Endometrial injury in women undergoing in vitro fertilisation (IVF)

Sarah F Lensen; Sarah Armstrong; Ahmed Gibreel; Carolina O Nastri; Nick Raine-Fenning; Wellington P Martins

doi:10.1002/14651858.CD009517.pub4

. 2021 Jun 10;2021(6):CD009517. doi: 10.1002/14651858.CD009517.pub4

Endometrial injury in women undergoing in vitro fertilisation (IVF)

Sarah F Lensen ^1,^✉, Sarah Armstrong ², Ahmed Gibreel ³, Carolina O Nastri ⁴, Nick Raine-Fenning ⁵, Wellington P Martins ⁴

Editor: Cochrane Gynaecology and Fertility Group

PMCID: PMC8190981 PMID: 34110001

Abstract

Background

Implantation of an embryo within the endometrial cavity is a critical step in the process of in vitro fertilisation (IVF). Previous research has suggested that endometrial injury (also known as endometrial scratching), defined as intentional damage to the endometrium, can increase the chance of pregnancy in women undergoing IVF.

Objectives

To assess the effectiveness and safety of endometrial injury performed before embryo transfer in women undergoing in vitro fertilisation (IVF) including intracytoplasmic sperm injection (ICSI) and frozen embryo transfer.

Search methods

In June 2020 we searched the Cochrane Gynaecology and Fertility Group Specialised Register, CENTRAL, MEDLINE, Embase, CINAHL, LILACS, DARE and two trial registries. We also checked the reference sections of relevant studies and contacted experts in the field for any additional trials.

Selection criteria

Randomised controlled trials comparing intentional endometrial injury before embryo transfer in women undergoing IVF, versus no intervention or a sham procedure.

Data collection and analysis

We used standard methodological procedures recommended by Cochrane. Two independent review authors screened studies, evaluated risk of bias and assessed the certainty of the evidence by using GRADE (Grading of Recommendation, Assessment, Development and Evaluation) criteria. We contacted and corresponded with study investigators as required. Due to the high risk of bias associated with many of the studies, the primary analyses of all review outcomes were restricted to studies at a low risk of bias for selection bias and other bias. Sensitivity analysis was then performed including all studies. The primary review outcomes were live birth and miscarriage.

Main results

Endometrial injury versus control (no procedure or a sham procedure)

A total of 37 studies (8786 women) were included in this comparison. Most studies performed endometrial injury by pipelle biopsy in the luteal phase of the cycle before the IVF cycle. The primary analysis was restricted to studies at low risk of bias, and included eight studies. The effect of endometrial injury on live birth is unclear as the result is consistent with no effect, or a small reduction, or an improvement (odds ratio (OR) 1.12, 95% confidence interval (CI) 0.98 to 1.28; participants = 4402; studies = 8; I² = 15%, moderate‐certainty evidence). This suggests that if the chance of live birth with IVF is usually 27%, then the chance when using endometrial injury would be somewhere between < 27% and 32%.

Similarly, the effect of endometrial injury on clinical pregnancy is unclear (OR 1.08, 95% CI 0.95 to 1.23; participants = 4402; studies = 8; I² = 0%, moderate‐certainty evidence). This suggests that if the chance of clinical pregnancy from IVF is normally 32%, then the chance when using endometrial injury before IVF is between 31% and 37%. When all studies were included in the sensitivity analysis, we were unable to conduct meta‐analysis for the outcomes of live birth and clinical pregnancy due to high risk of bias and statistical heterogeneity.

Endometrial injury probably results in little to no difference in chance of miscarriage (OR 0.88, 95% CI 0.68 to 1.13; participants = 4402; studies = 8; I² = 0%, moderate‐certainty evidence), and this result was similar in the sensitivity analysis that included all studies. The result suggests that if the chance of miscarriage with IVF is usually 6.0%, then when using endometrial injury it would be somewhere between 4.2% and 6.8%.

Endometrial injury was associated with mild to moderate pain (approximately 4 out of 10), and was generally associated with some minimal bleeding.

The evidence was downgraded for imprecision due to wide confidence intervals and therefore all primary analyses were graded as moderate certainty.

Higher versus lower degree of injury

Only one small study was included in this comparison (participants = 129), which compared endometrial injury using two different instruments in the cycle prior to the IVF cycle: a pipelle catheter and a Shepard catheter. This trial was excluded from the primary analysis due to risk of bias. In the sensitivity analysis, all outcomes reported for this study were graded as very‐low certainty due to risk of bias, and as such we were not able to interpret the study results.

Authors' conclusions

The effect of endometrial injury on live birth and clinical pregnancy among women undergoing IVF is unclear. The results of the meta‐analyses are consistent with an increased chance, no effect and a small reduction in these outcomes. We are therefore uncertain whether endometrial injury improves the chance of live birth or clinical pregnancy in women undergoing IVF. Endometrial injury does not appear to affect the chance of miscarriage. It is a somewhat painful procedure associated with a small amount of bleeding. In conclusion, current evidence does not support the routine use of endometrial injury for women undergoing IVF.

Plain language summary

Endometrial injury in women undergoing in vitro fertilisation (IVF)

Review question

To assess whether it is safe and effective to perform endometrial injury (also known as endometrial scratching), in women undergoing in vitro fertilisation (IVF), including intracytoplasmic sperm injection (ICSI) and frozen embryo transfer.

