Barrier agents for adhesion prevention after gynaecological surgery

Gaity Ahmad; Kyungmin Kim; Matthew Thompson; Priya Agarwal; Helena O'Flynn; Akshay Hindocha; Andrew Watson

doi:10.1002/14651858.CD000475.pub4

. 2020 Mar 22;2020(3):CD000475. doi: 10.1002/14651858.CD000475.pub4

Barrier agents for adhesion prevention after gynaecological surgery

Gaity Ahmad ^1,^✉, Kyungmin Kim ², Matthew Thompson ², Priya Agarwal ², Helena O'Flynn ¹, Akshay Hindocha ¹, Andrew Watson ³

Editor: Cochrane Gynaecology and Fertility Group

PMCID: PMC7085418 PMID: 32199406

Abstract

Background

Pelvic adhesions can form secondary to inflammation, endometriosis, or surgical trauma. Strategies to reduce pelvic adhesion formation include placing barrier agents such as oxidised regenerated cellulose, polytetrafluoroethylene, and fibrin or collagen sheets between pelvic structures.

Objectives

To evaluate the effects of barrier agents used during pelvic surgery on rates of pain, live birth, and postoperative adhesions in women of reproductive age.

Search methods

We searched the following databases in August 2019: the Cochrane Gynaecology and Fertility (CGF) Specialised Register of Controlled Trials, MEDLINE, Embase, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycINFO, the Cochrane Central Register of Controlled Trials (CENTRAL), Epistemonikos, and trial registries. We searched reference lists of relevant papers, conference proceedings, and grey literature sources. We contacted pharmaceutical companies for information and handsearched relevant journals and conference abstracts.

Selection criteria

Randomised controlled trials (RCTs) on the use of barrier agents compared with other barrier agents, placebo, or no treatment for prevention of adhesions in women undergoing gynaecological surgery.

Data collection and analysis

Three review authors independently assessed trials for eligibility and risk of bias and extracted data. We calculated odds ratios (ORs) or mean differences (MDs) with 95% confidence intervals (CIs) using a fixed‐effect model. We assessed the overall quality of the evidence using GRADE (Grades of Recommendation, Assessment, Development and Evaluation) methods.

Main results

We included 19 RCTs (1316 women). Seven RCTs randomised women; the remainder randomised pelvic organs. Laparoscopy (eight RCTs) and laparotomy (11 RCTs) were the primary surgical techniques. Indications for surgery included myomectomy (seven RCTs), ovarian surgery (five RCTs), pelvic adhesions (five RCTs), endometriosis (one RCT), and mixed gynaecological surgery (one RCT). The sole indication for surgery in three of the RCTs was infertility. Thirteen RCTs reported commercial funding; the rest did not state their source of funding.

No studies reported our primary outcomes of pelvic pain and live birth rate.

Oxidised regenerated cellulose versus no treatment at laparoscopy or laparotomy (13 RCTs)

At second‐look laparoscopy, we are uncertain whether oxidised regenerated cellulose at laparoscopy reduced the incidence of de novo adhesions (OR 0.50, 95% CI 0.30 to 0.83, 3 RCTs, 360 participants; I² = 75%; very low‐quality evidence) or of re‐formed adhesions (OR 0.17, 95% CI 0.07 to 0.41, 3 RCTs, 100 participants; I² = 36%; very low‐quality evidence).

At second‐look laparoscopy, we are uncertain whether oxidised regenerated cellulose affected the incidence of de novo adhesions after laparotomy (OR 0.72, 95% CI 0.42 to 1.25, 1 RCT, 271 participants; very low‐quality evidence). However, the incidence of re‐formed adhesions may have been reduced in the intervention group (OR 0.38, 95% CI 0.27 to 0.55, 6 RCTs, 554 participants; I² = 41%; low‐quality evidence).

No studies reported results on pelvic pain, live birth rate, adhesion score, or clinical pregnancy rate.

Expanded polytetrafluoroethylene versus oxidised regenerated cellulose at gynaecological surgery (two RCTs)

We are uncertain whether expanded polytetrafluoroethylene reduced the incidence of de novo adhesions at second‐look laparoscopy (OR 0.93, 95% CI 0.26 to 3.41, 38 participants; very low‐quality evidence). We are also uncertain whether expanded polytetrafluoroethylene resulted in a lower adhesion score (out of 11) (MD ‐3.79, 95% CI ‐5.12 to ‐2.46, 62 participants; very low‐quality evidence) or a lower risk of re‐formed adhesions (OR 0.13, 95% CI 0.02 to 0.80, 23 participants; very low‐quality evidence) when compared with oxidised regenerated cellulose.

No studies reported results regarding pelvic pain, live birth rate, or clinical pregnancy rate.

Collagen membrane with polyethylene glycol and glycerol versus no treatment at gynaecological surgery (one RCT)

Evidence suggests that collagen membrane with polyethylene glycol and glycerol may reduce the incidence of adhesions at second‐look laparoscopy (OR 0.04, 95% CI 0.00 to 0.77, 47 participants; low‐quality evidence). We are uncertain whether collagen membrane with polyethylene glycol and glycerol improved clinical pregnancy rate (OR 5.69, 95% CI 1.38 to 23.48, 39 participants; very low‐quality evidence).

One study reported adhesion scores but reported them as median scores rather than mean scores (median score 0.8 in the treatment group vs median score 1.2 in the control group) and therefore could not be included in the meta‐analysis. The reported P value was 0.230, and no evidence suggests a difference between treatment and control groups.

No studies reported results regarding pelvic pain or live birth rate.

In total, 15 of the 19 RCTs included in this review reported adverse events. No events directly attributed to adhesion agents were reported.

Authors' conclusions

We found no evidence on the effects of barrier agents used during pelvic surgery on pelvic pain or live birth rate in women of reproductive age because no trial reported these outcomes.

It is difficult to draw credible conclusions due to lack of evidence and the low quality of included studies. Given this caveat, low‐quality evidence suggests that collagen membrane with polyethylene glycol plus glycerol may be more effective than no treatment in reducing the incidence of adhesion formation following pelvic surgery. Low‐quality evidence also shows that oxidised regenerated cellulose may reduce the incidence of re‐formation of adhesions when compared with no treatment at laparotomy. It is not possible to draw conclusions on the relative effectiveness of these interventions due to lack of evidence.

No adverse events directly attributed to the adhesion agents were reported. The quality of the evidence ranged from very low to moderate. Common limitations were imprecision and poor reporting of study methods. Most studies were commercially funded, and publication bias could not be ruled out.

Plain language summary

Barrier agents for adhesion prevention after gynaecological surgery

Review question

This review of trials assessed the effects of barrier agents on pelvic pain, live birth rate, clinical pregnancy rate, adhesion formation, and adhesion score (a measure of adhesion severity) after pelvic surgery.

Background

A common problem following pelvic surgery is the occurrence of adhesions, where the surfaces of two separate pelvic structures (e.g. inner lining of pelvic wall or pelvic organs such as uterus, ovaries, bladder, or bowel) stick together. During pelvic surgery, strategies to reduce pelvic adhesion formation include placing a synthetic physical barrier between pelvic structures.

Study characteristics

We included 19 randomised controlled trials (RCTs) that included a total of 1316 women undergoing gynaecological surgery. These trials assessed different types of barrier agents for preventing adhesions and compared them with each other or with no treatment. The data are current to August 2019. Thirteen RCTs reported commercial funding; the other studies did not state their source of funding.

Key results

No studies reported the effects of barrier agents used during pelvic surgery on pelvic pain or live birth rate among women of reproductive age.

Low‐quality evidence suggests that oxidised regenerated cellulose and collagen membrane with polyethylene glycol plus glycerol may be more effective than no treatment in reducing the risk of adhesion formation following pelvic surgery.

One study reported the effect of collagen membrane with polyethylene glycol plus glycerol on postoperative adhesion score; however due to the way these data were reported, we are unable to interpret whether the intervention had any effect. No studies reported the effect of oxidised regenerated cellulose on adhesion score.

One study reported the effect of collagen membrane with polyethylene glycol plus glycerol on clinical pregnancy rate; however this evidence was found to be of very low quality. We are uncertain whether this intervention led to a higher clinical pregnancy rate than no treatment. No studies reported the effect of any other intervention on clinical pregnancy rate.

Two studies compared the effects of expanded polytetrafluoroethylene and oxidised regenerated cellulose on adhesion score and adhesion formation. However, this evidence was found to be of very low quality, and we are uncertain whether either intervention was more effective than the other. No studies compared the relative effects of these interventions on pelvic pain, live birth rate, or clinical pregnancy rate.

We found no conclusive evidence on the relative effectiveness of any reported interventions. No adverse events directly attributed to the adhesion agents were reported.

Quality of the evidence

The quality of the evidence ranged from very low to moderate. The most common limitations were imprecision (few participants and wide confidence intervals) and poor reporting of study methods. Most studies were commercially funded, and publication bias could not be ruled out.

