of tables and figures were used which included the 'Characteristics of included studies' table, 'Forest plots' and 'Summary of findings'.

The 'Characteristics of included studies' table presented information on individual studies. They contained the following subheadings; i) Methods ii) Participants iii) Intervention iv) Outcomes. The 'Methods' section detailed the study design, stating whether randomizations occurred and the duration of study. The 'Participants' section included details of the participants such as the settings, degree of haemorrhoids, age, sex and nationality. The 'Interventions' section listed the various intervention groups including the details of phlebotonics used, dose, frequency, mode of administration, side‐effects and duration. The 'Outcomes' section included the outcomes of interventions and time‐points measured in the study.

Forest plots were generated using the RevMan 5 software. Data was entered into RevMan5 by two reviewers.

A 'Summary of findings' table was produced using the GRADEprofiler and the RevMan 5 software. The standard template for this table is recommended by the Cochrane collaboration (Schunemann 2008). The following subheadings were used:

i) List of all important outcomes ii) Illustrative comparative risks iii) Absolute and relative magnitude of effect iv) Numbers of participants and studies addressing outcomes v) Quality of evidence rating (GRADE Working Group 2004) vi) Additional comments.

6) Data analysis and drawing conclusions:

For the purposes of this review, a meta‐analysis combining the statistical analysis of multiple studies was performed using RevMan 5. The meta‐analysis was used as well as the statistical power to enhance the statistical precision.

Measure of treatment effect

The meta‐analysis performed involved a two‐part process. The first part of the meta‐analysis involved calculating a summary statistic for the study. For dichotomous data we calculated the estimate for the treatment of effect using the odds ratio.

For continuous data the mean and the standard deviations were recorded. Where necessary the standard deviations were calculated from the standard errors or from the upper and lower limits of the confidence intervals provided. The two main summary statistics which were used were the Mean Difference (MD) and the Standardized Mean Difference (SMD). The SMD was used as a summary statistic where for example the outcome measure of interest is the relief of haemorrhoidal symptoms but the studies use different global assessment scales to measure the effect. The SMD expresses the size of the intervention effect in each study relative to the variability observed in that study.

SMD = (difference in mean outcome between groups)/(Standard deviation of outcome among participants)

The second stage of the meta‐analysis involved pooling together the data and calculating the weighted average of the effects of the intervention estimated in the individual studies. The weighted average was defined as;

Weighted average = (sum of (estimate x weight))/(sum of weights)

For the meta‐analysis of dichotomous data, four specific types of methods were considered depending on the result of the studies obtained. These were 'Mantel‐Haenszel', 'Peto', or 'Inverse Variance' for fixed‐effect meta‐analysis or 'DerSimonian and Laird' for random effects meta‐analysis. The Mantel‐Haenszel (Mantel 1959; Greenland 1985) was used if the data present was sparse or the study size was small. The Peto's odds ratio method was used to pool relevant odds ratios together.

Unit of analysis

Only randomised controlled trials were included for analysis in this review. No trials were cross‐over or cluster‐randomized controlled trials.

Missing data

It was considerably difficult to contact investigators regarding missing data. Many contact details of the authors were not included in studies. We considered incomplete data to have been addressed if 85% or more participants had been included in the final analysis. If less than 85% were included but steps were taken to ensure that these did not bias the results, this was deemed satisfactory.

If the above were not clear, we intended to calculate the statistics on an intention‐to‐treat analysis (ITT) basis. Where summary statistics were not available, we intended to calculate these summary statistics (e.g. standard deviations) from the available data supplied.

Assessment of heterogeneity

The degree of heterogeneity was assessed using the Chi² test and the I² test (Higgins 2003). It should be noted that a Chi ² test is poor at detecting the true heterogeneity among studies. A statistically significant result indicates heterogeneity while a non‐significant result is certainly no evidence for non‐heterogeneity.

An I² statistic expresses the proportion of  total variation across studies that is due to heterogeneity as supposed to chance. Values for I² lie between 0% and 100%. For the purpose of this review we categorized significant heterogeneity as an I² value >50% whereas non‐significant heterogeneity as an I² value <50%. If significant heterogeneity occurred (I²>50%) a random‐effects model was used whereas if low heterogeneity occurred (I²<50%) a fixed‐effect model was used.

Assessment of reporting bias

Attempts to gain further information were sought from trial authors however practically this was not possible since a large proportion of the trials had been carried out many years ago.

A funnel plot constructed of the effect estimates against their standard error (for overall symptoms improvements) suggests that there is publication bias resulting in asymmetry. However there were too few trials to perform Egger's regression test. 

Data synthesis

Data was analysed using a fixed‐effect model for cases of low heterogeneity (I²<50 %) and a random‐effects model for high heterogeneity between studies (I²>50%).

Subgroup analysis and assessment of heterogeneity

Various factors considered as possible causes for heterogeneity were differing formulations of phlebotonic and differing characteristics of the study populations (such as age or gender). Further investigations into subgroup analysis will be carried out in the future when sufficient studies are available.

A random methods meta‐analysis was performed to incorporate heterogeneity among studies. A meta‐analysis within the subgroups was also performed. RevMan 5 was used to perform the sub‐group analysis.

Sensitivity Analysis

We had planned to perform a secondary sensitivity analysis as per protocol, based on the presence and absence of randomised controlled trials (RCT's) published as letters and abstracts. This was not possible as there were no further RCT's published as letters and abstracts which were not already included in the meta‐analysis.

Results

Description of studies

Description of Studies

See ;Characteristics of included studies; Characteristics of excluded studies

Results Of Search

Two review authors (NP and DL) screened the results of the search. A total of seventy‐four results detailing randomised controlled trials were obtained, (MEDLINE =19; EMBASE= 23; CENTRAL = 32). After the removal of duplicates thirty‐two studies were obtained of which twenty‐four were included and eight were excluded. No trials were found through pharmaceutical companies.

Included Studies

We identified twenty‐four studies for inclusion in this review. Even though we intended to obtain studies involving the use of phlebotonics administered orally or topically at any dose in the comparisons listed under 'Types of Intervention', the majority of the included studies reported phlebotonics versus control interventions. 
 The majority of studies were performed in high income countries (20 out of 24). Seven of the trials were performed in Italy (Annoni 1986; Basile 2001; Basile 2002; Cospite 1992; Cospite 1994; La Torre 2004; Squadrito 2000). Three studies were performed in France (Godeberge 1994; Chauvenet 1994; Debien 1996).Two of the studies were performed in Singapore (Ho 1995; Ho 2000), two in China (Jiang 2006; A ba‐bai‐ke‐re 2011) and two in Turkey (Colak 2003; Mentes 2001). Each of the following countries had one study each: Germany (Belcaro 2010), Greece (Dimitroulopoulos 2005), India (Misra 2000) and Slovenia (Mlakar 2005). Four studies came from low income countries: Thailand (Thanapongsathorn 1992 and Panpimanmas 2010), Indonesia (Wijayanegara 1992) and Venezuala (Sarabia 2001). As noted the majority of trials were performed in continental Europe and Asia reflecting the increased use of phlebotonics in these regions.

Participants

The majority of studies included in this review involved both male and female adult participants. The only exception was one study, (Wijayanegara 1992), which only included pregnant women. One study, (Panpimanmas 2010), specifically included participants under the age of fifty and another (Squadrito 2000) specifically included only participants over the age of thirty.

Interventions

Of the twenty‐four trials included in this review, twenty trials evaluated the use of phlebotonic versus a control intervention. Of these twenty, four studies (Colak 2003; La Torre 2004; Mlakar 2005; A ba‐bai‐ke‐re 2011) evaluated the use of phlebotonics and controls post‐haemorrhoidectomy. One study (Wijayanegara 1992) specifically evaluated the use of phlebotonics with controls in pregnant women. Another study (Ho 2000) also evaluated the use of phlebotonics versus control versus rubber band ligation and another study (Panpimanmas 2010) also evaluated the use of phlebotonics with control and medical intervention (C.quadrangularis).

The remaining four studies included two studies which evaluated the use of phlebotonics versus phlebotonics (Cospite 1992; Sarabia 2001). Another which evaluated phlebotonics versus Infrared photocoagulation (Dimitroulopoulos 2005) and one study evaluated phlebotonics with a medical intervention (Debien 1996).

Different formulations of phlebotonics were also used. The most commonly used preparations were Micronized Purified Flavonoid Fraction (MPFF) which were used in thirteen studies (Misra 2000; Ho 2000; Colak 2003, Cospite 1994; Godeberge 1994; La Torre 2004; Ho 1995; Mlakar 2005; Dimitroulopoulos 2005; Jiang 2006; Sarabia 2001; Thanapongsathorn 1992; Panpimanmas 2010). Two studies used Diosmin (Chauvenet 1994; A ba‐bai‐ke‐re 2011) with a control intervention. Cospite's study (Cospite 1992) compared Daflon (an MPFF of Diosmin) with Diosmin.Troxerutin‐carbazochrome was used in three studies (Basile 2002; Basile 2001; Squadrito 2000) and Calcium Dobesilate in two studies (Mentes 2001; Sarabia 2001). Two studies (Annoni 1986; Wijayanegara 1992) used hydroxyethylrutoside. Belcaro's study (Belcaro 2010) used Pycnogel^R which is a flavonoid medication derived from a French maritime pine bark.