Background

Couples who have trouble getting pregnant may seek fertility treatments to help them conceive, such as IVF. In an IVF cycle, eggs are collected from the woman and are combined with sperm in the laboratory to create embryos. Embryos are transferred into the womb in the hope that they will implant and establish a pregnancy. Implantation is the process by which an embryo embeds itself into the lining of the womb; it is the first step toward establishing a successful pregnancy. It has been suggested that the chances of implantation are increased if endometrial injury is performed before replacement of the embryo into the womb.

Study characteristics

We included 38 clinical trials (8915 women) which had tested the effects of endometrial injury on the outcomes of IVF. The studies were conducted in different populations of women, and the way the endometrial injury was conducted also differed between studies in terms of the instrument used and the timing of the procedure in relation to the IVF cycle. Many of the studies were poor quality and at high risk of bias, and therefore we performed the main analyses only including studies that were not at high risk of bias. Of the 38 included studies, only eight were included in the primary analyses.

Key results

It is unclear whether endometrial injury affects the chance having a baby from IVF. The results suggest that, if the chance of having a baby from IVF is usually about 27%, then the chance of having a baby when using endometrial injury before IVF would be somewhere between less than 27% and 32%. Similarly, for the outcome of pregnancy, if the chance of getting pregnant from IVF is assumed to be about 32%, then the chance of pregnancy when using endometrial injury before IVF is somewhere between 31% and 37%.

Endometrial injury does not appear to affect the chance of having a miscarriage from IVF. The endometrial injury procedure causes mild to moderate pain and a small amount of vaginal bleeding, although this is short‐lived. This evidence does not support the routine use of endometrial injury for women undergoing IVF.

One small study compared endometrial injury using two different instruments in the cycle prior to the IVF cycle. All outcomes reported for this study were graded as very‐low certainty due to risk of bias, and as such we were not able to interpret the study results.

Certainty of the evidence

For the primary analyses, the evidence is of moderate certainty. The evidence was reduced because the results were imprecise, and consistent with endometrial injury having no effect, being beneficial, and being harmful to the chance of getting pregnant or having a baby.

Summary of findings

Background

Description of the condition

Assisted reproductive techniques (ART) include treatments and procedures requiring in vitro handling of human oocytes and sperm, or of embryos, with the objective of achieving pregnancy and live birth. The most common forms of ART include in vitro fertilisation (IVF) and intracytoplasmic sperm injection (ICSI). We use the term 'IVF' to refer to IVF, ICSI and frozen embryo transfer. IVF is a widely used treatment for infertility, but success rates remain relatively low, with less than 30% of treatment cycles resulting in a live birth (Gunby 2010). A key determinant of treatment success is implantation of the embryo, which depends on two factors: the quality of the embryo and the receptivity of the endometrium. Even when good‐quality embryos are transferred, implantation may not occur. The cumulative chance of achieving a pregnancy improves with second and third attempts, but thereafter the incremental improvement per additional cycle is reduced (Luke 2012). Recurrent implantation failure (RIF) is a clinical situation with many definitions; most published articles define RIF as failure of two to six previous treatment cycles (Polanski 2014). Implantation is a complex process, which is thought to be due in part to endometrial receptivity. One intervention suggested to improve endometrial receptivity is physical injury to the endometrium, also known as endometrial injury or endometrial scratching.

Description of the intervention

Endometrial injury, also known as endometrial scratching, is defined as intentional damage to the endometrium performed with the objective of improving the reproductive outcomes of women or couples desiring pregnancy. The most common intervention is endometrial injury performed using a pipelle. In the context of an IVF cycle, endometrial injury is performed some time prior to embryo transfer and can be performed as an office procedure, with or without ultrasound guidance. To perform endometrial injury, a speculum is inserted into the vagina and the external cervical os is located. A pipelle or similar device is then introduced through the external os and advanced until it reaches the uterine fundus and a sample of the endometrium is retrieved by suction and rotation within the uterine cavity. The movements made to obtain the sample are believed to result in some beneficial disturbance or "injury" to the endometrium, which may facilitate implantation and therefore increase the chance of pregnancy.

How the intervention might work

The underlying mechanism of how endometrial injury may improve endometrial receptivity remains unclear, however several pathways have been hypothesised. The first hypothesis suggests that the mechanical effect of local injury to the proliferative endometrium induces endometrial decidualisation, a process that naturally occurs in preparation for pregnancy and therefore favours implantation (Li 2009; Zhou 2008). A second hypothesis is that the injury induces a wound‐healing response, which involves recruitment of immune system cells to the site of healing (Siristatidis 2014), as it is associated with a significant increase in the secretion of cytokines, interleukins, growth factors, macrophages and dendritic cells ‐ all of which are beneficial for embryo implantation (Gnainsky 2010; Li 2009). Recruited immune cells are capable of living for months and are able to differentiate into tissue‐resident macrophages or dendritic cells, thus playing a direct role in decidual development and embryo implantation (Siristatidis 2014). Uterine natural killer (NK) cells are a major source of immunoregulatory cytokines in the endometrium and are thought to be reduced in numbers during controlled ovarian stimulation (Siristatidis 2014). Endometrial injury may increase the quantity of NK cells within the endometrium, restoring these to sufficient numbers (Junovich 2011). Cytokines, growth factors and NK cells are also responsible for increased angiogenesis, thereby providing adequate blood flow to the tissue and preventing embryo rejection (Siristatidis 2014). A third hypothesis is related to the observation that ovarian stimulation during IVF leads to abnormal maturation of the endometrium, such that it is advanced at the time of embryo transfer and may be less receptive to an implanting embryo (Lass 1998; Ubaldi 1997). This hypothesis suggests that endometrial injury retards endometrial maturation, leading to better synchronicity between the endometrium and the transferred embryo (Li 2009).