Summary of findings

Background

Description of the condition

Pelvic adhesions can form as the result of pelvic inflammation, endometriosis, or surgical trauma. The incidence of pelvic adhesions at second‐look laparoscopy in the first few weeks after surgery has been reported to be between 25% and 92% (Okabayashi 2014). Consequences of adhesion formation include subfertility, development of chronic abdominal pain, and dyspareunia (difficult or painful sexual intercourse) (SRS 2007). A recent study demonstrated that in women with a known reason for small‐bowel obstruction, adhesions were the single most common cause (Ten Broek 2013).

Cutting, surgical denudation, ischaemia, desiccation, or abrasion can cause peritoneal trauma during surgery. Subsequent healing in the peritoneal cavity occurs through a combination of mesothelial regeneration and fibrosis, resulting in adhesion formation between damaged serosal surfaces (diZerega 1990).

Minimally invasive techniques such as laparoscopy reduce the risk of de novo (new) adhesion formation but do not eliminate it entirely. Studies included within this review therefore assessed both laparoscopy and laparotomy.

Description of the intervention

Several barrier agents with different characteristics are commercially available. Oxidised regenerated cellulose was the first tested synthetic mechanical barrier agent to cover traumatised peritoneum in the pelvis. It is applied over raw tissue surfaces at the last stage of surgery after haemostasis has been achieved, and it is designed to form a gelatinous protective coat within eight hours of application. Following this, it is broken down into its monosaccharide constituents and is designed to be absorbed within two weeks. Concerns with use of oxidised regenerated cellulose include migration and the need for meticulous haemostasis. Use of oxidised regenerated cellulose in the presence of bleeding may promote fibrin deposition at sites of incomplete haemostasis, resulting in adhesion formation rather than prevention (Wiseman 1999).

Another commercially available barrier agent is expanded polytetrafluoroethylene surgical membrane. This inert and permanent barrier acts by preventing cellular growth. However, it must be sutured to remain in place, and this may increase the incidence of adhesions while prolonging operating time. Much debate is ongoing regarding the need to remove expanded polytetrafluoroethylene after peritoneal healing is complete. This debate stems from studies investigating expanded polytetrafluoroethylene in vascular and pericardial grafts, which found no significant long‐term adverse effects when the barrier was not removed (Jacobs 1996). In addition, a prospective multi‐centre observational study investigating the long‐term use of expanded polytetrafluoroethylene without removal reported only one case of postoperative infection, which did not require removal of the membrane (Hurst 1998).

Other products include sodium hyaluronate with carboxymethylcellulose, an adhesion barrier agent composed of chemically derived sodium hyaluronate and carboxymethylcellulose. It is designed to be absorbed from the peritoneal cavity within seven days and completely excreted from the body within 28 days (Diamond 1996). Sodium hyaluronate with carboxymethylcellulose consists of chemically modified hyaluronic acid and carboxymethylcellulose. It separates denuded planes of tissue for up to seven days before it is absorbed. As with oxidised regenerated cellulose, migration is the cause of some concern; however no evidence suggests that its effectiveness is altered by the presence of blood. Although evidence for its use in gynaecological surgery is limited, sodium hyaluronate with carboxymethylcellulose has been widely studied in the context of colorectal surgery. Formation of adhesions and incidence of small‐bowel obstruction were reduced in five randomised controlled trials (RCTs) investigating its use in various colorectal and general surgical procedures (Ten Broek 2013).

Another product is a dual‐sided hydrophilic film. One side is a smooth surface consisting of porcine collagen, polyethylene glycol, and glycerol, and the opposite side is a porous surface made of lyophilised porcine collagen. The film is designed to degrade in the body within three weeks of application, and it has demonstrated significant reduction in adhesion formation in rat models (Gruber‐Blum 2011).

Fibrin sheet is a sheet‐type fibrin sealant with a solid layer of human fibrinogen, thrombin, and aprotinin coating the active surface of equine collagen stained with riboflavin (Mais 1995a; Pellicno 2003).

Fluid and pharmacological methods used to prevent adhesion formation are investigated in another Cochrane Review (Ahmad 2014).

How the intervention might work

Theoretically, inert physical materials that are able to prevent mechanical contact between serosal surfaces for longer than three days have the potential to be helpful in preventing adhesion formation. This would allow independent healing of each traumatised peritoneal surface. Some barrier agents (e.g. expanded polytetrafluoroethylene) need to be sutured into place, requiring extra operating time, especially during laparoscopic procedures.

Why it is important to do this review

This review is concerned with the effectiveness of barrier agents placed in the peritoneal cavity for preventing adhesions and in some cases for improving fertility. Pharmacological adjuncts are reviewed elsewhere. Women present with the secondary effects of adhesions, including dyspareunia, subfertility, bowel obstruction, and chronic pelvic pain. These problems can greatly impact quality of life and may necessitate further surgery. No clinical consensus or guidance is available regarding the most effective anti‐adhesion agent. Assessment of the evidence on effectiveness of barrier agents is therefore important.

Objectives

To evaluate the effects of barrier agents used during pelvic surgery on rates of pain, live birth, and postoperative adhesions in women of reproductive age.

Methods

Criteria for considering studies for this review

Types of studies

We included published or unpublished randomised controlled trials (RCTs) in which either women or pelvic structures were the unit of randomisation. Non‐randomised studies (e.g. studies with evidence of inadequate sequence generation such as alternate days or participant numbers) were excluded, as they are associated with high risk of bias. In the case of cross‐over trials, only data from before the cross‐over would have been included, as any additional pelvic surgery increases the risk of adhesion formation. However, no cross‐over trials were identified.

Types of participants

Women undergoing pelvic surgery for infertility or for other indications. Studies investigating adhesion prevention in non‐gynaecological specialties were excluded. Types of surgery performed could include open or laparoscopic procedures.

Types of interventions

Trials comparing physical barrier agents (oxidised regenerated cellulose, expanded polytetrafluoroethylene, sodium hyaluronate with carboxymethylcellulose, fibrin sheet, collagen membrane with polyethylene glycol and glycerol) used during pelvic surgery versus any other physical barrier agent or placebo or no treatment were included.

Studies of fibrin glue and Sepracoat (Genzyme Corporation) were excluded, as these are not physical barrier agents.

Types of outcome measures

Primary outcomes

Pelvic pain (improvement/worsening/no change in pain at second‐look laparoscopy (SLL)), as measured by validated pain scales, for example, visual analogue pain scale (VAS) scores, the McGill Pain Questionnaire (MPQ), a pain improvement rating scale, general pain experience, or a gynaecological pain questionnaire
Live birth rate

Secondary outcomes

Adhesion score, recorded on whichever scale the study authors used but with preference given to the modified American Fertility Society (mAFS) score
Number of participants with adhesions at SLL
Clinical pregnancy rate (pregnancy confirmed on ultrasound scan)
Miscarriage rate, defined as loss of pregnancy before 24 weeks of gestation
Ectopic pregnancy rate
Number of participants with improvement in quality of life (QoL) at SLL, recorded on whichever scale was chosen by study authors
Adverse events

Live birth rate and clinical pregnancy rate are relevant when studies have specifically investigated use of the barrier agent in procedures performed to improve fertility. This does not apply to some studies.

Articles that met the inclusion criteria but did not report any of the outcomes considered within this review were also included within the qualitative analysis.

Search methods for identification of studies

We searched for all published and unpublished RCTs comparing the use of barrier agents versus any other active intervention or placebo/no treatment without language restriction. Searches were designed and conducted by the Information Specialist for Cochrane Gynaecology and Fertility.

Electronic searches

We searched:

Cochrane Gynaecology and Fertility Group (CGFG) Specialised Register of Controlled Trials; Procite platform, searched 21 August 2019 (Appendix 1);
Central Register of Controlled Trials (CENTRAL); OVID platform, searched 21 August 2019 (Appendix 2);
MEDLINE; OVID platform, searched from 1946 to 21 August 2019 (Appendix 3);
Embase; OVID platform, searched from 1980 to 21 August 2019 (Appendix 4); and
PsycINFO; OVID platform, searched from 1806 to 21 August 2019 (Appendix 5).

These searches were conducted by the Information Specialist for the Cochrane Gynaecology and Fertility Group.

Other electronic searches included trial registers for the US National Institutes of Health, the World Health Organization (WHO) international trial registry platform, the Database of Abstracts of Reviews of Effects (DARE), OpenGrey, PubMed, Epistemonikos, and Google Scholar.

We conducted the last search in August 2019.

Searching other resources

We also searched the reference lists of relevant publications, review articles, and included studies and contacted experts in the field to request additional data. We handsearched relevant journals and conference abstracts that were not covered in the CGF register, in liaison with the Information Specialist. Two review authors screened retrievals.