Disease verification

The included studies evaluated the use of phlebotonics for either symptomatic acute on chronic haemorrhoidal disease, chronic haemorrhoidal disease or symptom improvement post‐haemorrhoidectomy. All studies with the exception of two (Ho 1995; Mlakar 2005) stated explicit exclusion criteria.

The following studies evaluated the effect of phlebotonics post‐haemorrhoidectomy (A ba‐bai‐ke‐re 2011; Basile 2001; Basile 2002; Colak 2003; La Torre 2004; Ho 1995; Mlakar 2005). They all stated appropriate inclusion criteria with all patients being eligible for surgery.

The majority of studies evaluated the use of phlebotonics for acute haemorrhoidal disease (Belcaro 2010; Cospite 1992; Cospite 1994; Debien 1996; Chauvenet 1994; Dimitroulopoulos 2005; Godeberge 1994; Ho 2000; Jiang 2006; Misra 2000; Mentes 2001; Panpimanmas 2010; Sarabia 2001; Squadrito 2000; Thanapongsathorn 1992). These studies all stated explicit inclusion criteria and screening of these patients were often based on symptoms, signs, clinical examination and pre‐stated inclusion criteria. Studies of Wijayanegara 1992; Annoni 1986 included patients with chronic haemorrhoidal disease states. For further details of individual specific inclusion criteria please see Characteristics of included studies.

Excluded Studies

We excluded eight studies on the following grounds;

1) One study did not describe a randomised controlled clinical trial (Garner 2002)

2) Two studies described poor methods of randomizations (Mlakar 2008; Sandhu 2004) and one study poor methodology (Marsicano 1995).

3) Three studies upon closer evaluation into the literature did not use phlebotonics (Smith 1998; Tan 2006 and Quijano 2005)

4) (Kubchandani 2001) was a conference abstract describing various randomised controlled trials already included in the meta‐analysis.

Risk of bias in included studies

Allocation concealment

Allocation concealment was only adequately described in two studies (A ba‐bai‐ke‐re 2011; Jiang 2006) . In twenty‐two studies, allocation concealment was deemed 'unclear'.

Blinding

All of the studies described, with the exception of three (Belcaro 2010; Debien 1996; La Torre 2004), stated adequate methods of blinding (either as double or single‐blinded) and it was unlikely that blinding was broken.

Incomplete data outcome

One study (Panpimanmas 2010) reported incomplete data outcome which was significant to induce a plausible effect size. In one study (Ho 1995) it was unclear as to whether incomplete data outcomes had been properly addressed. In the remaining twenty‐two studies, incomplete data outcome was deemed to have been appropriately addressed.

Selective reporting

In two studies (Ho 1995; La Torre 2004) both described a situation in which selective reporting was present and not adequately addressed. In three studies (Colak 2003; Annoni 1986; Godeberge 1994) it was unclear as to whether or not there was selective reporting. The remaining nineteen studies adequately showed a low risk of selective reporting.

Other potential source of bias

In thirteen studies it was unclear whether any other sources of bias were present, (Annoni 1986; Cospite 1992; Cospite 1994; Debien 1996; Godeberge 1994; Ho 1995; La Torre 2004; Misra 2000; Panpimanmas 2010; Sarabia 2001; Squadrito 2000; Thanapongsathorn 1992; Wijayanegara 1992). This was largely due to the lack of vital information such as sources of funding. Nine studies were deemed clear of other sources of bias (A ba‐bai‐ke‐re 2011; Basile 2001; Belcaro 2010; Chauvenet 1994; Colak 2003; Dimitroulopoulos 2005; Ho 2000; Jiang 2006; Mlakar 2005). In two studies (Basile 2002; Mentes 2001), other sources of bias were present. In Basile's study the research was sponsored by a pharmaceutical company and Mentes' study had very poor compliance amongst the participants which resulted in the randomizations stopping prematurely amongst the control group.

Sequence Generation

Any study which had a quasi‐random method of sequence generation was excluded. For the twenty‐four studies that were included for analysis in this review, eleven described an adequate sequence generation (A ba‐bai‐ke‐re 2011; Annoni 1986; Basile 2002; Colak 2003; Dimitroulopoulos 2005; Debien 1996; Ho 2000; Jiang 2006; La Torre 2004; Mentes 2001; Squadrito 2000). In thirteen studies the method of sequence generation was unclear (Basile 2001; Belcaro 2010; Chauvenet 1994; Cospite 1992; Cospite 1994; Godeberge 1994; Ho 1995; Misra 2000; Mlakar 2005; Panpimanmas 2010; Sarabia 2001; Thanapongsathorn 1992; Wijayanegara 1992).

Effects of interventions

See: Table 1

i) Primary Outcomes

Pain

A variety of methods were used to measure pain. These were pain and pain scores post‐haemorrhoidectomy.

Two studies (Jiang 2006; Colak 2003) recorded the number of participants who reported pain (dichotomous variable see Figure 1). The use of phlebotonics was associated with a beneficial but not statistically significant effect for pain (OR 0.11; 95% CI 0.01 to 1.11)(P=0.06) but this was also associated with associated significant heterogeneity (I²= 85%).

1.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.1 Pain (dichotomous variable)

The following studies (A ba‐bai‐ke‐re 2011; Basile 2002; Basile 2001; Squadrito 2000; Mlakar 2005; La Torre 2004; Belcaro 2010 ) all reported pain scores, however the results could only be pooled from two studies (Mlakar 2005; La Torre 2004 ) due to the different formatting of results (see Figure 2). These two studies specifically involved post‐haemorrhoidectomy patients.

2.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.2 Pain scores (post‐haemorrhoidectomy).

Although several studies (A ba‐bai‐ke‐re 2011; Basile 2001; Basile 2002; Squadrito 2000) could not be pooled due to the different formatting of results and lack of essential study data, they demonstrated the beneficial use of phlebotonics in treating symptomatic haemorrhoids. Belcaro's (Belcaro 2010) study compared phlebotonics with a control and this demonstrated a beneficial effect in using phlebotonics for haemorrhoids. However this could not be pooled into a meta‐analysis since it did not involve post‐haemorrhoidectomy patients and was simply an isolated comparison with symptomatic haemorrhoids.

The use of phlebotonics was associated with a beneficial but not statistically significant effect for pain scores post‐haemorrhoidectomy (SMD ‐1.04; 95% CI ‐3.21 to 1.12) (P=0.35) with significant statistical heterogeneity (I²= 96%).

Bleeding

Different methods were used to measure bleeding. Bleeding (dichotomous variable) and bleeding post‐haemorrhoidectomy were the different parameters used.

Several studies reported the number of patients who experienced bleeding as an outcome measure (Jiang 2006; Misra 2000; A ba‐bai‐ke‐re 2011; Ho 1995) and these were pooled into a forest plot (see Figure 3). Two of the studies (Jiang 2006; Misra 2000) were pooled together and they demonstrated the statistically significant beneficial effect of phlebotonics in treating symptomatic haemorrhoids (OR 0.12; 95% CI 0.04 to 0.37)( P=0.0002) with no statistical heterogeneity (I²=0%). Two studies (A ba‐bai‐ke‐re 2011; Ho 1995) were pooled together (Figure 4) as these compared bleeding post‐haemorrhoidectomy with week 2 taken as a common end‐point. Phlebotonics also demonstrated a statistically significant effect in alleviating bleeding post‐haemorrhoidectomy (OR 0.18; 95% CI 0.06 to 0.58)(P=0.004) with no statistical heterogeneity I²= 0%.

3.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.3 Bleeding.

4.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.4 Bleeding post‐haemorrhoidectomy taken at 2 weeks (dichotomous variable).

Four studies reported bleeding scores (Belcaro 2010; La Torre 2004; Basile 2001; Basile 2002). Basile's studies (Basile 2002; Basile 2002) could not be pooled into a forest plot due to lack of essential study data available and different formatting of results however they both demonstrated a significant beneficial effect in using phlebotonics. Belcaro's study and La Torre's study both had zero outcomes for their continuous data and therefore these results could not be pooled into a forest plot.

Pruritis

Pruritus symptom scores were also reported in two studies, (Basile 2001; Basile 2002). These could not be pooled together due to the formatting of results presented and lack of essential data needed to calculate values such as the standard deviations. Two studies however reported the number of patients experiencing pruritus (A ba‐bai‐ke‐re 2011; Jiang 2006) and these were pooled together (Figure 5).There was a significant beneficial effect in using phlebotonic medication for pruritis (OR 0.23; 95% CI 0.07 to 0.79) (P= 0.02) and there was no statistical heterogeneity between the trials (I²=0%).