Why it is important to do this review

The first published study examining this intervention suggested it could improve implantation (Barash 2003), and this sparked a series of trials investigating this IVF'add on'. Subsequently the intervention was adopted as a routine intervention in many clinics worldwide (Lensen 2016b; Spencer 2016). Several laboratory studies have indicated that endometrial injury may induce changes in the endometrium that improve reproductive outcomes in women undergoing IVF cycles. It is necessary to identify, evaluate and summarise the evidence for endometrial injury as a fertility treatment or 'add on' in women undergoing IVF cycles, to enable evidence‐based medicine to be practised.

If endometrial injury is beneficial, it will help many women and couples to conceive from IVF. Similarly, if endometrial injury is of no benefit, or is detrimental to a couple's chances of a having a baby through IVF, then it can be abandoned as a technique, which currently costs patients as much as £400 in some clinics (Lensen 2016b), and is often considered to be painful.

Objectives

To assess the effectiveness and safety of endometrial injury performed before embryo transfer in women undergoing in vitro fertilisation (IVF), including intracytoplasmic sperm injection (ICSI) and frozen embryo transfer.

Methods

Criteria for considering studies for this review

Types of studies

Only randomised controlled trials (RCTs) were considered eligible; quasi‐ and pseudo‐randomised trials were not included. Cross‐over trials were eligible for inclusion for completeness, but only data from the first phase would be pooled in the meta‐analysis because this design is not valid in the context of subfertility trials (Vail 2003). However, no cross‐over trials were identified .

Types of participants

Women undergoing IVF (including ICSI and embryo transfer).

Types of interventions

The studied intervention is intentional endometrial injury performed within six months before IVF. We planned to include studies evaluating the following comparisons: (1) endometrial injury versus no intervention or a sham procedure; (2) higher versus lower degree of endometrial injury (e.g. pipelle versus hysteroscopy); and (3) different numbers of interventions (e.g. one procedure versus the procedure performed twice at two different time points).

Types of outcome measures

Primary outcomes

Effectiveness

Live birth per woman randomised.

Ongoing pregnancy, defined as a clinical pregnancy of 12 or more weeks' gestation, was used as a surrogate for live birth in cases where studies did not report live birth but reported ongoing pregnancy.

Adverse events

Miscarriage per woman randomised. Where possible, the definition used was the loss of a clinical pregnancy. In some cases, the number of miscarriages was calculated as the difference between the number of live births and clinical pregnancies, and in some cases the number of miscarriages was taken as reported in the paper.

Secondary outcomes

Effectiveness

Clinical pregnancy per woman randomised, defined as the presence of one or more gestational sacs or heartbeats on ultrasound at approximately six to eight weeks gestation.

Adverse events

Multiple gestation per woman randomised.
Pain reported during the intervention, as measured by any validated qualitative or quantitative scale.
Abnormal bleeding during or after the intervention.

Search methods for identification of studies

We searched for RCTs in accordance with a search strategy developed in consultation with the Information Specialist for the Cochrane Gynaecology and Fertility Group (CGF). We applied no language restriction.

Electronic searches

We performed the updated search on 15 June 2020 in the following databases:

the CGF Specialised Register, ProCite platform, searched 15 June 2020 (Appendix 1);
CENTRAL via the Cochrane Register of Studies (CRSO), Web platform, searched 15 June 2020 (Appendix 2);
MEDLINE, OVID platform, searched from 1946 to 15 June 2020 (Appendix 3);
Embase, OVID platform, searched from 1980 to 15 June 2020 (Appendix 4);
PsycINFO, OVID platform, searched from 1806 to 15 June 2020 (Appendix 5);
CINAHL, the Cumulative Index to Nursing and Allied Health Literature, EBSCO platform, searched from 1961 to 15 June 2020 (Appendix 6);
LILACS, Latin American Caribbean Health Sciences Literature, Web platform, searched 15 June 2020 (Appendix 7);
Epistemonikos, Web platform, searched 15 June 2020 (Appendix 8);
DARE, the Database of Abstracts of Reviews of Effects, OVID platform, 2005 to November 2012 (Appendix 9).

We searched for ongoing trials on clinicaltrials.gov (http://www.clinicaltrials.gov/), and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (http://ictrptest.azurewebsites.net/Default.aspx). We searched for conference abstracts on the Web of Science and for grey literature on OpenGrey, and we handsearched for abstracts from the European Society of Human Reproduction and Embryology 2020.

Searching other resources

We searched the reference lists of included and excluded trials, other systematic reviews of this intervention and contacted experts in the field for any additional trials.