Data collection and analysis

Selection of studies

After an initial screen of titles and abstracts identified by the search, we retrieved the full text of all potentially eligible studies. Three review authors (PA, KK, MT) independently examined these full‐text articles for compliance with inclusion criteria. Disagreements on study eligibility were resolved by consultation with a fourth review author (GA). Review authors corresponded with study investigators to clarify study eligibility (e.g. with respect to participant eligibility criteria and allocation method). Figure 1 shows a flow diagram of study selection.

Data extraction and management

Three review authors (PA, KK, MT) independently extracted data from eligible studies using a data extraction form that was designed and pilot‐tested by the review authors. Disagreements were resolved by consultation with a fourth review author (GA). Data extracted included study characteristics and outcome data. When studies were followed by multiple publications, the main trial report was used as the reference, and additional details were derived from secondary papers. We corresponded with study investigators to ask for further data on methods and/or results, as required.

Assessment of risk of bias in included studies

We assessed each included trial for the following criteria using the Cochrane risk of bias assessment tool: sequence generation; allocation concealment; blinding of participants, personnel, and outcome assessors; completeness of outcome data; selective outcome reporting; and other potential sources of bias (Higgins 2011). We presented conclusions in the 'Risk of bias' table and incorporated them into the interpretation of review findings by performing sensitivity analyses. Two review authors (KK, MT) independently performed all assessments of the quality of clinical trials. All discrepancies were resolved by GA and PA.

Care was taken to search for within‐study reporting bias, as seen in trials failing to report obvious outcomes (e.g. pregnancy rate, major complications) or reporting them in insufficient detail.

Measures of treatment effect

Results of dichotomous variables are presented as Mantel‐Haenszel odds ratios (ORs) with 95% confidence intervals (CIs), and results of continuous variables are presented as mean differences (MDs).

We reversed the direction of effects of individual studies, when required, to ensure consistency across trials. We treated ordinal data as continuous outcomes. We presented 95% CIs for all outcomes. When data used to calculate ORs or MDs were not available, we planned to utilise the most detailed numerical data available, which might facilitate similar analyses of included studies. We compared the magnitude and direction of effects reported by studies versus how they are presented in the review, taking account of legitimate differences.

Unit of analysis issues

For within‐participant designs, 'effective sample sizes' were calculated to allow for statistical synthesis, that is, for a trial randomly assigning ovaries within each participant rather than randomly assigning participants, numbers were calculated to simulate as nearly as possible the odds ratio and the confidence interval as if the study design had randomly assigned participants, not ovaries. This was achieved as follows.

We calculated the odds ratio and the 95% CI for the matched design.
We constrained the control rate in the hypothetical parallel design to make it equal to that observed in the matched design.
We constrained group sizes for the parallel design to make them equal to each other.
Based on the two constraints above, we calculated the numbers of 'successes' and 'failures' in each group to reproduce as nearly as possible the OR (95% CI) of the matched design.

As a result of this type of re‐analysis of data, studies with within‐participant design that may have previously demonstrated significance may no longer do so.    Several studies of within‐participant design comparing oxidised regenerated cellulose versus no treatment were presented and analysed wrongly as having a parallel design. We recognised that failure to account for pairing and failure to take account of doubling the sample size by using ovaries instead of participants as the unit of randomisation may have yielded spurious results. For these studies, the least favourable outcome for oxidised regenerated cellulose that was compatible with the reported results was assumed. We attempted to obtain correct results tables from study authors but were unsuccessful.

Dealing with missing data

Data were analysed on an intention‐to‐treat (ITT) basis as far as possible, and attempts were made to obtain missing data from the original investigators. When data for primary outcomes could not be obtained, we planned to undertake imputation of individual values. For secondary outcomes, only available data were analysed.    If studies reported sufficient detail to calculate mean differences but no information on associated standard deviation (SD), we assumed that the outcome had a standard deviation equal to the highest SD from other studies within the same analysis.

Assessment of heterogeneity

Clinical and methodological characteristics of individual studies were considered to ensure that any pooling was clinically meaningful. The I² statistic was calculated to assess statistical heterogeneity. An I² measurement over 50% was taken to indicate substantial heterogeneity.

Assessment of reporting biases

In view of the difficulties involved in detecting and correcting publication bias and other reporting biases, study authors aimed to minimise their potential impact by ensuring a comprehensive search for eligible studies, and by remaining alert for duplication of data. We aimed to minimise the impact by ensuring that a robust and comprehensive search was performed. We planned to create a funnel plot of 10 or more studies that were included in a meta‐analysis to explore the possibility of small‐study effects (a tendency for estimates of the intervention effect to suggest that it is more beneficial in smaller studies).

Data synthesis

Statistical analysis was performed in accordance with the guidelines developed by Cochrane. Data from the primary studies were combined in RevMan using the fixed‐effect model. We planned to report standardised mean differences (SMDs) if similar outcomes were reported on different scales. An increase in OR, SMD, or MD was indicated to the right of the central line of the forest plot, and a decrease was indicated to the left of the central line. Whether this favoured treatment or no treatment depended on the outcome analysed, but the axes were labelled accordingly. Analyses were stratified by type of barrier agent, type of surgery, and type of control.

Subgroup analysis and investigation of heterogeneity

If we had detected substantial heterogeneity, we planned to explore possible explanations via sensitivity analyses. We planned to take any statistical heterogeneity into account when interpreting the results, especially if any variation in the direction of effect was evident.

Sensitivity analysis

We conducted sensitivity analyses for the important review outcomes to determine whether the conclusions were robust to arbitrary decisions made regarding eligibility of studies and analysis. These analyses included consideration of whether review conclusions differed if:

eligibility was restricted to studies without high risk of bias;
a random‐effects model had been adopted; or
relative risk had been used as the summary effects measure.

Overall quality of the body of evidence: 'Summary of findings' table

We prepared 'Summary of findings' tables using GRADEpro software and Cochrane methods (GRADEpro GDT 2015; Higgins 2011). These tables evaluate the overall quality of the body of evidence for the main review outcomes (pelvic pain, live birth rate, clinical pregnancy rate, improvement or worsening of adhesion scores, and incidence of adhesions) using GRADE (Grades of Recommendation, Assessment, Development and Evaluation) criteria (study limitations (i.e. risk of bias), consistency of effect, imprecision, indirectness, and publication bias). Judgements about evidence quality (high, moderate, or low) have been justified, documented, and incorporated into reporting of results for each outcome.

Results

Description of studies

Results of the search

At the 2020 update three studies were retrieved in full text. One new study was identified and was included within the meta‐analysis in this update (Canis 2014), and two were excluded; there are now 19 included studies and 10 excluded. See Figure 1 for an overview of search results.

At the 2015 update eighteen studies met inclusion criteria, and eight studies were excluded. See study tables under Characteristics of included studies. Reasons why studies were excluded are detailed in the Characteristics of excluded studies tables.

Included studies

Trial design and setting

Number of studies

We identified 19 RCTs that met the inclusion criteria (1309 participants).

Multi‐centre trials

Ten were multi‐centre studies (Azziz 1993; Canis 2014; Diamond 1996; Franklin 1995; Haney 1995; Myomectomy ASG 1995; Nordic APSG 1995; Sekiba 1992; Takeuchi 2005; Tinelli 2011).

Design

In seven parallel‐group trials, the unit of randomisation was the participant (Canis 2014; Diamond 1996; Mais 1995a; Mais 1995b; Takeuchi 2005; Tinelli 2011 Wallweiner 1998). The remainder were within‐participant trials in which the unit of randomisation was the ovary, uterine sites, or the pelvic sidewall. One study was a multi‐arm trial comparing fibrin gel and a fibrin sheet versus control (Takeuchi 2005). Comparison of the fibrin sheet arm versus the control arm is included in this review.

Support/sponsorship

Thirteen trials stated sponsorship. Seven trials were sponsored by Johnson & Johnson (Azziz 1993; Franklin 1995; Li 1994; Nordic APSG 1995; Saravelos 1996; Sekiba 1992; van Geldorp 1994), one was sponsored by Genzyme (Diamond 1996), one by Covidien (Canis 2014), and a further two trials by manufacturers of expanded polytetrafluoroethylene (Haney 1995; Myomectomy ASG 1995). One trial received additional sponsorship from the Medical Research Council (MRC) in Canada (Greenblatt 1993). Another trial received support from Johnson & Johnson after the study was completed to help with the analysis, but the final content of the paper was under the sole control of the principal investigator (Keckstein 1996).

Participants

Primary indications for surgery

Myomectomy was the indication in six trials (Canis 2014; Diamond 1996; Mais 1995b; Myomectomy ASG 1995; Takeuchi 2005; Tinelli 2011), and ovarian surgery was the indication in four trials (Greenblatt 1993; Keckstein 1996; Saravelos 1996; van Geldorp 1994). Three trials were restricted to women undergoing adhesiolysis for infertility (Azziz 1993; Nordic APSG 1995; Sekiba 1992). Participants also underwent adhesiolysis in Li 1994, although the indications were subgrouped into infertility or pelvic pain. The indication for the Haney 1995 trial was also pelvic adhesions, although the primary reason for surgery was not exclusively infertility, and this trial was not subgrouped. The indications were endometriosis in Mais 1995a and Wallweiner 1998, and bilateral ovarian disease in Franklin 1995.