5.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.5 Pruritus.

ii) Secondary Outcome

Discharge or Leakage

Two studies (Cospite 1994; Jiang 2006) reported the number of patients who experienced discharge or leakage (Figure 6). The studies showed that phlebotonics had a statistically significant beneficial effect for discharge or leakage (OR 0.12; 95% CI 0.04 to 0.42)(P=0.0008) with no statistical heterogeneity (I²=0%).

6.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.6 Discharge or Leakage.

Two studies (Basile 2001; Basile 2002) both reported symptom scores of discharge on a VAS scale, however due to missing data and the formatting of results presented, they could not be pooled together in the meta‐analysis. They did however demonstrate the beneficial use of phlebotonics. A ba‐bai‐ke‐re's study (A ba‐bai‐ke‐re 2011) demonstrated the beneficial use of phlebotonics in improving discharge and leakage, however since this was an isolated study involving post‐haemorrhoidectomy patients this could not be pooled with other studies.

Overall symptom improvement (positive outcome)

Five studies could be pooled together (Figure 7) which assessed the overall symptom improvement comparing phlebotonics with controls. Four studies (Annoni 1986; Cospite 1994; Mentes 2001; Thanapongsathorn 1992) were pooled together into a subgroup. One study (Wijayanegara 1992) assessed the improvement in symptoms dealing with pregnant women only and this was pooled into a subgroup analysis. Overall, there is a statistically significant beneficial effect in using phlebotonic medication (OR 15.99; 95% CI 5.97 to 42.84) (P<0.00001). There was however significant statistical heterogeneity between studies (I²= 39). The subgroup analysis also revealed that this effect was statistically significant for pregnant women (P<0.00001) and non‐pregnant women (P<0.00001) with minimal statistical heterogeneity within the subgroup (I²=0%).

7.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.7 Overall symptom improvement.

Two studies (Godeberge 1994; Panpimanmas 2010) assessed the improvement for each individual symptom but not the global symptom improvement and for this reason this data could not be pooled into the meta‐analysis. Godeberge's study demonstrated that phlebotonics had a beneficial effect in symptom improvement compared with a control intervention. The Panpimanmas study however did not show any difference for symptom improvement between a phlebotonic or control intervention.

Post‐operative analgesic consumption

Three studies (Basile 2001; Basile 2002; Colak 2003) only evaluated post‐operative haemorrhoidectomy patients and recorded the number of patients who required analgesic consumption (with day 2 taken as a common end‐point). The three studies were pooled together (Figure 8). One study (Mlakar 2005), records the total number of analgesic consumption of Tramadol and Naproxen in both control and treatment groups, however these results could not be pooled with any other data for the purposes of the meta‐analysis. Phlebotonics were associated with less analgesic consumption (OR 0.54; 95% CI 0.30 to 0.99)(P=0.05) and no statistical heterogeneity (I²= 0%). This is a reflection of their beneficial effect in alleviating pain caused post‐operatively.

8.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.8 Analgesic intake (post‐operative day 2)

Adverse events (side‐effects)

Seven studies reported the number of adverse events which took place (A ba‐bai‐ke‐re 2011; Basile 2001; Basile 2002; Belcaro 2010; Cospite 1994; Ho 2000; Misra 2000). Overall the number of adverse events which took place as a result of taking phlebotonics were few and often consisted of mild gastro‐intestinal side‐effects. The results demonstrate that there is no difference between the two groups (phlebotonics and control) with regards to this outcome measure (RD ‐0.00; 95% CI ‐0.04 to 0.04)(I²=0%), see Figure 9.

9.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.9 Adverse events (side‐effects)

Non‐Compliance

Only three studies mentioned non‐compliance for those patients who completed their respective study. This demonstrated no significant difference between phlebotonic and control groups (RD 0.00; 95% CI ‐0.04 to 0.05)(I²=35%), see Figure 10.

10.

Forest plot of comparison: 1 Phlebotonic versus placebo or control intervention, outcome: 1.10 Non‐Compliance.

Discussion

The evidence suggests that there is a potential benefit in using phlebotonics for symptomatic haemorrhoids as well as for symptom relief post‐haemorrhoidectomy. In this discussion we aim to justify this argument, explore heterogeneity and address the main limitations encountered in this review.

The majority of the data supports the hypothesis that phlebotonics alleviate the symptoms of haemorrhoidal disease and post‐haemorrhoidectomy. Outcomes such as bleeding (P=0.0002 ), bleeding post‐haemorrhoidectomy (0.004), pruritus (P=0.02), discharge and leakage (P=0.0008) and overall symptom improvement (P <0.00001) indicate a statistically significant beneficial effect. Other outcomes such as pain (P=0.06), pain scores (P=0.35) and post‐operative analgesic consumption (P=0.05) indicate that there are beneficial effects, but they were not statistically significant. Evidence from the trials suggests there were no issues regarding adverse events of phlebotonics or patient compliance. The majority of data demonstrates the beneficial use of phlebotonics with some outcomes being statistically significant and because of this, we recommend their use. However limitations were encountered and these need to be taken into consideration.

Among the limitations, many trials used subjective clinical parameters such as symptom scales. These scales could not be objectively defined and varied widely between studies and the resulting self‐assessment scales were therefore subjective. Several trials (Colak 2003; Mlakar 2005; Basile 2001; Basile 2002) used a linear visual analogue scale which required the patients to assess their own symptoms and other studies used a patients' self‐evaluation scale. These methods of assessments would have resulted in a subjective interpretation of the scales used by different patients. However despite this, there were also more objective quantitative measurements of assessment. These included dichotomous variables such as the presence of bleeding, pruritus, discharge or leakage. This would have resulted in a higher quality of results.

Other major methodological limitations encountered related to issues regarding dietary and lifestyle measures. Haemorrhoidal disease is more likely to occur as a result of low‐fibre diets, poor nutrition, lack of exercise, obesity and poor lifestyle intervention (Misra 2000). Although dietary advice was given in some cases, the individual diet of patients in the majority of trials could not be rigorously controlled, thus leading to bias in the results obtained. Several trials (Mentes 2001; Jiang 2006) clearly state that although dietary advice was given, there was no feasible method of ensuring that strict compliance was adhered to and this would have undoubtedly been the case in the majority of studies, thus leading to bias. Even the geographical location would have had an effect on the diet with Misra's study (Misra 2000) suggesting that the high‐fibre diets used by the North Indian subcontinent was beneficial in treating haemorrhoids. Misra's study then subsequently made the conclusion that micronized purified flavonoid fraction (MPFF) had a positive effect above and beyond the good dietary intervention received by the North Indian community.

There was also little similarity in the overall summary estimates used by investigators to describe the incidence, severity and duration of haemorrhoidal disease symptoms. Consequently for some outcomes, it was not possible to pool all of the data together. Most investigators required their participants to record the clinical severity of symptoms and signs and used similar scales to estimate this. Continuous outcome measures included the use of the linear visual analogue scales (VAS scales), whereas other trials used a symptom severity scale which rated symptoms on a scale (e.g. none, mild, moderate or severe). Several trials used dichotomous results to report the presence or absence of a particular sign such as bleeding. For several studies, it was not possible to use the data since essential values such as the standard deviation, standard error of the mean or confidence intervals were not provided and could not be calculated from the available data. Attempts were made to contact the authors involved to provide the missing data, but in many cases this was not possible due to the lack of contact information available.

The majority of studies compared phlebotonics with a control intervention. Very few trials compared these with surgical or medical interventions and because of this, a conclusion could not be made with regard to these outcomes. Only two studies compared phlebotonics with other herbal medical therapies. The study by Panpimanmas (Panpimanmas 2010) compared MPFF with a placebo as well as Cissus quadrangularis. Cissus quadrangularis is a herbal extract found mainly in Asia. This study concluded that C.quandragularis and Daflon showed no difference in the treatment of acute haemorrhoids. Debien's study (Debien 1996) compared phlebotonics with Gingko biloba herbal extract. Because the comparisons involved two entirely different substances it was decided not to pool these outcomes together into a forest plot.

Only two studies obtained compared phlebotonic drugs with procedures such as rubber band ligation (Ho 2000) and infrared photocoagulation (Dimitroulopoulos 2005). These two studies used different procedures and because of this it was inappropriate to pool these comparisons into a meta‐analysis. Ho's study concluded that although the addition of MPFF to fibre supplements accelerates the relief of bleeding from non‐prolapsed internal haemorrhoids, no statistically significant advantage could be found over rubber band ligation. The study by Dimitroulopoulos compared the effect of a treatment combining IRP and oral MPFF versus each treatment alone on bleeding cessation in patients with acute internal haemorrhoids. The study demonstrated no statistical difference between the IRP group and the MPFF group. However five days of treatment combining oral MPFF with local application of IRP was beneficial in reducing the bleeding of internal haemorrhoids.