Data collection and analysis

Selection of studies

Review authors (from SFL, AG, SA) assessed eligibility independently in duplicate and in a standardised manner. Any disagreements were resolved by discussion or consultation with a third review author.

Data extraction and management

We (from SFL, AG, CON, WPM, NR‐F) extracted data from eligible trials using a data extraction form that had been designed and pilot‐tested by the review authors. When studies had multiple publications, we used the main trial report as the reference and obtained additional details from secondary papers. We attempted to correspond with study investigators to resolve data queries as required. Data extraction was undertaken in duplicate and disagreements were resolved by discussion.

Assessment of risk of bias in included studies

Two review authors independently assessed methodological quality and data collection by using the Cochrane risk of bias assessment tool (Higgins 2011). We assessed selection bias (random sequence generation and allocation concealment), attrition bias (incomplete outcome data), reporting bias (selective reporting), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessors) and other biases (other problems that could put a trial at high risk of bias, including lack of adequate trial registration). Studies were considered low risk for performance bias if any sham procedure was used in the control arm which was likely to blind participants to their allocation. Only studies reporting subjective outcomes such as pain during the procedure or bleeding following the procedure were judged for detection bias, as it was considered that knowledge of trial allocation would not affect the assessment of objective outcomes such as clinical pregnancy and live birth. Due to recent publications which have suggested potential methodological issues with this cohort of studies (Li 2019), we elected to conduct a number of additional assessments on the included studies, some of which were considered to fall under the umbrella of 'Other' risk of bias. These included: lack of prospective trial registration, using Retraction Watch to search for publication retractions among trial authors, high publication rate of randomised trials by authors in the last 10 years (with particular focus on cases where authors had published many trials as first author), and careful scrutiny for any unlikely or implausible study characteristics such as high recruitment rates (per centre per month), inconsistencies in reporting of trial data (e.g. pregnancies that are not accounted for in studies that follow women to live birth). We presented and described all judgements in the conclusions portion of the Risk of bias in included studies tables.

Measures of treatment effect

We summarised the effects of the intervention as odds ratios (ORs) for binary outcomes (live birth, clinical pregnancy, miscarriage) and as mean differences (MDs) for continuous outcomes (pain). We evaluated the precision of the estimates by using the 95% confidence interval (CI). We considered the clinical relevance of all comparisons while taking into account the precision of the estimates. We planned to determine the number needed to treat for an additional beneficial outcome (NNTB) or for an additional harmful outcome (NNTH) when a significant difference was observed.

Unit of analysis issues

We used the number of woman randomised as the denominator for all outcomes. No ‘per cycle’ data were included.

Dealing with missing data

We analysed the data on an intention‐to‐treat (ITT) basis as far as possible and attempted to obtain missing data from the trial researchers. We assumed that participants who dropped out after randomisation (e.g. because of cycle cancellation), or who were lost‐to‐follow up or withdrew, did not achieve clinical pregnancy or live birth; no other assumptions were made.

Assessment of heterogeneity

We pooled data in a meta‐analysis only when risk of bias was not substantial and the clinical and methodological characteristics of the included studies were considered to be sufficiently similar for meta‐analysis to provide a clinically meaningful summary. We assessed heterogeneity using I² and considered I² > 50% to indicate substantial heterogeneity.

Assessment of reporting biases

We tried to minimise the potential impact of reporting biases by performing a comprehensive search for eligible studies including manual searching of conference proceedings and trial registration websites, and by looking for duplication of data. We performed funnel plot analysis when more than 10 studies were included in a comparison. When possible, we used published protocols and prospective trial registration pages for included studies to investigate selective reporting (i.e. comparisons of outcomes listed in the study protocol versus outcomes reported in papers).

Data synthesis

We carried out statistical analysis of the data in Review Manager (RevMan 5.4.1 and RevMan Web). When we considered studies to be sufficiently similar, we combined data using a fixed‐effect model.

For this update of the review, we planned three comparisons.

Endometrial injury versus no injury/sham procedure.
Higher versus lower degree of endometrial injury (e.g. pipelle versus Tao brush).
Different numbers of interventions (e.g. one procedure versus multiple procedures).

However, we only identified trials for the first and second comparison.

These three comparisons are new to this updated review. In the previous version of the review, comparisons included a timing of component, depending on when the endometrial injury was performed in the IVF cycle. In this review, the effect of timing of endometrial injury was instead investigated in subgroup analysis. Further, comparison three has been added.

Subgroup analysis and investigation of heterogeneity

We conducted five different subgroup analyses to assess treatment effects within different subgroups.

We planned to conduct subgroup analyses based on the presence of recurrent implantation failure (RIF), because women who have impaired implantation ability are, theoretically, a group with better potential to benefit from the intervention. In some cases it was possible to obtain subgrouped data from the paper or from author correspondence, in which case the study was split across subgroups. The trials were broadly grouped as per published classifications (Polanski 2014) into:

women with recurrent or previous implantation failure (≥2 previous embryo transfers);
women having their first or second IVF cycle (≤1 previous embryo transfers);
trials in which the inclusion criteria permitted any women regardless of previous IVF exposure, or where it was unclear.