Type of surgery

Laparotomy was performed in nine trials (Azziz 1993; Canis 2014; Diamond 1996; Franklin 1995; Li 1994; Myomectomy ASG 1995; Nordic APSG 1995; Sekiba 1992; van Geldorp 1994), and laparoscopy was performed in the remainder (Greenblatt 1993; Haney 1995; Keckstein 1996; Korell 1994; Mais 1995a; Mais 1995b; Saravelos 1996; Takeuchi 2005; Wallweiner 1998). Tinelli 2011 performed both laparoscopy and laparotomy.

Suturing

Oxidised regenerated cellulose was occasionally sutured into place (Keckstein 1996; Li 1994; Sekiba 1992), whereas expanded polytetrafluoroethylene was always sutured into place (Haney 1995; Korell 1994; Myomectomy ASG 1995).

Adjuvants to surgery

One trial compared oxidised regenerated cellulose versus no treatment; hydrocortisone was also instilled in both treatment and control groups (Li 1994).

Microsurgical techniques

In three trials, it was noted that microsurgery was performed (Azziz 1993; Li 1994; Nordic APSG 1995).

Interventions

Thirteen RCTs compared oxidised regenerated cellulose with no treatment ‐ six at laparoscopy (Keckstein 1996; Mais 1995a; Mais 1995b; Saravelos 1996; Tinelli 2011; Wallweiner 1998), and seven at laparotomy (Azziz 1993; Franklin 1995; Li 1994; Nordic APSG 1995; Sekiba 1992; Tinelli 2011; van Geldorp 1994)
Two RCTs compared expanded polytetrafluoroethylene versus oxidised regenerated cellulose (Haney 1995; Korell 1994)
One RCT compared expanded polytetrafluoroethylene versus no treatment (Myomectomy ASG 1995)
One RCT compared sodium hyaluronate with carboxymethylcellulose versus no treatment (Diamond 1996)
One RCT compared fibrin sheet versus no treatment (Takeuchi 2005)
One RCT compared collagen film with polyethylene glycol and glycerol versus Ringer's lactate (Canis 2014)

Timing of second‐look laparoscopy (SLL)

Only two studies stated a mean time after laparoscopy (Diamond 1996 ‐ 23 days; Haney 1995 ‐ 30 days). Large variation was noted in timing of the SLL both within and between studies. Only three studies performed all SLL procedures within six weeks (Greenblatt 1993; Haney 1995; Myomectomy ASG 1995). Only four further studies performed SLL after eight weeks (Keckstein 1996; Mais 1995a; Mais 1995b; Wallweiner 1998). Remaining studies performed SLL after 10 days to 20 weeks.

Tinelli 2011 did not routinely perform SLL but instructed participants to have any further surgery completed within a six‐year period. Adhesions were assessed at the second surgery. Timing of the second surgery varied between 2.3 and 2.5 years.

Studied outcomes

Primary outcomes

Pelvic pain

No trials reported this outcome.

Live birth rate

No trials reported this outcome.

Secondary outcomes

Adhesion score

Only one study reported mean adhesion score as an outcome (Haney 1995). Another study reported median adhesion score as an outcome (Canis 2014).

Number of participants with adhesions at SLL

Nine trials reported the incidence of de novo adhesions as an outcome (Canis 2014; Diamond 1996; Greenblatt 1993; Korell 1994; Mais 1995b; Myomectomy ASG 1995; Saravelos 1996; Takeuchi 2005; Tinelli 2011). Ten trials reported the incidence of re‐formation adhesions as an outcome (Azziz 1993; Franklin 1995; Haney 1995; Keckstein 1996; Li 1994; Mais 1995a; Nordic APSG 1995; Sekiba 1992; van Geldorp 1994; Wallweiner 1998).

Clinical pregnancy rate

One study reported this outcome (Canis 2014).

Miscarriage rate

No trials reported this outcome.

Ectopic pregnancy rate

One study reported this outcome (Canis 2014).

Quality of life

No trials reported this outcome.

Adverse effects

Fourteen studies reported adverse effects as an outcome (Azziz 1993; Canis 2014; Diamond 1996; Franklin 1995; Haney 1995; Keckstein 1996; Mais 1995a; Mais 1995b; Myomectomy ASG 1995; Nordic APSG 1995; Saravelos 1996; Takeuchi 2005; Tinelli 2011; van Geldorp 1994). In the remaining studies, the absence of adverse effects was not stated.

Excluded studies

Ten studies were excluded. In nine studies, the agent used as the intervention was not a barrier agent, and one study did not investigate use of a barrier agent during a gynaecological procedure (To 1992).

Risk of bias in included studies

See Figure 2 and Figure 3 for a summary diagram.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

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

Allocation

Sequence generation

Ten studies were rated as having unclear risk of bias. Study authors did not detail the randomly assigned component despite stating that these were randomised studies (Azziz 1993; Franklin 1995; Greenblatt 1993; Korell 1994; Li 1994; Nordic APSG 1995; Saravelos 1996; Sekiba 1992; van Geldorp 1994; Wallweiner 1998). Nine studies were deemed to have low risk of bias for sequence generation (Canis 2014; Diamond 1996; Haney 1995; Keckstein 1996; Mais 1995a; Mais 1995b; Myomectomy ASG 1995; Takeuchi 2005; Tinelli 2011).

Concealment of allocation

Eight studies were deemed to have low risk of selection bias (Canis 2014; Diamond 1996; Haney 1995; Keckstein 1996; Li 1994; Mais 1995a; Mais 1995b; Myomectomy ASG 1995). The remaining 11 studies were deemed to have unclear risk of selection bias. Although five of these studies used sealed envelopes, investigators did not detail the methods used to develop and monitor the allocation process with sealed envelopes (Azziz 1993; Franklin 1995; Nordic APSG 1995; Saravelos 1996; Sekiba 1992).

Blinding

Blinding of participants and personnel (performance bias)

Due to the nature of the intervention and the lack of comparable placebo agents, no study achieved blinding of the primary surgeon. However, by revealing the treatment allocation only after completing the initial surgical intervention, seven studies were deemed to have minimised performance bias (Azziz 1993; Canis 2014; Diamond 1996; Franklin 1995; Haney 1995; Keckstein 1996; Li 1994).

Twelve studies were rated to have unclear risk of performance bias (Greenblatt 1993; Korell 1994; Mais 1995a; Mais 1995b; Myomectomy ASG 1995; Nordic APSG 1995; Saravelos 1996; Sekiba 1992; Takeuchi 2005; Tinelli 2011; van Geldorp 1994; Wallweiner 1998), as study authors did not clearly state whether participants were blinded and/or whether treatment allocation was revealed before or during the initial surgery.

Blinding of outcome assessment (detection bias)

Nine studies clearly stated that assessors at SLL were blinded (Canis 2014; Diamond 1996; Greenblatt 1993; Li 1994; Mais 1995a; Mais 1995b; Saravelos 1996; Sekiba 1992; Tinelli 2011). Two of these studies recorded the SLL in video, and blinded independent assessors watched those videos to grade the severity of adhesions (Canis 2014; Diamond 1996). However, in Canis 2014, video recording was not performed in seven of the SLL patients, meaning that these patients were not assessed by a blinded independent surgeon, potentially increasing the risk of detection bias. Therefore, we included only the data from independently performed SLL from Canis 2014 in our final data analysis.

Eight studies did not clearly state that the surgeon performing the SLL was blinded to the initial intervention (Franklin 1995; Keckstein 1996; Korell 1994; Myomectomy ASG 1995; Nordic APSG 1995; Takeuchi 2005; van Geldorp 1994; Wallweiner 1998).

Two studies were deemed to have high risk of detection bias, as the surgeons at SLL were not blinded to the intervention group (Azziz 1993; Haney 1995).

Fourteen studies blinded surgeons to randomisation at second‐look laparoscopy, three studies did not discuss blinding (Haney 1995; Myomectomy ASG 1995; van Geldorp 1994), and two studies did not use blinding (Azziz 1993; Takeuchi 2005).

Incomplete outcome data

Four studies were rated as high risk, 10 as unclear risk, and five as low risk.

Dropouts

Azziz 1993 reported a large number of dropouts. A total of 198 participants were randomly assigned, but 64 were excluded from analysis because of inadequate documentation in 23, surgical technique and evaluation inconsistent with the protocol in 36, concurrent therapy in conflict with the protocol in three, and participant refusal to undergo second‐look laparoscopy in two (Wiseman 1999), demonstrating high risk of bias. Keckstein 1996 also reported a large number of dropouts, with 8 of 25 participants not returning for SLL. No reason was given for this. Canis 2014 reported a large number of dropouts as well, with 7 of 61 patients not undergoing second‐look laparoscopy and 21 of 61 not returning for three‐year follow‐up.