Statistical heterogeneity between studies was encountered and several reasons for this are identified. One of the key factors was the dose of phlebotonics. Different formulations were used and subsequently the therapeutic regime greatly varied between the studies. Another factor resulting in statistical heterogeneity were the different study populations used. Some trials were carried out in mainland Europe whereas others in Asia. The dietary requirements and lifestyles of these two groups would have differed greatly with populations in Asia receiving a high‐fibre diet and a more active lifestyle than their European counterparts (Misra 2000). This would have subsequently resulted in a different response rate of haemorrhoids to the treatment provided depending on their geographical location.

Summary of main results

Studies reporting the effect of phlebotonics for haemorrhoidal disease suggest that there is a potential benefit in their use. Several outcomes suggest that there is a statistically significant beneficial effect whereas other outcomes, although beneficial were not statistically significant. There was however some significant statistical heterogeneity between trials and methodological limitations were encountered.

There is limited data comparing the effects of phlebotonics with medical treatment or procedures such as rubber band ligation and sclerotherapy.

Potential biases in the review process

All trials assessed the effect of phlebotonics in treating haemorrhoidal disease. For several trials there was missing data and many results could not be pooled together. Although these were addressed in the final discussion, they would have inevitably led to bias in the meta‐analysis.

There was also the potential to miss trials or grey literature, which mentioned haemorrhoidal symptoms (under the spectrum of anorectal pathologies) as secondary outcome measures. These would have been less well‐publicised trials. Conducting a thoroughly comprehensive search and identifying the relevance of each paper would have largely avoided this.

Agreements and disagreements with other studies or reviews

There have been no previous meta‐analyses evaluating the effectiveness of phlebotonic medications as a whole in their treatment of haemorrhoidal disease. However, there have been review articles evaluating the effectiveness of flavonoids in the treatment of haemorrhoids. One meta‐analysis (Alonso‐Coello 2006) mentioned that flavonoids seemed to have a beneficial effect in the symptomatic treatment of haemorrhoids. However, the authors also mentioned that flaws in the study design existed and these weakened any inferences of the effect of phlebotonics on symptomatic haemorrhoids. The review article also used fewer outcome measures to evaluate the effect of the medications. A review article (Hain 2011) made a very similar conclusion in that the true effect of flavonoid medications were still unknown.

Whilst we greatly appreciate the arguments presented in these reviews, the scope of our review is slightly different in that we evaluated phlebotonics as a whole group (flavonoids and non‐flavonoid medications) and thus included more studies. These previous reviews presented a rigorous evaluation of flavonoids only. We have however presented data which supports the beneficial use of phlebotonics with more trials included. We also evaluated more outcome measures. To enhance our conclusion further we certainly recommend more rigorous trials to be carried out in the future taking into account various methodological limitations.

Authors' conclusions

Implications for practice.

The evidence presented in this review suggests that there is a benefit in using phlebotonics for symptomatic haemorrhoids as well as symptom alleviation post‐haemorrhoidectomy and there is little concern regarding their overall safety from the trials presented. Adverse events from the studies mainly consisted of mild isolated gastro‐intestinal disturbances. We do therefore recommend the use of phlebotonics in treating haemorrhoidal disease as there is a proven benefit. However although there is a beneficial effect, not all of the pooled data is statistically significant. The methodological quality is largely moderate with associated risk of bias in included studies (see Figure 11 and Figure 12) and some of the confidence intervals were wide with statistical heterogeneity. The funnel plot constructed (Figure 13) also indicates that there is some publication bias present.

11.

Methodological quality graph: Review authors' judgements about each methodological quality item presented as percentages across all included studies.

12.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

13.

Funnel plot of comparison: 1 Phlebotonic versus placebo or control intervention, for outcome: 1.7 Overall symptom improvement.

None of the trials examined the use of phlebotonics for thrombosed haemorrhoids and very limited data exists examining their role in pregnant women or the more severe stages of haemorrhoidal disease. We therefore cannot recommend their use in these specific subgroups of patients.

In summary, we recommend the use of phlebotonics for haemorrhoids in clinical practice. However, more robust clinical trials are needed to further support our conclusion and to make further recommendations with regards to specific sub‐groups of patients and other medical and surgical treatment options.

Implications for research.

There is a greater prevalence of haemorrhoids among the Caucasian adult male and female population (age groups of 45‐60 years). The numerous trials obtained mainly compared the use of phlebotonics with control interventions for the less severe stages of haemorrhoidal disease. This review demonstrates the benefit in using phlebotonics for haemorrhoids and a conclusion can be made with regards to this. To further support this conclusion, we outline several recommendations for a future research agenda. We suggest the use of similar end‐points for outcome measures so that more data can be pooled appropriately. We recommend that investigators succinctly describe their methods of randomizations and allocation concealment so that the methodological quality of the trials is greater. Although phlebotonics are a heterogenous class of drugs, we recommend that future studies use similar or identical agents as well as therapeutic regimes in studies which have already taken place. This will inevitably minimize the statistical heterogeneity between trials.

There are very few trials which evaluate the use of phlebotonics and procedures such as rubber band ligation or sclerotherapy. We recommend that more future trials evaluate these comparisons. We also recommend that more trials investigate the effect of phlebotonics with thrombosed haemorrhoids.

What's new

Date	Event	Description
6 March 2012	Amended	For editors

History

Protocol first published: Issue 3, 2003
 Review first published: Issue 8, 2012

Date	Event	Description
15 November 2010	Amended	Converted to new review format.

Data and analyses

Comparison 1. Phlebotonic versus placebo or control intervention.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pain	2	190	Odds Ratio (M‐H, Random, 95% CI)	0.11 [0.01, 1.11]
2 Pain scores (post‐haemorrhoidectomy)	2	113	Std. Mean Difference (IV, Random, 95% CI)	‐1.04 [‐3.21, 1.12]
2.1 Up to seven days	1	63	Std. Mean Difference (IV, Random, 95% CI)	0.05 [‐0.45, 0.54]
2.2 Up to ten days	1	50	Std. Mean Difference (IV, Random, 95% CI)	‐2.17 [‐2.87, ‐1.46]
3 Bleeding	2	190	Odds Ratio (M‐H, Fixed, 95% CI)	0.12 [0.04, 0.37]
3.1 Bleeding at day 7	2	190	Odds Ratio (M‐H, Fixed, 95% CI)	0.12 [0.04, 0.37]
4 Bleeding post‐haemorrhoidectomy (dichotomous variable)	2	314	Odds Ratio (M‐H, Fixed, 95% CI)	0.18 [0.06, 0.58]
5 Pruritus	2	176	Odds Ratio (M‐H, Fixed, 95% CI)	0.23 [0.07, 0.79]
5.1 Post‐haemorrhoidectomy	1	86	Odds Ratio (M‐H, Fixed, 95% CI)	0.25 [0.06, 0.97]
5.2 Acute haemorrhoidal episodes	1	90	Odds Ratio (M‐H, Fixed, 95% CI)	0.16 [0.01, 3.42]
6 Discharge and Leakage	2	139	Odds Ratio (M‐H, Fixed, 95% CI)	0.12 [0.04, 0.42]
7 Overall symptom improvement	5	368	Odds Ratio (M‐H, Random, 95% CI)	15.99 [5.97, 42.84]
7.1 Pregnant women	1	97	Odds Ratio (M‐H, Random, 95% CI)	61.63 [17.49, 217.25]
7.2 Non‐pregnant women	4	271	Odds Ratio (M‐H, Random, 95% CI)	9.63 [3.80, 24.41]
8 Analgesic consumption	3	202	Odds Ratio (M‐H, Fixed, 95% CI)	0.54 [0.30, 0.99]
9 Adverse events	7	511	Risk Difference (M‐H, Fixed, 95% CI)	‐0.00 [‐0.04, 0.04]
10 Non‐Compliance	3	288	Risk Difference (M‐H, Fixed, 95% CI)	0.00 [‐0.04, 0.05]

1.1. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 1 Pain.

1.2. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 2 Pain scores (post‐haemorrhoidectomy).

1.3. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 3 Bleeding.

1.4. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 4 Bleeding post‐haemorrhoidectomy (dichotomous variable).

1.5. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 5 Pruritus.

1.6. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 6 Discharge and Leakage.

1.7. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 7 Overall symptom improvement.

1.8. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 8 Analgesic consumption.

1.9. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 9 Adverse events.

1.10. Analysis.

Comparison 1 Phlebotonic versus placebo or control intervention, Outcome 10 Non‐Compliance.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

A ba‐bai‐ke‐re 2011.