In undertaking the previous update to this review (Nastri 2015), we observed that in several RCTs, some degree of intrauterine manipulation was also performed in the control groups. Such manipulation occurred by insertion of an instrument ‐ such as a hysteroscope or a uterine sound ‐ into the uterine cavity as part of the standard treatment or as a sham procedure. As such intrauterine manipulation probably causes some degree of endometrial injury, it may reduce the observed differences in reproductive outcomes caused by the intervention. To assess this, we performed a second subgroup analysis according to the type of control intervention provided:

no intrauterine manipulation in the control group;
intrauterine manipulation in the control group.

A third subgroup analysis concerns the timing of the endometrial injury procedure relative to the IVF cycle. This analysis was undertaken as it is suggested that the timing of the procedure may be important, as studies performing the procedure in the luteal phase of the cycle prior to embryo transfer have reported benefit from the procedure, whereas studies performing the procedure on the day of oocyte retrieval reported harm (Karimzade 2010). In the previous iteration of this review (Nastri 2015), the timing of the endometrial injury was embedded in the main comparisons, but in this update the effect of timing of the injury was examined with the following subgroup analyses:

follicular phase prior cycle;
luteal phase prior cycle;
early follicular phase IVF cycle;
late follicular phase IVF cycle.

Studies performing endometrial injury more than once across two different time points, or permitting the procedure to be performed anytime in a broad window, were not considered for this subgroup analysis.

A fourth subgroup analysis introduced in this update to the review concerns the intensity of the intervention, as a less disruptive procedure, such as one conducted with a Tao brush, may cause a different effect to the endometrium than a procedure conducted with a more invasive instrument such as a Novak curette. The subgroups used were:

low intensity;
moderate intensity;
high intensity.

Lastly, it was considered that the timing and intensity of the procedure may both be important to the effect of the procedure: for example, an intense procedure performed close to the time of embryo transfer may be harmful, whereas an intense procedure performed in the cycle prior to the embryo transfer could be beneficial, allowing sufficient healing time prior to possible implantation. For this reason, the fifth subgroup analysis was a combination of factors in the third and fourth analyses: intensity and timing of the procedure:

follicular phase prior cycle and low intensity;
follicular phase prior cycle and moderate intensity;
follicular phase prior cycle and high intensity;
luteal phase prior cycle and low intensity;
luteal phase prior cycle and moderate intensity;
luteal phase prior cycle and high intensity;
early follicular phase IVF cycle and low intensity;
early follicular phase IVF cycle and moderate intensity;
early follicular phase IVF cycle and high intensity;
late follicular phase IVF cycle and low intensity;
late follicular phase IVF cycle and moderate intensity;
late follicular phase IVF cycle and high intensity.

Sensitivity analysis

The primary analysis was restricted to studies at low risk of selection bias and other bias, for all review outcomes. We conducted sensitivity analyses to determine whether review conclusions were robust to arbitrary decisions made regarding eligibility and analysis. These analyses included consideration of whether review conclusions would have been different if eligibility had been restricted to studies without high risk of bias in any domain, and including all studies regardless of risk of bias. This was a post‐hoc decision and a change from the review protocol.

Summary of findings and assessment of the certainty of the evidence

We prepared a summary of findings table to evaluate the overall certainty of the body of evidence for the main review outcomes (live birth, clinical pregnancy, miscarriage) using GRADE (Grades of Recommendation, Assessment, Development and Evaluation) criteria: study limitations (i.e. risk of bias), consistency of effect, imprecision, indirectness, publication bias, for the comparison 'Endometrial injury versus control' (no procedure or a sham procedure). We prepared a second table for the comparison 'Higher versus lower degree of injury'. We justified and documented judgements about evidence certainty (high, moderate, low and very low) and incorporated this into reporting of results for live birth, clinical pregnancy and miscarriage (GRADEpro GDT, Higgins 2019).

Results

Description of studies

Results of the search

The first electronic search was on 14 December 2011, and new searches were carried out on 11 March 2014, 19 January 2015 and May 2017, February 2019 and June 2020. For this update, from 675 records screened on the basis of title and abstract, we identified 57 potentially eligible records; two review authors independently assessed these trials completely for eligibility and 24 new studies were included, giving 38 studies in total (Figure 1.

Included studies

Study design and setting

We included 38 parallel‐design RCTs in this review. All were two‐arm trials except for one study which had four arms, comparing endometrial injury in the luteal versus proliferative phases with two placebo procedures at the same time (Liu 2017). The majority of studies were published as full‐text articles, however six were available as conference abstracts only (Hur 2012, Karim Zadeh 2008; Merriam 2017; Metwally 2020; Polanski 2015; Zygula 2016), however extensive author correspondence was available in some cases. One study identified from the trial registration webpage was not published at all, however the authors provided the outcome data (Wolff 2018).