Tinelli 2011 reported participants undergoing a second surgical procedure within six years of initial surgery: 546 of 694 participants underwent a second surgery. In contrast to other trials, a second‐look laparoscopy at a specified time interval was not performed on all participants, leading to the exclusion of 148 participants from the results; this increases the risk of attrition bias.

Withdrawals and intention‐to‐treat analysis

Withdrawals were stated in eleven studies (Azziz 1993; Canis 2014; Diamond 1996; Greenblatt 1993; Haney 1995; Keckstein 1996; Li 1994; Myomectomy ASG 1995; Saravelos 1996; Takeuchi 2005; Tinelli 2011), and three studies did not state withdrawals (Franklin 1995; Nordic APSG 1995; Sekiba 1992); in five studies, no dropouts were reported (Mais 1995a; Mais 1995b; Takeuchi 2005; van Geldorp 1994; Wallweiner 1998). No studies with dropouts performed an ITT analysis. A total of six participants from all studies were reported to have been unable to undergo second‐look laparoscopy because of pregnancy.

Selective reporting

Two studies were rated as high risk (Li 1994; Nordic APSG 1995), and the other 17 studies were rated as unclear risk. The Nordic study reported larger numbers of participants randomly assigned in meeting abstracts before the study was published in a journal (Nordic APSG 1995). The explanation for this discrepancy was that some participants who did not meet inclusion criteria were recruited and were not included in the final analysis (personal communication, Wiseman 1999). In the United Kingdom, centres other than Sheffield were randomly assigning participants in a study of oxidised regenerated cellulose versus no treatment in the early 1990s (personal communication, A Watson), but the data have never been published (Li 1994). Attempts to identify these data by contacting investigators and the sponsoring company (Ethicon, Bridgewater, New Jersey, USA) have proved fruitless thus far. In van Geldorp 1994, another centre was recruiting participants (investigator: Trimbos‐Kemper, The Netherlands), but these results have been reported only briefly in a review (Wiseman 1999). Data regarding the use of oxidised regenerated cellulose must be interpreted with caution because of the high risk of bias.

Additional information was sought from some study authors; at the time of this writing, six replies had been received (Azziz 1993; Franklin 1995; Keckstein 1996; Myomectomy ASG 1995; Saravelos 1996; van Geldorp 1994).

Haney 1995 reported statistically significant reductions in the adhesion score, but it is not clear whether these analyses took account of the within‐participant design. It should be noted that the incidence of adhesions in the oxidised regenerated cellulose arm of the trial is greater than would be expected from the RCTs of oxidised regenerated cellulose versus no barrier agent. As it was not clear from the publication whether the surgeon was unblinded at the time of the second‐look laparoscopy, this result should be treated with some caution.

Other potential sources of bias

Statistical analysis

Of the 10 studies using a within‐participant design and reporting adhesion formation, only three presented and analysed matched data appropriately (Azziz 1993; Saravelos 1996; Sekiba 1992). Appropriate data were extracted from the reports of two others (Franklin 1995; Haney 1995), which were incorrectly analysed in the original. See Methods section for a full description. One further study failed to present comparative data and was excluded from all analyses (Greenblatt 1993).

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4

Summary of findings for the main comparison. Analysis 1 ‐ Oxidised regenerated cellulose vs no treatment at laparoscopy for adhesion prevention after gynaecological surgery.

Oxidised regenerated cellulose vs no treatment at laparoscopy for adhesion prevention after gynaecological surgery
Population: women having gynaecological surgery  Settings: surgical  Intervention: oxidised regenerated cellulose (Interceed) Comparison: no treatment at laparoscopy
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	Number of participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	No treatment	Oxidised regenerated cellulose
Pelvic pain	Not reported in any study in this comparison
Live birth rate	Not reported in any study in this comparison
Adhesion score	Not reported in any study in this comparison
Incidence of adhesions ‐ de novo  Incidence of adhesions at second‐look laparoscopy	479 per 1000	315 per 1000  (216 to 433)	OR 0.50   (0.3 to 0.83)	360  (3 studies)	⊕⊝⊝⊝  Very low^a,b,c,d
Incidence of adhesions ‐ re‐formation (or mixture)  Incidence of adhesions at second‐look laparoscopy	746 per 1000	333 per 1000  (171 to 546)	OR 0.17  (0.07 to 0.41)	100  (3 studies)	⊕⊝⊝⊝  Very Low^a,c,d
Clinical pregnancy rate	Not reported in any study in this comparison
The basis for the assumed risk* is the median control group risk across studies. 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.

Expanded polytetrafluoroethylene vs oxidised regenerated cellulose (Interceed) for adhesion prevention after gynaecological surgery
Population: women having gynaecological surgery  Settings: surgical  Intervention: expanded polytetrafluoroethylene Comparison: oxidised regenerated cellulose
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	Number of participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	Oxidised regenerated cellulose	Expanded polytetrafluoroethylene
Pelvic pain	Not reported in any study in this comparison
Live birth rate	Not reported in any study in this comparison
Adhesion score  Non‐validated score out of 11 at SLL	Mean adhesion score was ‐3.79 lower  (5.12 to 2.46 lower) in the expanded polytetrafluoroethylene group			58  (1 study)	⊕⊝⊝⊝  Very low^a,b,c
Incidence of adhesions ‐ de novo  Incidence at second‐look laparoscopy	149 per 1000	141 per 1000  (44 to 374)	OR 0.93   (0.26 to 3.41)	38  (1 study)	⊕⊝⊝⊝  Very low^c,d
Incidence of adhesions ‐ re‐formation (or mixture)  Incidence at second‐look laparoscopy	567 per 1000	146 per 1000  (26 to 512)	OR 0.13  (0.02 to 0.8)	23  (1 study)	⊕⊝⊝⊝  Very low^c,d	Confidence interval crossed the line of no effect when a risk ratio rather than an odds ratio was calculated (RR 0.36, 95% CI 0.13 to 1.01)
Clinical pregnancy rate	Not reported in any study in this comparison
The basis for the assumed risk* is the median oxidised regenerated cellulose group risk across studies. 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; RR: risk 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.

Collagen membrane with polyethylene glycol and glycerol vs Ringer's lactate
Patient or population: patients with adhesion prevention after gynaecological surgery Settings: surgical Intervention: collagen membrane with polyethylene glycol and glycerol Comparison: Ringer's lactate
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	Number of participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	Ringer's lactate	Collagen membrane with polyethylene glycol and glycerol
Pelvic pain	Not reported in any study in this comparison
Live birth rate	Not reported in any study in this comparison
Adhesion score	Median adhesion score in the control group was 0.8 compared with 1.2 in the intervention group			47 (1 study)	⊕⊝⊝⊝ Very low^a,b,c
Incidence of adhesions ‐ de novo Incidence at second‐look laparoscopy	479 per 1000	36 per 1000  (0 to 415)	OR 0.04 (0.00 to 0.77)	47  (1 study)	⊕⊕⊝⊝  Low^a,c
Incidence of adhesions ‐ re‐formation or mixture Incidence at second‐look laparoscopy	Not reported in any study in this comparison
Clinical pregnancy rate	235 per 1000	637 per 1000  (298 to 879)	OR 5.69 (1.38 to 23.48)	39  (1 study)	⊕⊝⊝⊝  Very low^a,c,d
The basis for the assumed risk* is the median control group risk across studies. 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.