Methods	Randomized observer‐blinded randomised controlled trial.
Participants	Mean age of controls 53.2 years and mean age of Diosmin was 51.3 years. Male and female participants. China, university hospital surgical department. Eighty‐six consecutive patients aged 12‐75 years with grade III and grade IV acute mixed haemorrhoids and indication for haemorrhoidectomy.
Interventions	Milligan‐Morgan haemorrhoidectomy performed on all patients with the exception of three who were subject to a haemorrhoidopexy. The patients were then subject to two groups; Group 1; Diosmin 500mg, three tablets twice daily, after meals for three days followed by two tablets twice daily from day four to day seven. Group 2; Identical placebo regimen.
Outcomes	Assessment of pain via patient self‐questionnaires; verbal response scales and visual analogue scales. Number of patients experiencing post‐operative symptoms of bleeding, heaviness, pruritis and mucosal discharge at pre‐specified time intervals of two and eight weeks.
Notes	Exclusion criteria were patients complicated with fistula or anal fissure, inflammatory bowel disease, dermatitis, proctitis, pregnancy, severe cardiovascular state or pulmonary complications.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Sealed envelopes were used
Blinding (performance bias and detection bias) All outcomes	Low risk	Observer blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up
Selective reporting (reporting bias)	Low risk	Outcome measures were stated and these were addressed in the results
Other bias	Low risk	No baseline imbalance
Adequate sequence generation	Low risk	Computer‐based randomizations

Annoni 1986.

Methods	Double‐blinded randomised controlled trial
Participants	Mean age of participants was 42.1 +/‐ 12.48 years Male and female Italy, University hospital setting Fifty patients (33 male and 17 female) with 2nd, 3rd and 4th degree haemorrhoids were enrolled
Interventions	Patients were randomised to receive two treatments; Group 1; 20 patients (12 male and 8 female) were treated with O‐rutosidea (HR), 4 g per day in total (2 sachets in 1/2 glass water twice a day) Group 2; Comparable placebo with similar therapeutic regime
Outcomes	Haemorrhoidal symptoms based on a visual analogue scale (VAS score) assessing the symptoms of pain, discharge, bleeding and inflammation. Assessment was performed by doctors by rectal examination and symptoms graded on a 4 point severity scale (0= absent, 1=mild, 2=moderate, 3=severe)
Notes	Exclusion criteria were treatment with NSAIDS or steroids in the previous 30 days or surgical treatment in the past 6 months. Also excluded were those treated with analgesics, anti‐inflammatories, local anaesthetics or ice.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded and unlikely that blinding was broken
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete data outcome
Selective reporting (reporting bias)	Unclear risk	Outcomes assessed were stated and these outcomes were addressed
Other bias	Unclear risk	No baseline imbalance however Arcapia Lda is a pharmaceutical company which supported the research
Adequate sequence generation	Low risk	Randomly allocated central number table

Basile 2001.

Methods	Randomized double‐blinded placebo controlled phase IV study.
Participants	Average age of participants were 47 years in the Troxerutin group and 41 years in the placebo group. Male and female participants. Italy, university hospital setting. Thirty patients were enrolled within the age group of 18‐60 years with confirmed haemorrhoidal disease which had an indication for either surgery or rubber band ligation.
Interventions	Following haemorrhoidectomy patients were randomised to either two groups; Group 1; Troxerutin 150 mg and carbazochrome 1.5 mg in combination. 3ml ampoules twice a day for five consecutive days after surgical procedure. Group 2; Placebo, IM 3 ml ampoules twice a day for five consecutive days.
Outcomes	Haemorrhoidal symptoms based on a visual analogue score (VAS), pain, discharge, bleeding and pruritus. Pain medication intake. Time taken to restore physiological defecation. Oedema evaluation based on four‐point evaluation scale (0=none, 1=mild, 2=moderate, 3=severe). Camera pictures taken at T1 and T4. Blood coagulation tests.
Notes	Exclusion criteria were unwillingness to co‐operate, poor motivation, emotional or intellectual problems, participation in other studies involving marketed products, history of allergies or hypersensitivities, pregnancy or lactation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data reported
Selective reporting (reporting bias)	Low risk	Primary and secondary outcome measures as assessed by the study protocol were addressed
Other bias	Low risk	No other bias to be reported
Adequate sequence generation	Unclear risk	Not specified

Basile 2002.

Methods	Double‐blinded randomised controlled study.
Participants	Mean age of participants was 45.6 +/‐ 7.4 years. Male and female participants. Italy, university hospital setting. Sixty patients (33 male and 27 female) aged 18‐65 years were enrolled for haemorrhoidectomy or rubber band ligation and then randomised into two treatment groups.
Interventions	Following a haemorrhoidectomy or rubber band ligation patients were randomised to receive one of two treatments; Group 1; troxerutin 150mg plus carbazochrome 1.5 mg in 3 ml ampoules twice a day for five consecutive days after surgery starting on the day of surgery. Group 2; Comparable placebo with similar therapeutic regime.
Outcomes	Haemorrhoidal symptoms based on a visual analogue scale (VAS score) assessing the symptoms of pain, discharge, bleeding and inflammation. Oedema based on a four‐point scale; (0=absent, 1=mild, 2=moderate, 3=severe).
Notes	Exclusion criteria were chronic or severe disease (diabetes mellitus, coagulation disorders/oncological diseases) altered hepatic function, renal function, portal hypertension or pelvic tumours. Unwillingness to co‐operate, poor motivation, emotional problems, participation in other studies involving investigational or marketed products within one month preceding study entry, a history of allergy to the phlebotonics used, pregnancy and lactation were also exclusion criteria.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Yes, double‐blinded controlled study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete outcome data
Selective reporting (reporting bias)	Low risk	The outcomes to be addressed were stated and all outcomes were addressed
Other bias	High risk	No baseline imbalance. Arcapia Lda is a pharmaceutical company which supported the research
Adequate sequence generation	Low risk	Randomly allocated using a cental number table

Belcaro 2010.

Methods	Randomized controlled clinical trial evaluating PycnogenolR treatment for acute external haemorrhoidal episodes.
Participants	Average age for the oral PycnogenolR group and the placebo were 49.3 years and 48.9 years respectively. Male and female adult participants. Germany, university hospital setting. Eighty‐four participants suffering from an acute episode of external haemorrhoids (lasting 24‐48 hours prior to exclusion were chosen).
Interventions	Patients were randomised to the following treatment groups; Group A; High dose PycnogenolR. Group B; Comparable placebo. Group C; topical 0.5% Pycnogenol R cream in combination with high dose Pycnogenol R oral treatment as in group A. Group D; Oral Pycnogenol R in combination with topical sham cream.
Outcomes	Frequently observed signs and symptoms which included acute intravascular thrombosis, acute severe perianal pain, purple/black edematous and tense subcutaneous perianal mass, tenderness, ischaemia/necrosis of the overlying skin and bleeding.
Notes	Exclusion criteria were; i) Patients requiring any other cardiovascular treatment. ii) Recently received surgery. iii) Diagnosed with infections/tumours. iv) Severely handicapped. v) The need for surgery (bleeding, thrombosis, prolapse, pain not responding to medical treatment or other complications). vi) Portal hypertension.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Primary and secondary outcomes of interest have been reported
Other bias	Low risk	Free of other sources of bias
Adequate sequence generation	Unclear risk	Sequence generation not fully specified

Chauvenet 1994.

Methods	Double‐blinded controlled multicentric randomised controlled clinical trial
Participants	94 participants male and female in a French hospital setting. Patients were suffering from either external haemorrhoidal thrombosis requiring medical treatment or internal congestive haemorrhoidal crisis.
Interventions	Diosmin; 1800 mg Diosmin for four days divided into 3 doses, (morning, noon and evening). Then 1200 mg Diosmin divided into three doses (morning, noon and evening) for four days Placebo; Similar placebo regimen.
Outcomes	Pain and oedema/congestive status on D0 and D7 (main criteria) Global score of other signs D0 and D7, overall assessment of investigator on D7 and pain relief
Notes	Exclusion criteria were anal fissure, haemorrhoidal prolapse, thrombosed haemorrhoidal prolapse, cryptitis, phlebitis, anti‐inflammatory drugs such as NSAIDs or steroids and intolerance to Diosmin
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded and blinding unlikely to be broken
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete data outcome
Selective reporting (reporting bias)	Low risk	No selective reporting
Other bias	Low risk	No baseline imbalance
Adequate sequence generation	Unclear risk	Not specified

Colak 2003.

Methods	Prospective randomised controlled observer‐blinded clinical study.
Participants	Median age for MFF group was 41 (22‐86) years and median age of control group was 45 (32‐63) years. Male and female participants. Turkey, university hospital setting. 112 patients who had symptomatic 3rd or 4th degree haemorrhoids requiring haemorrhoidectomy were enrolled.
Interventions	Post‐haemorrhoidectomy the patients were randomly assigned into the following groups; Group 1; MFF for 1 week after haemorrhoidectomy. Group 2; Control group for 1 week after haemorrhoidectomy (not receiving MFF).
Outcomes	Pain scores after haemorrhoidectomy based on a linear analogue scale (0‐10). Number of patients who required intramuscular analgesic injections. Comparison of urinary retention, hospital stay, and patient satisfaction levels.
Notes	Patients with concomitant anal disease such as fistulae, fissures, Crohn's, ulcerative colitis, rectal cancer or those taking oral anticoagulants were excluded.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Blinding took place of key outcome assessors
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	Study protocol was available and primary and secondary outcome were assessed
Other bias	Low risk	No other source of bias to be specified
Adequate sequence generation	Low risk	Computer‐generated table

Cospite 1992.