Most trials were single‐centre trials and were conducted in Brazil (Nastri 2013), China (Liu 2017; Xu 2015), Belgium (Mackens 2020), Egypt (Maged 2018; Sherif 2018), Hong Kong (Mak 2017; Yeung 2014), India (Narvekar 2010; TK 2017;), Iran (Aflatoonian 2016 Karim Zadeh 2008; Karimzadeh 2009; Karimzade 2010; Safdarian 2011), Italy (Pecorino 2018), Israel (Baum 2012), Poland (Zygula 2016), South Korea (Hur 2012), Spain (Izquierdo Rodriguez 2020), Turkey (Gurgan 2019; Guven 2014; Inal 2012), the UK (Polanski 2015), and the USA (Eskew 2018; Merriam 2017; Wolff 2018). Eleven studies were conducted at more than one centre: in Canada (Hilton 2019), China (Tang 2020), Denmark (Berntsen 2020; Olesen 2019), Egypt (Gibreel 2015), France (Frantz 2019), Iran (Shahrokh‐Tehraninejad 2016), the Netherlands (van Hoogenhuijze 2020), the UK (Metwally 2020); one in Egypt and Saudi Arabia (Shohayeb 2012), and one across 13 centres in New Zealand, Australia, Belgium, Sweden and the UK (Lensen 2019).

Author correspondence with 13 trials resulted in additional or modified outcome data not presented in the published materials (Eskew 2018; Gurgan 2019; Izquierdo Rodriguez 2020; Maged 2018; Merriam 2017; Olesen 2019; Pecorino 2018; Polanski 2015; Nastri 2013; Shohayeb 2012; Tang 2020; van Hoogenhuijze 2020; Wolff 2018).

Participants

The studies included 8915 women: 4467 women in the intervention groups and 4448 women in the control groups. Seventeen studies included only women who had undergone previous unsuccessful IVF attempts (TK 2017; Baum 2012; Berntsen 2020; Gibreel 2015; Gurgan 2019; Karim Zadeh 2008; Karimzadeh 2009; Narvekar 2010; Olesen 2019; Pecorino 2018; Shahrokh‐Tehraninejad 2016; Shohayeb 2012; Tang 2020; TK 2017; van Hoogenhuijze 2020; Wolff 2018; Zygula 2016); 13 included women regardless of the number of previous attempts (Aflatoonian 2016; Eskew 2018; Guven 2014; Izquierdo Rodriguez 2020; Lensen 2019; Mackens 2020; Mak 2017; Merriam 2017; Nastri 2013; Polanski 2015; Safdarian 2011; Xu 2015; Yeung 2014); five studies included only women undergoing their first IVF cycle (Hur 2012; Maged 2018; Metwally 2020; Liu 2017; Karimzade 2010); and two recruited women undergoing either their first or second cycle (Frantz 2019; Hilton 2019). One study included women who were either undergoing their first IVF cycle, or had not had any previous unsuccessful cycles (Sherif 2018).

Interventions

All except one of the included studies (Merriam 2017), compared endometrial injury versus either no procedure or a placebo procedure, prior to or as part of an IVF cycle. Most trials used a pipelle to perform the endometrial injury, however a Novak curette was used in three studies (Karimzade 2010; Karim Zadeh 2008; Shohayeb 2012), a modified cook catheter was used in one trial (Sherif 2018), scissors in one trial (Gurgan 2019), and forceps in another (Berntsen 2020). Most studies compared endometrial injury to no procedure, however a number of studies used a placebo procedure involving placement of the catheter or cotton bud inside the cervix (Baum 2012; Liu 2017; Mak 2017; Wolff 2018), drying the cervix with gauze (Nastri 2013), placement of catheter next to cervix (Eskew 2018), and insertion of a uterine sound (Gibreel 2015). Additionally, a number of the studies conducted the endometrial injury concurrently with hysteroscopy (Berntsen 2020; Gurgan 2019; Narvekar 2010; Shohayeb 2012), or instillation of granulocyte colony‐stimulating factor (Xu 2015), and the control arms in these studies also underwent these procedures in all but two trials (Berntsen 2020; Gurgan 2019). In one study, the researchers compared a pipelle curette with a Shepard insemination catheter (Merriam 2017).

Most studies performed the procedure only once, however it was performed twice in five studies: between days 9–12 and days 21–24 in the cycle before the IVF cycle (Baum 2012), twice between days 21‐26 of the prior cycle (Gibreel 2015), twice within one week during the luteal phase of the prior cycle (Inal 2012), once during hysteroscopy between days 7‐10 and again between days 24‐25 of the prior cycle (Narvekar 2010), and twice within 48 hours during the luteal phase of the prior cycle (TK 2017).

Most studies conducted the procedure in the cycle immediately prior to the embryo transfer cycle, either in the luteal phase (Aflatoonian 2016; Eskew 2018; Frantz 2019; Gibreel 2015; Hilton 2019; Inal 2012; Izquierdo Rodriguez 2020; Karim Zadeh 2008; Karimzadeh 2009; Maged 2018; Mak 2017; Merriam 2017; Metwally 2020; Nastri 2013; Olesen 2019; Pecorino 2018; Polanski 2015; Safdarian 2011; Shahrokh‐Tehraninejad 2016; TK 2017; van Hoogenhuijze 2020; Wolff 2018; Yeung 2014; Zygula 2016), or follicular phase (Berntsen 2020; Gurgan 2019; Shohayeb 2012; Tang 2020). A smaller number of studies conducted the procedure in the embryo transfer cycle, either in the early follicular phase (Guven 2014; Hur 2012; Mackens 2020; Sherif 2018), or in the late follicular phase (Karimzade 2010; Xu 2015). Four studies performed endometrial injury at multiple time points, either performing the procedure on multiple occasions or permitting the procedure to take place at any time across a broad window before embryo transfer (Baum 2012; Liu 2017; Lensen 2019; Narvekar 2010).