Date	Event	Description
13 September 2019	New search has been performed	Review has been updated; 1 new study has been included (Canis 2014)
13 September 2019	New citation required but conclusions have not changed	Evidence regarding new interventions has been included; however due to low‐quality evidence, we made no changes to our overall conclusions at this update

Date	Event	Description
19 March 2015	New search has been performed	Review has been updated; 1 new study has been included (Tinelli 2011)
19 March 2015	New citation required but conclusions have not changed	We made no change to our conclusions at this update
2 June 2008	Amended	Review has been converted to new review format
20 December 2007	New citation required and conclusions have changed	Substantive amendments have been made
20 December 2007	New search has been performed	Review has been updated

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Incidence of adhesions	6		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 De novo	3	360	Odds Ratio (M‐H, Fixed, 95% CI)	0.50 [0.30, 0.83]
1.2 Re‐formation (or mixture)	3	100	Odds Ratio (M‐H, Fixed, 95% CI)	0.17 [0.07, 0.41]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mean adhesion score (non‐validated score)	1	58	Mean Difference (IV, Fixed, 95% CI)	‐3.79 [‐5.12, ‐2.46]
2 Incidence of adhesions	2		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2.1 De novo	1	38	Odds Ratio (M‐H, Fixed, 95% CI)	0.93 [0.26, 3.41]
2.2 Re‐formation (or mixture)	1	23	Odds Ratio (M‐H, Fixed, 95% CI)	0.13 [0.02, 0.80]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mean adhesion score (non‐validated score)	1	60	Mean Difference (IV, Fixed, 95% CI)	‐0.14 [‐0.67, 0.39]
2 Incidence of de novo adhesions per participant	1	62	Odds Ratio (M‐H, Fixed, 95% CI)	1.2 [0.42, 3.41]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Incidence of adhesions	1	47	Odds Ratio (M‐H, Fixed, 95% CI)	0.04 [0.00, 0.77]
2 Median adhesion score			Other data	No numeric data
3 Clinical pregnancy rate	1	39	Odds Ratio (M‐H, Fixed, 95% CI)	5.69 [1.38, 23.48]
4 Ectopic pregnancy rate	1	39	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Methods	Unit of randomisation: pelvic sidewall  Method of randomisation: sealed envelopes, computer‐generated  Time of randomisation: at completion of surgery  Blinding: second‐look laparoscopy performed by the same surgeon
Participants	Patients undergoing pelvic adhesiolysis by laparotomy  N = 134 (268 pelvic sidewalls) Dropouts: 64 participants were excluded for surgical procedures and/or evaluations inconsistent with the protocol (39), for inadequate documentation (23), and for not returning for second‐look laparoscopy (2)  Some participants underwent additional surgical procedures (salpingostomy, myomectomy, tubal anastomosis, endometriosis treatment)  Indication for surgery: infertility  Pre‐existing adhesions: yes  Cause of adhesions: not stated (endometriosis included)  Microsurgery: yes Age, years: > 18 (not further specified) Location: multi‐centre trial at 13 centres in North America  Timing and duration: 1986 to 1989
Interventions	Oxidised regenerated cellulose on pelvic sidewall vs uncovered opposite sidewall Other adjuvants used: none
Outcomes	Adhesions at second‐look laparoscopy: at 10 days to 14 weeks Incidence Per pelvic sidewall Per organ Matched pair analysis According to use of sutures According to presence of adhesions Extent: Area Area differential % improvement Area: measured (not specified) Severity (graded as filmy or severe) According to type of initial adhesion Adverse effects No pregnancy outcomes
Notes	Trial supported by manufacturers of Interceed (Johnson & Johnson, Inc.)  Additional information provided by first study author  Documented with photographs and drawings  Reports of an earlier trial with a smaller sample size were published in 1989 and 1990 (see references)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Assignment performed by computerised algorithm; however study authors state that this was performed by the study's sponsor, who manufactured the intervention of interest
Allocation concealment (selection bias)	Unclear risk	Sealed envelopes were used for allocation; however no further details regarding the process of concealment are provided
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Unit of randomisation was by ovary, so each participant received both intervention and control Surgeon was not blinded, but treatment assignment was revealed only intraoperatively: "after completion of all operative procedures... a sealed envelope disclosed assignment of treated and untreated sidewall"
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Surgeon who performed the initial laparotomy was also the surgeon who performed the second‐look laparoscopy. As no independent observers were used, this could be a source of bias
Incomplete outcome data (attrition bias)   All outcomes	High risk	Dropouts: 64 participants were excluded for surgical procedures and/or evaluations inconsistent with the protocol (39), for inadequate documentation (23), and for not returning for second‐look laparoscopy (2)
Selective reporting (reporting bias)	Unclear risk	Data presented as numbers of cases and percentages; no data conversion. P values stated for outcomes. No omission of outcomes and no subsets of data. However, no pre‐published protocol for reference purposes
Other bias	Low risk	No other bias identified

Methods	Unit of randomisation: women Method of randomisation: random balanced table in numbered, sealed envelopes Time of randomisation: following operation, before closure of abdominal cavity Blinding: patients blinded, videotapes of surgery reviewed by independent blinded surgeons Multi‐centre trial
Participants	Patients undergoing myomectomy at laparotomy for excision of at least 1 myoma > 60 mm in diameter N = 61 33 randomised to intervention (collagen membrane with polyethylene glycol and glycerol (PREVADH)) and 28 randomised to control (Ringer's lactate) Dropouts: 2 patients from the intervention group withdrew due to subsequent operations; 5 patients (3 from the intervention group and 2 from the control group) did not undergo second‐look laparoscopy. Videos of surgery in 7 patients (4 from the intervention group and 3 from the control group) were not assessed by independent blinded surgeons Age, years: 20 to 42; median 34.0 in intervention group and 34.5 in control group Exclusion criteria: endometriosis (stage > 1), preoperative embolisation, submucous myoma, chronic corticotherapy, immunosuppressive or immunomodulatory treatment, treatment with a gonadotrophin‐releasing hormone (GnRH) analogue within 3 months of the study, adnexal adhesions discovered during surgery Baseline characteristics: no significant differences between groups stated Indication for surgery: preservation of fertility Pre‐existing adhesions: none Microsurgery: no Additional surgical procedures: none stated Location: 11 centres in France Timing: May 2006 to June 2008
Interventions	Collagen membrane with polyethylene glycol and glycerol (PREVADH) vs instillation of 500 mL Ringer's lactate Other adjuvants used: none stated Second‐look laparoscopy: Timing: 10 to 20 weeks following initial surgery Surgeon blinded: no Adhesiolysis: not stated
Outcomes	Adhesions at second‐look laparoscopy: Adhesion score (AFS and mAFS) Adhesion incidence Incidence of severe adhesions Pregnancy outcomes: Clinical pregnancy rate Ectopic pregnancy rate Adverse events
Notes	Study sponsored and funded by Covidien Intention to treat Surgery documented by video; however videos not analysed in 7 cases Power calculation: stated 34 patients needed in each arm Prospectively registered at US clinical trials database: NCT01388907
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random allocation was performed using a balanced random numbers table
Allocation concealment (selection bias)	Low risk	Blinded, sealed, numbered envelopes were opened in the operating theatre to decide treatment allocation
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Study does not clearly state that patients were blinded, but most patients had the initial surgery and an SLL Surgeons were not blinded, but the randomisation process was applied intraoperatively to reduce the risk of performance bias: "after the uterus was sutured, but before closing the abdominal wall, patients were randomly allocated to one of two treatment groups, application of the anti‐adhesive film to the surgical site (P‐Group), or instillation of 500 mL Ringer's lactate solution into the pelvic cavity (R‐group, control)"
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Independent blinded surgeons were asked to review videos recorded during the second‐look laparoscopy to assess for adhesion development
Incomplete outcome data (attrition bias)   All outcomes	High risk	3‐year assessments completed in only 61% of patients
Selective reporting (reporting bias)	Unclear risk	No clear omissions of outcomes; data presented clearly, no subsets of data omitted. No protocol identified for comparison.
Other bias	Low risk	Coelioscopy second‐look adhesions performed later in control group than in intervention group (4.1 vs 2.9 months) Recruitment stopped early despite initial sample size calculations requiring 80 patients Interim analysis performed without type 1 error adjustment Effectiveness of anti‐adhesion film in preserving fertility interpreted with caution as confounding infertility factors unavailable and intention to preserve fertility may have changed during 3‐year follow‐up

Methods	Unit of randomisation: women  Method of randomisation: computer‐generated  Timing of randomisation: end of surgery  Blinding: surgeon unaware of randomisation at second‐look laparoscopy  Multi‐centre trial
Participants	Women undergoing uterine myomectomy at laparotomy with at least 1 posterior uterine incision ≥ 1 cm. Similar baseline characteristics, including age, size and number of myomas, and number of uterine incisions N = 127  Microsurgery: no  Dropouts: 6 (withdrew or excluded because of new medications)  Age, years: 34  Location: USA (19 centres)  Timing: 1993 to 1995  Pre‐existing adhesions  Cause of adhesions: not stated
Interventions	Sodium hyaluronate and carboxymethylcellulose vs no treatment
Outcomes	Adhesion formation at the time of second‐look laparoscopy (mean days 23, range 7 to 70)  Outcomes included number of adhesion sites, severity score, and surface area of uterine adhesions  No pregnancy outcomes reported
Notes	Power calculations: not stated  Documented by video  Multi‐centre trial  Sponsored by Genzyme  No intention‐to‐treat analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Clear method of randomisation derived from computer‐generated randomisation list No clear source of bias Quote: "at each centre each enrolled patient was assigned consecutively a study number that corresponded to an identically numbered sealed study envelope, which determined the patient's treatment assignment via a computer derived randomisation list. After completion of myomectomy, patients were randomised at each centre into two groups"
Allocation concealment (selection bias)	Low risk	Numbered, sealed envelopes created before patient enrolment used to provide allocation, decreasing risk of bias Quote: "at each centre each enrolled patient was assigned consecutively a study number that corresponded to an identically numbered sealed study envelope, which determined the patient's treatment assignment via a computer derived randomisation list. After completion of myomectomy, patients were randomised at each centre into two groups"
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Study authors do not explicitly state that patients were blinded, but all patients had the initial myomectomy and the SLL: "after completion of the myomectomy and associated procedures, but before closure of the abdominal cavity, patients were randomized at each center into two groups: no treatment controls and patients to receive the Seprafilm study device" Surgeons were not blinded, but randomisation took place intraoperatively to minimise performance bias: "after myomectomy, just before closure of the abdominal cavity, women randomized for no treatment (n = 68) received neither Seprafilm nor a placebo"
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Videotapes of the SLL were reviewed by an independent blinded observer: "videotapes recorded during laparoscopy subsequently were reviewed at a central site by a single independent observer blinded to the patient's treatment assignment, to serve as the basis for the efficacy data on Seprafilm"
Incomplete outcome data (attrition bias)   All outcomes	Low risk	6 (out of 127) participants were withdrawn
Selective reporting (reporting bias)	Unclear risk	Data presented graphically and as percentages. P values stated for outcomes. No omission of outcomes and no subsets of data. Although this was an infertility study, no pregnancy outcomes stated
Other bias	Low risk	No other bias identified