Methods	Randomized double‐blinded controlled trial.
Participants	59 adult patients with haemorrhoids. University hospital setting, Italy. Patients had at least one acute episode in the 2 months prior to entry.
Interventions	Group 1; Daflon 500mg. Group 2; Diosmin 900 mg.
Outcomes	Grading of anal symptoms on a four‐point scale (0‐3). Frequency and duration of acute episodes. Patient satisfaction scales. Investigator satisfaction scales.
Notes	Exclusion criteria were associated recto‐sigmoidal conditions, progressive local parasitic infestations, anal fissure, complete and effective prior haemorrhoidectomy, entropic treatment in the previous two months and treatment with any other experimental drug, concomitant therapy with anti‐inflammatory drugs, anticoagulants or platelet anti‐aggregates.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded conditions and unlikely to be broken
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete data outcome
Selective reporting (reporting bias)	Low risk	No selective reporting
Other bias	Unclear risk	Other bias not specified
Adequate sequence generation	Unclear risk	Not specified

Cospite 1994.

Methods	Randomized, double‐blinded placebo controlled study.
Participants	Mean age of participants was 44.6 years in Daflon group and 42.6 years in placebo group. Male and female regardless of age or sex. 100 participants were recruited with a history of haemorrhoidal disease confirmed by endoscopic examination suffering from an uncomplicated and acute attack of haemorrhoids lasting no more than three days.
Interventions	Group 1; Daflon 500 mg for seven days at the dosage of six tablets daily (3 bd) during the first four days and four tablets daily (2 bd) during the following three days. Group 2; Placebo.
Outcomes	Number of patients with symptoms; proctorrhagia, anal discomfort, pain, anal discharge, proctitis.
Notes	Exclusion criteria were; need of a surgical procedure, associated anal fissure, permanent prolapse, patients treated with phlebotonics, anticoagulant, anti‐inflammatory drugs, those taking analgesic or topical medications not authorized from the study protocol.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded provision maintained and unlikely to be broken
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete data outcome sufficient to introduce a plausible effect size
Selective reporting (reporting bias)	Low risk	Outcome measures were pre‐specified and these were stated
Other bias	Unclear risk	Other bias not specified
Adequate sequence generation	Unclear risk	Not specified

Debien 1996.

Methods	Multicentric randomised controlled trial
Participants	Patients who were aged eighteen who were suffering from a haemorrhoid crisis defined by pain lasting for at least 48 hours and of intensity superior or equal to 6/10 on an analogue scale. Clinical exam had to reveal one of the following; congestion, external haemorrhoid oedema, thrombosis with no surgical indication and swelling of internal haemorrhoids.
Interventions	Group 1; Diosmin. One tablet morning, noon and evening for seven days. Group 2; Gingko biloba‐heptaminol‐troxerutin. Two tablets in the morning, one tablet at midday and one tablet the evening.
Outcomes	Intensity of pain was recorded on a Visual analogue scale daily two hours after taking the medication. Three days after the medication was complete patients were evaluated for any functional or physical signs of a haemorrhoid crisis. Compliance and adverse effects were also recorded.
Notes	Exclusion criteria included any other associated pathologies including rectal, sigmoid, colon, intestinal or proctological other than haemorrhoid crisis. If there were any of the following treatment; venotonics, anticoagulation, antithrombotics, anti‐inflammatories (NSAIDS or steroids) or any previous topical treatment to haemorrhoids. Also patients with intolerance to the treatment, pregnant women, lactating women and sexually active women who did not take contraception were also excluded.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not present
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not mentioned
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow up
Selective reporting (reporting bias)	Low risk	No reporting bias present
Other bias	Unclear risk	Not specified
Adequate sequence generation	Low risk	Computer based randomizations

Dimitroulopoulos 2005.

Methods	Prospective randomised controlled single‐blinded study.
Participants	Mean age was 49.2 years (range, 29‐71 years). Male and female participants. Presenting with rectal bleeding due to grades I, II and III acute internal haemorrhoids with no previous treatment for haemorrhoids within the last 6 months and without colonic disease. Greece, university hospital setting. A total of 351 patients presenting with rectal bleeding due to grades I, II and III acute internal haemorrhoids were enrolled into the study.
Interventions	Patients (117 each) were randomly assigned into the following groups; Group 1; MPFF and IRP group; Patients received 6 tablets (3000mg) of oral MPFF daily in 2 equally divided doses, after breakfast and before dinner for five days. Group 2; MPFF group; Patients assigned to give MPFF dose as above. Group 3; IRP group; Patients underwent an IRP procedure.
Outcomes	Primary end‐point; rate of bleeding cessation (expressed in percentage of patients with no bleeding) within five days after intervention. Secondary end‐point; prevention of relapse within 90 days after intervention.
Notes	Exclusion criteria were pregnancy, history of acute haemorrhoidal attack occuring at least 6 months before enrolment, grade IV haemorrhoids, concomitant large‐bowel and anal canal diseases, history of pelvic radiation, use of anti‐coagulation, anti‐platelet agents and non‐steroidal anti‐inflammatory drugs.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Blinding of observers took place
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete outcome data
Selective reporting (reporting bias)	Low risk	No selective outcome reporting took place
Other bias	Low risk	No other obvious source of bias
Adequate sequence generation	Low risk	Random assignment using sealed envelopes

Godeberge 1994.

Methods	Double‐blinded, placebo controlled, randomised controlled trial.
Participants	Mean age of patients in Daflon group was 46 years and men age of patients in Placebo group was 48 years. Male and female participants aged over 18 years. Hospital setting France. One hundred and twenty outpatients (54 men,66 women) suffering from an acute episode of symptomatic haemorrhoidal disease during the previous two months.
Interventions	Group 1; Daflon 500mg, two tablets daily. Group 2; placebo compound.
Outcomes	Changes in number, mean duration, mean severity of attacks. Change in mean score per feature. Change in overall symptom and sign score. Percentage of patients improved in each group.
Notes	Exclusion criteria were indication for non‐medical treatment (permanent irreducible prolapse, external thrombosed pile, anal fissure; documented concomitant disease, previous haemorrhoidectomy; treatment with any of the following at inclusion or in the previous two months (venotropic agent, anticoagulant, platelet antiaggregant, anti‐inflammatory agent).
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded provisions maintained and unlikely to be broken
Incomplete outcome data (attrition bias) All outcomes	Low risk	Plausible effect size among missing outcomes not enough to have a clinically relevant impact on intervention effect estimate
Selective reporting (reporting bias)	Unclear risk	Insufficient information to permit judgement of 'yes' or 'no' answer
Other bias	Unclear risk	Other bias not specified
Adequate sequence generation	Unclear risk	Not specified

Ho 1995.

Methods	Randomized controlled observer‐blinded clinical trial.
Participants	Mean age of patients was 39.5 years. Male and female. Singapore, general hospital setting. 228 consecutive patients with symptomatic prolapsed irreducible haemorrhoids were recruited for a standardized diathermy excision of three piles without ligation of pedicles.
Interventions	Post elective haemorrhoidectomy patients were randomised into two groups (114 each); Group 1; Daflon 500 mg (two tablets three times daily for three days, followed by one tablet three times daily for the four days). Group 2; Control group.
Outcomes	Reactionary haemorrhage. Secondary haemorrhage (occuring 7‐14 days post‐op).
Notes	Exclusion criteria were not stated.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Observer‐blinded only
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Unclear
Selective reporting (reporting bias)	High risk	No selective reporting
Other bias	Unclear risk	No exclusion criteria were stated
Adequate sequence generation	Unclear risk	Not specified

Ho 2000.

Methods	Randomized controlled observer‐blinded clinical trial.
Participants	Mean age of participants were 44.7 years in the FIBER group, 44.4 years in the BAND group and 41.4 years in the MPFF group. Male and female adult participants. Singapore, general hospital setting. 162 patients with bleeding non‐prolapsed haemorrhoids were recruited and randomly assigned to the three groups.
Interventions	162 patients were randomly assigned to the following treatment groups; Group 1; FIBER group. 3.5g sachet of ispaghula husk twice daily for three months. Group 2; BAND group. Rubber band ligation of internal piles. Group 3; MPFF group. Micronized Purified Flavonoidic Fraction (DaflonR) in addition to a similar dose and duration of ispaghula husk.
Outcomes	Time taken for relief of haemorrhoidal bleeding. Recurrence of haemorrhoidal bleeding at 6 month follow‐up.
Notes	Exclusion criteria were previous treatment for the episode of bleeding, no history of bleeding disorders and total colonoscopy was performed to exclude other sources of per rectum bleeding.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Key study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Primary and secondary outcome measurements were detailed and these were addressed in the results
Other bias	Low risk	No other sources of bias
Adequate sequence generation	Low risk	Random allocation with sealed envelopes.