Outcomes

Thirty trials reported live birth; this includes four trials that reported only ongoing pregnancy, which was used as a surrogate for live birth (Aflatoonian 2016; Frantz 2019; Karimzade 2010; Maged 2018), and one trial that reported ongoing pregnancy/live birth (Mak 2017).

31/38 trials reported miscarriage, or the miscarriage rate was calculated from the available rates of live birth minus clinical pregnancy
38/38 trials reported clinical pregnancy
20/38 trials reported multiple pregnancy
10/38 trials reported pain
9/38 trials reported bleeding.

Excluded studies

A total of 24 studies were excluded for the following reasons: the study was not randomised (11 studies); the trial was never initiated or was terminated early with no available data (eight studies); the intervention was not eligible (three studies); no data were available (one study); and there was information available from authors that suggested the data from one trial were unreliable (one1 study) (Figure 1, Characteristics of excluded studies).

Ongoing studies

A further 19 studies appear to be ongoing according to either the trial registration page and/or from author correspondence (Characteristics of ongoing studies).

Studies awaiting classification

A total of eight studies are pending classification (Studies awaiting classification).

Risk of bias in included studies

In general the studies included in this review were small (median sample size 157, interquartile range (IQR )100 to 280) and therefore had limited power to detect clinically‐relevant differences in the outcomes included in this review. Many of the trials were also associated with serious risks of bias; every trial had one or more unclear or high risk of bias (Figure 2, Characteristics of included studies).

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

Allocation

Random sequence generation

The majority of studies used adequate methods for random sequence generation and were deemed to be at low risk for this domain. Three studies claimed to be randomised but did not report the method used for randomisation (Karim Zadeh 2008; Zygula 2016) or reported apparently conflicting information regarding the process of recruitment and allocation (Aflatoonian 2016); and were therefore judged to be at unclear risk of bias. Another trial reported that quote: "participants were randomly selected on the basis of their agreement to undergo endometrial biopsy" and it is not clear if the study was truly randomised (Safdarian 2011). One study is described as being non‐randomised on the trial registration page and is described in the paper as being a quote: "prospective case‐control study" however the authors confirmed in correspondence that computer‐based randomisation was used, therefore this study was graded as low risk (Guven 2014).

Allocation concealment

Eighteen studies used adequate methods for concealment of the random sequence and were deemed to be at low risk of selection bias.

Eleven studies did not report an attempt to conceal the allocation (Aflatoonian 2016; Baum 2012; Gurgan 2019; Hur 2012; Karim Zadeh 2008; Karimzade 2010; Liu 2017; Safdarian 2011; Sherif 2018; Xu 2015; Zygula 2016); and were judged to be at unclear risk of bias. A further three trials reported to use envelopes but it is unclear whether these met the SNOSE (sequentially numbered opaque sealed envelopes) criteria (Guven 2014; Inal 2012; Mak 2017).

Six studies were graded as high risk: two trial teams confirmed the envelopes used were not SNOSE (sequentially numbered, opaque sealed envelopes), and therefore these trials were graded as high risk (Gibreel 2015; Merriam 2017); two studies reported that patients drew numbers out of a bag to determine their allocation, and no safeguards were described to prevent the patients from replacing the paper and drawing another, therefore these studies were graded as high risk (Karimzadeh 2009; Shahrokh‐Tehraninejad 2016); a further two studies reported during author correspondence to have used an open list of allocations (Izquierdo Rodriguez 2020; Wolff 2018).

Blinding

Performance bias

Most studies were rated as high risk for performance bias as no blinding of participants or personnel was employed. In a number of cases, imbalances in the number of women reaching embryo transfer or undergoing single versus double‐embryo transfer were observed, which while minor imbalances, may have resulted from knowledge of trial allocation (Lensen 2019; Metwally 2020; van Hoogenhuijze 2020).

Studies which implemented a sham procedure were rated as low risk of bias, although it remains uncertain whether some of the sham procedures used would truly blind participants to their allocation. A number of sham procedures involved instrumentation of the cervix and uterine cavity (Gibreel 2015; Liu 2017; Pecorino 2018) or involved endocervical manipulation (Baum 2012; Mak 2017; Wolff 2018), and others involved placement of an instrument next to the cervix only (Eskew 2018; Nastri 2013). One study compared instillation of granulocyte colony stimulating factor alone with instillation in addition to endometrial injury (Xu 2015), and another study performed the procedure in addition to a hysteroscopy procedure which all women underwent (Shohayeb 2012); in these two cases it was not clear whether the participants were actually blind to their allocation or whether they were informed of it, therefore these trials were rated as unclear.

Detection bias

For the purposes of this review, lack of blinding was not considered to be a source of detection bias in studies reporting only objective outcomes such as clinical pregnancy. These trials were rated as low risk for detection bias even when no blinding was employed. For the studies reporting on pain and/or bleeding during or following the procedure, outcomes which were self‐reported by participants, lack of blinding was considered a risk of detection bias. These outcomes were reported in 10 trials; three implemented a sham procedure therefore the participants were considered blind to their allocation (Liu 2017; Nastri 2013; Pecorino 2018), and the remainder were rated as high risk (Frantz 2019; Lensen 2019; Mackens 2020; Merriam 2017; Polanski 2015; van Hoogenhuijze 2020; Yeung 2014).