Methods	Unit of randomisation: ovary  Method of randomisation: blocks of 2 with sealed envelopes  Timing of randomisation: during surgery  Blinding: yes 57 participants randomly assigned  2 exclusions  55 participants (110 ovaries) analysed Trial supported by manufacturers of Interceed (Johnson & Johnson Medical, Inc.)
Participants	Patients undergoing bilateral ovarian surgery by laparotomy (N = 55)  Dropouts: not stated  Indication for surgery: bilateral ovarian disease (adhesions 37; tubal occlusion 7; PID history 4; endometriosis 5; other 2); not exclusively participants with infertility  Pre‐existing adhesions: some (39/55)  Cause of adhesions: not stated (endometriosis included)  Microsurgery at surgeon's discretion Age, mean, years: 29.9 (range 22 to 41) Location: 6 centres (Houston, Texas; Detroit, Michigan; Richmond, Virginia; Grand Rapids, Michigan ‐ USA; Sydney ‐ Australia; Danderyd ‐ Sweden)  Timing and duration: 1990 to 1993
Interventions	Oxidised regenerated cellulose on ovary vs uncovered opposite ovary Other adjuvants used: none Second‐look laparoscopy Timing: 10 to 98 days after initial surgery Surgeon unaware: not stated Adhesiolysis: yes
Outcomes	Adhesions at second‐look laparoscopy: second or third week Incidence: Per pelvic sidewall Matched pair analysis According to diagnosis Extent: Area (measured with laparoscopic probe) Area differential % improvement According to diagnosis (area differential) According to time interval to SLL (area differential) Severity: Graded 0 to 3 (0 = no adhesions; 1 = filmy adhesions; 2 = organised adhesions; 3 = cohesive adhesions) Grade differential Pregnancy outcomes: nil  Adverse effects
Notes	Power calculations: nil  Preliminary report of this trial published as abstract in 1993 (see references)  Documentation: video  Sponsored by Johnson & Johnson
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No clear description of method of randomisation provided Quote: "before beginning the study, a treatment randomisation scheme was generated. Patients were randomised in blocks of two to provide balance with respect to the left and right ovaries"
Allocation concealment (selection bias)	Unclear risk	Sealed envelopes used to conceal allocations; unclear but appear to have been created after randomisation already decided Quote: "randomised treatment assignments were placed in sealed envelope for each patient, thereby blinding the surgeon to the treatment assignment during surgery"
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Unit of randomisation was by ovary, so each participant received both intervention and control Surgeon was not blinded, but treatment assignment was revealed only intraoperatively: "randomized treatment assignments were placed in sealed envelopes for each patient, thereby blinding the surgeon to the treatment assignment during surgery. At that time, a sealed envelope was opened, disclosing the assignment of either the left or right ovary for wrapping with Interceed"
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	It is not stated whether the surgeon performing SLL was blinded to which ovary received which intervention
Incomplete outcome data (attrition bias)   All outcomes	Low risk	2 exclusions (out of 57 participants): 1 due to complete lysis of adhesions, and the second due to severe adhesions of pelvic wall at SLL
Selective reporting (reporting bias)	Unclear risk	Data presented as numbers of cases; no data conversion. P values stated for outcomes. No omission of outcomes and no subsets of data. No pre‐published protocol identified
Other bias	Low risk	No other bias identified

Methods	Unit of randomisation: participant  Method of randomisation: unknown  Time of randomisation: unknown  Blinding: unknown
Participants	Patients undergoing myomectomy by laparoscopy  N = 38  Dropouts: not known  Myomectomy details unknown Indication for surgery: fibroids  Pre‐existing adhesions: no  Microsurgery: no Age, mean, SD: unknown Location: Germany  Timing and duration: unknown
Interventions	Oxidised regenerated cellulose applied to myomectomy incision(s) vs Gortex applied to myomectomy site(s) Other adjuvants used: none Second‐look laparoscopy Timing: unknown Surgeon unaware: unknown Adhesiolysis: unknown
Outcomes	Adhesions at second‐look laparoscopy Incidence (per participant) Severity: graded as none, mild, and severe Adverse effects: unknown Pregnancy: unknown
Notes	Power calculations not stated  Documentation not stated  No source of funding stated
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not stated within study
Allocation concealment (selection bias)	Unclear risk	Not stated within study
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Not stated within study
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not stated within study
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	No mention of whether withdrawals or dropouts occurred during the study
Selective reporting (reporting bias)	Unclear risk	Data presented as numbers of cases and percentages. P values stated for significant outcomes. No omission of outcomes and no subsets of data. However, no pre‐published protocol identified
Other bias	Low risk	No other bias identified

Methods	Unit of randomisation: myomectomy site  Method of randomisation: computerised randomisation table ‐ sealed envelopes  Timing of randomisation: at completion of surgery  Blinding: not stated whether surgeon was aware of randomisation at second‐look laparoscopy
Participants	Patients undergoing myomectomy (2 or more) by laparotomy  N = 28  1 exclusion (declined second‐look laparoscopy)  27 participants (54 myomectomy sites) analysed  All had 2 myomectomy incisions of similar length and > 2 cm apart located on the fundus and on the posterior uterine wall Indication for surgery: fibroids; not exclusively participants with infertility  Microsurgery: no  Pre‐existing adhesions: no Age, years, range: 18 to 43 Location: USA (Durham, North Carolina; Houston, Texas; Atlanta, Georgia; San Diego, California) and Canada (Montreal)  Timing and duration: 1991 to 1993
Interventions	Expanded polytetrafluoroethylene over myomectomy site vs uncovered myomectomy site Other adjuvants used: antibiotics Second‐look laparoscopy Timing: 2 to 6 weeks after initial surgery (mean 25 days) Surgeon unaware: not stated Adhesiolysis: yes
Outcomes	Adhesions at second‐look laparoscopy: 2 to 6 weeks after initial surgery Incidence (per myomectomy site) Extent: area Adhesion score Area: measured (not specified) Scoring (0 to 11) according to (1) extent: 0 = no adhesions; 1 = < 25%; 2 = 25% to 50%; 3 = 50% to 75%; 4 = > 75%; (2) type: 0 = none; 1 = filmy, transparent, avascular; 2 = opaque, transparent, avascular; 3 = opaque, capillaries present; 4 = opaque, large vessels; (3) tenacity: 0 = none; 1 = adhesions essentially fell apart; 2 = adhesions were lysed with traction; 3 = adhesions required sharp dissection  Adverse effects Pregnancy outcomes: nil
Notes	Documented with diagrams, photographs, and videotapes  Sponsored by Gore‐Tex Surgical Membrane  Multi‐centre trial  Power calculations: nil  Preliminary reports of this trial published as abstracts 1993 and 1994  No intention‐to‐treat analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation performed using a computerised randomisation table
Allocation concealment (selection bias)	Low risk	Assignments were concealed in sealed envelopes opened intraoperatively
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	No clear statement whether initial surgeon or participants were blinded
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	No clear statement whether surgeon performing second‐look laparoscopy or assessment at SLL was blinded
Incomplete outcome data (attrition bias)   All outcomes	Low risk	One participant withdrew from study before SLL
Selective reporting (reporting bias)	Unclear risk	Data presented as incidence of adhesions with CI and adhesion scores as means ± SEM. P values reported. No omissions or subsets of data. However, no pre‐published protocol identified
Other bias	Low risk	No other bias identified