Jiang 2006.

Methods	Randomized double‐blinded controlled clinical trial.
Participants	Mean age of participants were 43.2 years (range 18‐76 years). Male and female. Chinese nationality, university hospital setting. Ninety patients aged over eighteen experiencing an acute haemorrhoidal disease episode for less than 48 hours were enrolled.
Interventions	Patients were randomised into the following therapeutic regimen; Group 1; Micronized Purified Flavonoid Fraction (MPFF); six tablets per day for four days followed by four tablets per day for three days. Group 2; Active placebo with identical therapeutic regime.
Outcomes	Symptoms and signs that were recorded by investigators using a four‐point grading scale. Functional improvement in symptoms as recorded by the patient. Global efficacy of treatment as recorded by patients and investigators. Safety of treatment recorded on day four and day seven.
Notes	Exclusion criteria were the presence of severe haemorrhoidal manifestations, anal fissure, permanent prolapsed haemorrhoids, parasitic infection, systemic pathology requiring treatment. Phlebotonic medication, anti‐coagulant and anti‐platelet agents, analgesic or anti‐inflammatory drugs as well. Also excluded were patients less than 18 years old, those taken part in another clinical trial, patients with a past history of GI resection, alcohol or drug abuse, allergies to diosmin or any other phlebotonic medication, abnormal hepatic enzymes or abnormal renal function.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Yes explicitly stated allocation concealment
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded condition were applied
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete outcome data were present
Selective reporting (reporting bias)	Low risk	All assessments were reported and primary and secondary end‐points were stated as appropriate
Other bias	Low risk	Other bias not specified
Adequate sequence generation	Low risk	Computer generated sequential method

La Torre 2004.

Methods	Randomized controlled clinical trial evaluating the combination of micronized purified flavonoid fraction with short‐term routine antibiotics and anti‐inflammatory therapy.
Participants	Fifty‐patients with an average age of 56 years were (35‐74) years were recruited. Male and female participants aged over 18 years. Italy, university hospital setting. Fifty patients with an average age of fifty‐six with irreducible prolapsed haemorrhoids and an indication for haemorrhoidectomy, past history of haemorrhoids longer than six months, symptomatic irreducible prolapsed haemorrhoids and informed consent were selected.
Interventions	Patients were randomly assigned to the following treatment groups immediately after surgery; Group 1; Posthaemorrhoidectomy antibiotic and anti‐inflammatory therapy (control group). Group 2; Posthaemorrhoidectomy antibiotic and anti‐inflammatory and MPFF.
Outcomes	Self‐assessment of specific symptoms such as pain,bleeding, tenesmus and pruritus. These were scored with an observation period of 60 days.
Notes	Exclusion criteria were fistula, chronic anal fissure, inflammatory bowel disease, diabetes, other metabolic or endocrine disorders, coagulation disorders and previous anorectal surgery.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not specified
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete outcome data
Selective reporting (reporting bias)	High risk	Only analysed data on patient's self assessment of symptoms. Did not include data from PRs and anoscopy
Other bias	Unclear risk	Othe sources of bias not specified
Adequate sequence generation	Low risk	Computer generated random allocation

Mentes 2001.

Methods	Randomized double‐blinded controlled study.
Participants	Median age of Doxium and controlled diet group were 43 (range 19‐59) years and 42 (18‐60) years respectively. Turkish nationality, university hospital setting. 45 adult patients with 1o or 2o degree internal haemorrhoids. Internal symptoms had existed for longer than six months and rectal bleeding was currently present indicating an acute attack.
Interventions	Patients were randomised into the following groups; Group 1; (controlled group) n=16. Dietary and lifestyle advise measures only. Group 2; (Doxium R) n=30. Loading dose of Doxium R three tablets per day in three divided doses for one week followed by two tablets daily for one week and dietary and lifestyle advise.
Outcomes	Relief of symptoms (2=no relief of symptoms/worsening), 1=partial relief of symptoms, 0=significant relief or lack of symptoms). Anitis score. Cessation of bleeding plus lack of severe anitis.
Notes	Exclusion criteria were concurrent fistulae, chronic anal fissure, inflammatory bowel disease, diabetes or other metabolic or endocrine abnormalities, alcoholism, drug abuse, coagulation disorders, abnormal sexual habits, previous anorectal surgery or previous treatment of haemorrhoidal disease with any method other than diet or topical agents.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Blinding of key study personnel
Incomplete outcome data (attrition bias) All outcomes	Low risk	Incomplete data outcome addressed
Selective reporting (reporting bias)	Low risk	Study protocol was available and primary and secondary outcomes were addressed
Other bias	High risk	Poor compliance with dietary measures meant that randomizations was stopped early when the number of control patients reached 16 (instead of 30)
Adequate sequence generation	Low risk	Simple randomizations was used using a random table

Misra 2000.

Methods	Randomized double‐blinded controlled clinical trial.
Participants	Mean age of the MPFF and placebo group were 34.8 years and 33.3 years respectively. Male and female aged over 18 years of age. India, university hospital setting. 100 consecutive patients were chosen who had a past history of haemorrhoids of less than 18 months presenting with acute rectal bleeding of less than three days in association with visible grade I or II internal haemorrhoids upon proctoscopic examination.
Interventions	Group1; Micronized Purified flavonoid fraction; Two sequential phases. First phase involved taking six tablets per day for four days and four tablets per day for three days. In the second phase (prevention of relapse) patients whose bleeding had stopped by the end of the first phase took two tablets per day for 83 days. Group 2; Placebo; identical regimen.The placebo tablet had an identical appearance.
Outcomes	Primary outcome measure was the number of patients with no bleeding on the third completed day of treatment. Secondary outcomes measures were the number of patients in whom relapse of bleeding was prevented and the mean time between cessation of bleeding and relapse.
Notes	Exclusion criteria were presence of anal fissure, inflammatory bowel disease, colorectal cancer, pregnancy, previous laser treatment, use of a flavonoid drug 1 month before inclusion. Also treatment with analgesics, topical anti‐haemorrhoidal ointments, NSAIDS, steroids, anticoagulants or anti‐platelet agents.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded study and unlikely that provisions were broken
Incomplete outcome data (attrition bias) All outcomes	Low risk	Plausible effect size among missing outcomes not enough to have a clinically relevant impact on observed effect
Selective reporting (reporting bias)	Low risk	Primary and secondary outcome measures were stated and these were included
Other bias	Unclear risk	Other source of bias not specified
Adequate sequence generation	Unclear risk	Not specified

Mlakar 2005.

Methods	Randomized controlled observer‐blinded clinical trial.
Participants	Average age of participants were 49 years. Male and female. Slovenia, medical centre. 63 patients (44 females and 17 males) with 3rd degree haemorrhoids had ambulatory stapled haemorrhoidopexy under spinal anaesthesia during the period of one year and were randomised to either the control or Detralex group.
Interventions	Following haemorrhoidopexy the patients were randomised to the following groups; i) Detralex; two tablets three times daily for five days after the operation. ii) Control.
Outcomes	Self‐assessment of the presence of blood on defecation, degree of pain and consumption of analgesics for the first week after the operation.
Notes	Exclusion criteria were not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Yes, blinding took place of key study personnel
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete outcome data reported
Selective reporting (reporting bias)	Low risk	Primary and secondary outcomes of interest were stated and these were addressed
Other bias	Low risk	Clear of other source of bias
Adequate sequence generation	Unclear risk	Not specified

Panpimanmas 2010.

Methods	Prospective randomised double‐blinded controlled trial.
Participants	Majority of patients were in the 25‐40 year age category. Thailand, university hospital setting and general hospital setting. Trials were carried out in three different hospital settings. Five‐hundred and seventy patients (299 females, 271 males) aged under fifty years were recruited who had suffered an acute symptomatic haemorrhoidal episode within the last five days.
Interventions	Patients were randomised into three groups; Group 1; C.quadrangularis, three tablets twice a day for four days then two tablets twice a day for three days. Group 2; MPFF group, similar regime as in group. Group 3; Placebo group; identical therapeutic regime.
Outcomes	Patients were evaluated based on bleeding, discharge, pain, pruritis, erythema and direct patient interviews. Each symptom was scored on a graded severity scale from 0 to 3 on the first and seventh day of treatment.
Notes	Exclusion criteria were previous haemorrhoidectomy, treatment with anticoagulation or ASA, taking other anti haemorrhoidal drugs at the time, permanent prolapsed internal haemorrhoids requiring surgery, patients suffering from moderate to severe hypertension, cardiovascular diseases, renal failure, cirrhosis, pregnancy and lactation. Discontinuation criteria were severe side effects and life‐threatening situations.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded provisions maintained
Incomplete outcome data (attrition bias) All outcomes	High risk	Plausible effect size among missing outcomes enough to induce clinically relevant bias in intervention effects estimates
Selective reporting (reporting bias)	Low risk	All of the study's pre‐specified outcomes that are of interest have been reported in a pre‐specified way
Other bias	Unclear risk	Other source of bias not specified
Adequate sequence generation	Unclear risk	Not specified

Sarabia 2001.