Methods	Unit of randomisation: adnexae  Method of randomisation: sealed envelopes  Timing of randomisation: at completion of surgery  Blinding: yes 66 participants (132 ovaries)  No exclusions stated Trial supported by manufacturers of Interceed (Johnson & Johnson, Stockholm, Sweden)
Participants	Patients undergoing adhesiolysis by laparotomy  N = 66 (14 more were erroneously entered, as they did not meet entry criterion of pre‐existing adhesions)  Indication for surgery: infertility  Pre‐existing adhesions: yes  Cause of adhesions: not stated (endometriosis excluded)  Microsurgery: yes Age: not stated  No details on infertility status (duration of infertility, tubal status, ovulatory status, semen analysis) Location: Scandinavia, Sweden (Stockholm, Linkoping, Goteborg, Gavle, Umea, and Skovde); 1 in Finland (Oulo), 1 in Denmark (Aalborg)  Timing and duration: 1991 to 1993
Interventions	Oxidised regenerated cellulose on ovary and fallopian tube vs uncovered opposite side Other adjuvants used: none Second‐look laparoscopy Timing: 4 to 10 weeks after initial surgery Surgeon unaware: not stated Adhesiolysis: not stated
Outcomes	Outcomes analysed in review Adhesions at second‐look laparoscopy Incidence: Per ovary Per tube Per fimbria Adhesion score (0‐4) (max score = 24; 0 = no adhesions; 1 = minimal adhesions; 2 = mild adhesions: 1 or 2 simple thin strands less than 1 cm in width; 3 = moderate adhesions: more than 2 adhesions of type 2 or at least 1 solid adhesion; 4 = severe adhesion: more than type 3): In aggregate Per ovary Per tube Per fimbria Adverse effects Pregnancy outcomes: nil
Notes	Power calculations: nil  Sponsored by Johnson & Johnson  Multi‐centre trial  Duplicate publication with Larsson 93 (but different numbers) Preliminary reports of this trial published as abstracts in 1993 and 1994 (see references)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No clear statement regarding method of randomisation
Allocation concealment (selection bias)	Unclear risk	Sealed envelopes used for allocation concealment but unclear when sealed or opened. Quote: "a sealed envelope disclosed assignment of which side was to be treated with the adjuvant therapy"
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Unit of randomisation was right or left adnexa, so every woman also acted as her own control Not clearly stated whether surgeon was blinded, or when treatment allocation was revealed
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	No clear statement whether surgeon performing second‐look laparoscopy or assessment at SLL was blinded
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	No mention of whether withdrawals or dropouts occurred during the study
Selective reporting (reporting bias)	High risk	Data presented in graphical form, as well as in tables as means ± SEMs. P values reported in table. Data reported for adhesions on ovaries, fallopian tubes, fimbriae. Data reported elsewhere with different sample size. Despite being an infertility study, pregnancy outcomes not stated
Other bias	Low risk	No other bias identified

Methods	Unit of randomisation: ovaries  Method of randomisation: computer‐generated in opaque envelopes  Timing of randomisation: unclear  Blinding: double
Participants	Women (N = 21) undergoing treatment for polycystic ovarian syndrome by ovarian electrocautery  7 dropouts: 4 were pregnant, 2 withdrew, 1 had such dense adhesions at laparoscopy of both ovaries that no assessment could be made Pre‐existing adhesions: nil  Mean age, years: 28  No other cause for infertility found  Timing: 1994 to 1995 Country: UK (1 centre)
Interventions	Oxidised regenerated cellulose vs no treatment
Outcomes	Adhesion formation Incidence Extent Severity Further adhesiolysis  Timing of second‐look laparoscopy: 2 to 11 weeks  Pregnancy: outcomes given but ovaries randomly assigned, not participants
Notes	Supported by Johnson & Johnson  Video taken  Power calculations: nil  No intention‐to‐treat analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No clear statement provided regarding method of randomisation
Allocation concealment (selection bias)	Unclear risk	Sealed envelopes used for allocation concealment but unclear when sealed or opened. Quote: "a sealed envelope disclosed the assignment of which side was to be treated with Interceed"
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Unit of randomisation was right or left ovary, so every woman also acted as her own control Not clearly stated whether surgeon was blinded, or when treatment allocation was revealed
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Second‐look laparoscopy was video‐recorded and assessed by independent blinded assessors
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	1 participant excluded, as underwent ovarian drilling; 4 withdrew because of pregnancy; 2 withdrew because of personal choice (n = 21)
Selective reporting (reporting bias)	Unclear risk	Data reported as incidences and percentages. No data conversion. No P values reported
Other bias	Low risk	No other bias identified

Methods	Unit of randomisation: 546 participants  Time of randomisation: at completion of surgery  Blinding: single Method of randomisation: block randomisation was used to ensure balance in the numbers of participants in the 4 treatment arms. A statistician generated the entire randomisation sequence list in advance, and allocations were sequentially numbered from beginning to end  Exclusion criteria for the investigation: previous uterine or pelvic surgery; previous abdominal general surgery; presurgical treatment with GnRH analogues; gynaecological malignancy; pregnancy; use of any instillation such as 32% dextran‐70, corticosteroids, anticoagulants, or non‐steroidal anti‐inflammatory drugs; haematological or coagulation disorders; presence of ongoing pelvic infection
Participants	546 participants underwent surgery Type of surgery: laparoscopic or abdominal single or multiple intracapsular myomectomy Reasons for surgery: pelvic pain, menorrhagia and growth of fibroids as verified by ultrasound; some women requested myomectomy because of infertility Mean age between groups, years: 28.9 to 30.2 Timing of second‐look surgery between groups, years: 2.3 to 2.6 Dates: January 2003 to June 2009
Interventions	Participants were subdivided into groups for laparoscopic or abdominal surgery. Participants underwent single or multiple intracapsular myomectomy. Participants were randomly assigned to placement of oxidised regenerated cellulose absorbable adhesion barrier to the uterine incision or to control without barriers
Outcomes	Primary and secondary outcomes of analysis: presence and severity of adhesions for 4 groups: laparotomy with barrier, laparotomy without barrier, laparoscopy with barrier, and laparoscopy without barrier
Notes	Funding: no external source stated Power calculation: stated
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation list used to reduce bias Quote: "after completion of reconstructive uterine surgery, subjects were assigned to the treatment in a 1:1 ratio using a randomization list with random permutated blocks, length of 4"
Allocation concealment (selection bias)	Unclear risk	Methods of allocation concealment not stated in the text
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Not stated whether initial surgeon was blinded; however as control was no treatment, this appears unlikely. Not clearly stated whether participants were blinded
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Surgeons who performed second‐look laparoscopy and assessment were blinded to treatment allocation
Incomplete outcome data (attrition bias)   All outcomes	High risk	High attrition rate, intentionally excluding patients not requiring repeat surgery, even though these patients may have had asymptomatic adhesions Quote: "of the 694 enrolled patients, 546 (78%) were available for assessment of adhesions during a subsequent operation" Adhesions were assessed at second surgery only if the participant had undergone subsequent surgery over the 6 years. Arguably, study data may be skewed by not including the 21% of participants who did not require subsequent surgery. Reasons for subsequent surgery include "laparotomy for cesarean section, laparotomy or laparoscopy for ovarian cysts, recurrent fibroids, appendectomy, cholecystectomy, extrauterine pregnancy and infertility"
Selective reporting (reporting bias)	Unclear risk	Data reported as incidences and percentages. P values reported only for significant outcomes. No differences identified between methods and reported results. However, no pre‐publication protocol identified for reference purposes
Other bias	Low risk	No other bias identified

Study	Reason for exclusion
Diamond 1998	Excluded, as trial intervention was a solution ‐ not a barrier
Dunn 2002	Excluded, as trial intervention was a solution ‐ not a barrier
Korrell 1998	Excluded, as trial intervention was a solution ‐ not a barrier
Korrell 2000a	Excluded, as trial intervention was a solution ‐ not a barrier
Korrell 2000b	Excluded, as trial intervention was a solution ‐ not a barrier
Liu 2015	Excluded as trial intervention was a solution ‐ not a barrier
Reid 1993	Not an intervention that was sought by the objectives of the review
Reid 1997	Not an intervention that was sought by the objectives of the review
To 1992	Excluded, as no outcomes of interest to this review (fever and need for antibiotics)
Trew 2016	Excluded, as trial intervention was a solution ‐ not a barrier

PERMALINK

Barrier agents for adhesion prevention after gynaecological surgery

Gaity Ahmad

Kyungmin Kim

Matthew Thompson

Priya Agarwal

Helena O'Flynn

Akshay Hindocha

Andrew Watson

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

1.

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Overall quality of the body of evidence: 'Summary of findings' table

Results

Description of studies

Results of the search

Included studies

Trial design and setting

Number of studies

Multi‐centre trials

Design

Support/sponsorship

Participants

Primary indications for surgery

Type of surgery

Suturing

Adjuvants to surgery

Microsurgical techniques

Interventions

Timing of second‐look laparoscopy (SLL)

Studied outcomes

Primary outcomes

Pelvic pain

Live birth rate

Secondary outcomes

Adhesion score

Number of participants with adhesions at SLL

Clinical pregnancy rate

Miscarriage rate

Ectopic pregnancy rate

Quality of life

Adverse effects

Excluded studies