Methods	Randomized controlled study with an initial double‐blind period.
Participants	Age range was 18‐75 years. Male and female participants. Venezuala, university hospital setting. A total of fifty‐one patients with a diagnosis of acute haemorrhoidal crisis were randomised to receive either calcium dobesilate or a micronized flavonoid fraction of diosmin.
Interventions	Patients were randomised to either of the following groups; i) Calcium dobesilate 1 capsule (500 mg) TID for > 4 weeks, ii) Micronized flavonoid fraction of diosmin two tablets 500‐mg tablets TID for > 4 weeks. For the first ten days, patients in both groups were given dummy capsules or tablets to maintain the double‐blinded provision.
Outcomes	Assessment of haemorrhoidal symptoms on a four‐point scale (pain, discharge, bleeding, oedema, inflammation, pruritis and anal pressure). Use of pain medication, new acute episodes, patient perspective of efficacy, incidence of adverse events and compliance were assessed with direct patient interviews.
Notes	Exclusion criteria were previous rectal surgery, indication for surgery, chronic or serious disease, altered hepatic or renal function, pelvic tumour or portal hypertension.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete data outcome
Selective reporting (reporting bias)	Low risk	No selective outcome reporting
Other bias	Unclear risk	Other source of bias not specified
Adequate sequence generation	Unclear risk	Not specified

Squadrito 2000.

Methods	Double‐blinded, randomised parallel group controlled clinical trial
Participants	Average age for the control group and study group were 40.3 years and 40.6 years respectively. Male and female participants. Italy, university hospital setting. One hundred patients with acute uncomplicated haemorrhoids who were at least thirty years of age were enrolled provided their diagnosis was endoscopically confirmed.
Interventions	Patients were allocated into the following groups; Group 1; Troxerutin 150 mg and carbazochrome 1.5 mg twice daily by intramuscular route. Group 2; Saline containing carbazochrome 1.5mg to give the solution the same appearance as the combination product (control drug).
Outcomes	Subjective efficacy variable; (anal discomfort, spontaneous local bleeding, and pain at defecation). Objective efficacy variables (proctorrhagia, proctitis, anal prolapse). All evaluated by a four‐point scale 0=absent, 4=severe. Safety assessment.
Notes	Exclusion criteria were advanced haemorrhoids requiring surgery, type 1 diabetes, congestive heart failure treated with anti‐platelet drugs, presence of clinically significant renal, hepatic or other severe disease. Use of various medications such as fibrinolytic drugs, venous acting medications, investigational drugs within 30 days of screening, analgesic (other than 5% lidocaine ointment), NSAIDS within 7 days of screening or concomitant use of medications were not allowed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double blinded provisions maintained
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcome data reported
Selective reporting (reporting bias)	Low risk	Outcome measures were addressed in the study design and these were addressed
Other bias	Unclear risk	Other source of bias not specified
Adequate sequence generation	Low risk	Computer generated randomizations procedure

Thanapongsathorn 1992.

Methods	Double‐blinded randomised controlled study.
Participants	Average age of participants in placebo and Diosmin group were 32 years. Male and female. Thailand, university hospital setting. One‐hundred new and consecutive patients with acute symptoms due to first and second degree internal haemorrhoids were enrolled.
Interventions	Group 1; Oral Diosmin. Twelve tablets in three divided doses during the first four days and then two tablets twice daily for another ten days. Group 2; Placebo; identical therapeutic regimen.
Outcomes	Assessment of symptoms assessed in relation to subjective symptoms, which were; i) bleeding ii) pain iii) prolapse iv) burning v) pruritis vi) tenesmus vii) heaviness. Assessment of symptoms in relation to objective assessment which were i) swelling ii) congestion iii) bleeding iv) exudation v) prolapse.
Notes	Exclusion criteria were surgical anorectal diseases, previous internal haemorrhoid treatment, pregnancy, portal hypertension and haematological diseases.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded provisions maintained
Incomplete outcome data (attrition bias) All outcomes	Low risk	Attrition occurred in placebo group however missing outcomes not enough to have statistically significant impact on the observed effect size
Selective reporting (reporting bias)	Low risk	Primary and secondary outcome measures were stated and these were reported
Other bias	Unclear risk	Other source of bias not specified
Adequate sequence generation	Unclear risk	Not specified

Wijayanegara 1992.

Methods	Double‐blinded randomised placebo controlled study.
Participants	Pregnant women between 12 and 34 weeks gestation. Female participants, pregnant. Indonesia, antenatal clinic centres. A total of 100 patients were recruited to the trials based on certain inclusion criteria; i) severity of haemorrhoids (1st, 2nd and 3rd degree haemorrhoids) ii) Between weeks 12 and 34 gestation iii) a pregnancy with no medical or obstetric complications iv) No other treatment was given except for vitamins.
Interventions	Random allocation of patients into the following treatment groups; i) One tablet orally of hydroxyethyl‐rutoside taken orally twice daily for four weeks. ii) Corresponding placebo.
Outcomes	Global opinion of the patients involved. Overall evaluation of the efficacy of the clinicians.
Notes	Exclusion criteria included i) fourth degree haemorrhoids ii) All participants were informed not to use any other systemic or local treatment (except vitamins) including non‐prescription drugs during the trial period.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Inadequately described
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blinded and unlikely to be broken
Incomplete outcome data (attrition bias) All outcomes	Low risk	Incomplete outcome data adequately addressed
Selective reporting (reporting bias)	Low risk	All outcome measures of interest were reported in the study protocol and these were addressed
Other bias	Unclear risk	Other sources of bias not specified
Adequate sequence generation	Unclear risk	Not adequately described

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Garner 2002	Not a randomised controlled trial evaluating the effect of phlebotonics in treating haemorrhoids.
Kubchandani 2001	Conference proceedings describing randomised controlled trials.
Marsicano 1995	Poor methodology existed with inadequate description of sequence generation, blinding and allocation concealment. Baseline imbalance existed with crucial differences in characteristics between participants from both groups such as gender.
Mlakar 2008	Describes a quasi‐random method of randomizations in which patients are allocated based on the order of presentation.
Quijano 2005	Review evaluating various methods of conservative management of haemorrhoids in pregnancy.
Sandhu 2004	Describes a quasi‐random method of randomizations in which subjects were randomly allocated to their treatment of their choice after discussing the advantages and disadvantages of either method.
Smith 1998	Suppository preparations were not true phlebotonics upon closer evaluation of formulary but were predominantly antibiotic corticosteroid combinations.
Tan 2006	Substance evaluated was not true phlebotonic on closer evaluation.

Differences between protocol and review

We mainly addressed the outcome measures regarding symptom control, relief and relapse rates which were originally addressed in the protocol. We ensured that these were the main focus of our outcome measures. The symptoms of haemorrhoids were more specifically defined in the final review and we decided to separate them into primary and secondary categories based on their clinical importance. We defined the primary symptoms as bleeding, pain and pruritus and secondary outcomes were measures of discharge, overall symptom improvement and analgesic consumption.

The outcome measure regarding the adverse events reported just isolated events. The general consensus is that phlebotonics are a safe class of drug and due to the sparse nature of the adverse events, it was inappropriate to pool this outcome into a meta‐analysis.

We originally predefined the sub‐groups according to the following categories: 'thrombosed haemorrhoids, pregnant women, methodological quality and use of a validated questionnaire'. However, none of the studies reported using phlebotonics in thrombosed haemorrhoids, so this category could not be placed as a subgroup. The majority of the studies were also of a similar methodological quality and so a sub‐group analysis could not be performed in this case. In many cases there were also too few trials per outcome measure to perform a subgroup meta‐analysis. We did however included the 'pregnant women' into a subgroup meta‐analysis.

Contributions of authors

Dr Nirmal Perera (NP) and Dr Danae Liolitsa (DL) were involved in the data search, exclusion of studies and data extraction. NP and Dr Anna Croxford (AC) were involved independently in completing a systematic critical review in included studies. Dr Peter Lang (PL) and Dr Obioha Ukaegbu (OU) were involved in the grey literature search as well as the attempt to contact authors of studies where statistical data was insufficient. Mr Muhammed Iqbal Bhatti (MIB), Mr Mohammed Yassin (MY) and Mr Satheesh Iype (SI) were senior general surgeons involved as content experts.

Sources of support

Internal sources

• Ipswich Hospital Library, NHS, Not specified.

External sources

• Kings Lynn Hospital Library NHS trust, Not specified.

Declarations of interest

No conflicts of interest are present and no declarations of interest are present.

New