Incomplete outcome data

Most of the studies were rated as low risk as it appeared that there was no or only minimal attrition. For studies reported only as abstracts, unless additional information regarding attrition could be supplied by the authors, these were rated as unclear risk (Hur 2012; Karim Zadeh 2008; Metwally 2020; Zygula 2016). A further study was rated as unclear risk as it reported that all 142 women reached embryo transfer which seems unlikely, and raises the possibility that women not reaching embryo transfer were excluded (Liu 2017). One trial was rated as high risk as one of the inclusion criteria appears to only be able to be applied after randomisation (endometrium of >7 mm on day commencing progesterone supplementation) and therefore it seems likely that participants were excluded on this basis after randomisation (Shahrokh‐Tehraninejad 2016). Similarly, in another trial the eligibility criteria would have been applied post‐randomisation (grade I or II embryos to transfer and no failure to proceed to follicle retrieval) (Guven 2014). It is unclear how many women were excluded for these reasons. Another two trials were rated as high risk due to post‐randomisation exclusions in the intervention group for women who did not undergo the injury procedure (Merriam 2017) or those with abnormalities identified (Berntsen 2020; Gurgan 2019). Further, we observed an unexplained imbalance in the number of women with cancelled or failed cycles between the two groups in one study (Gurgan 2019), and in another study, all participants from one recruiting centre were excluded due to the centre closing down and follow‐up data being unavailable (Berntsen 2020).

Selective reporting

The majority of studies did not have adequate trial registration (26 studies); 11 were registered retrospectively, four were registered late (within six months of starting recruitment), nine were not registered at all, and in two cases it was unclear as the recruitment period was unknown. Studies that were not registered or registered retrospectively were rated as high risk for selective reporting, unless they reported important review outcomes including live birth and adverse events such as pain and/or bleeding, which occurred in two instances (Liu 2017; TK 2017). A total of 12 studies were registered prospectively, and these were graded as low risk if they reported all planned (review) outcomes; in one case the study was only available as an abstract which did not report all registered outcomes, and this study was rated as unclear as it is not known whether all registered outcomes will be reported by the trial team (Metwally 2020). For two trials it was unclear whether the trial registration was prospective or not, therefore these studies were rated as unclear risk (Sherif 2018; Wolff 2018).

Other potential sources of bias

Most of the studies were rated as high risk for this domain. A total of 10 trials stopped early due to: observing a significant effect (Eskew 2018; Frantz 2019; Mackens 2020; Nastri 2013; Karimzade 2010), difficulty recruiting (Berntsen 2020; Hilton 2019; TK 2017; Wolff 2018), or unclear reasons (Baum 2012). Studies stopping early for observing a positive effect were rated as high risk; even when stopping is on the basis of adequate stopping rules, the resulting data available for meta‐analysis have been demonstrated to produce a biased effect (Bassler 2010). In cases where only abstract or limited information was available, we graded the study as high risk for other bias due to the potential for missing information to impact on trial eligibility or outcome data (Hur 2012; Karim Zadeh 2008; Merriam 2017; Wolff 2018; Zygula 2016). In one study it was observed that one trial arm had a significantly higher fertilisation rate, which would have favoured the intervention arm, and this study was therefore rated as high risk (Sherif 2018). Additionally, all trials which were not registered prospectively or within six months of initiating recruitment, were rated as high risk of other bias, due to the potential for undisclosed protocol changes or deviations which can cause bias.

A number of studies were also identified as having additional concerns, some of which were seen as representing a high risk of 'Other bias'. These included: inconsistency between reported outcomes (e.g. implantation and clinical pregnancy rates, leading to uncertainty in the accuracy of the data) or inconsistency in reported timelines (e.g. time between completing recruitment and submitting paper did not allow for follow‐up to live birth) (Aflatoonian 2016; Inal 2012; Maged 2018; Pecorino 2018; Shohayeb 2012); a number of authors on studies had also been authors on retracted publications (Aflatoonian 2016; Gurgan 2019; Karimzade 2010; Karim Zadeh 2008; Karimzadeh 2009; Xu 2015); some studies had very high recruitment rates which may or may not be explained by the nature of the trial population or setting (Guven 2014; Karimzade 2010; Maged 2018; Safdarian 2011; Tang 2020), and some authors had a large number of published trials in recent years, which may represent an improbable workload (Aflatoonian 2016; Maged 2018; Shohayeb 2012). In most cases, the implications of these additional concerns is not clear, however we present them in this review for transparency.

In previous versions of the review, exposure of the control participants to intentional or inadvertent endometrial disruption was rated as high risk of bias, due to the potential for this exposure to dilute the observed effect of endometrial injury. However, it is now considered that this does not represent a risk of bias per se, and the effect of any manipulation in the control group is explored in the subgroup analysis.

Effects of interventions

See: Table 1; Table 2

Summary of findings 1. Endometrial injury compared to control in women undergoing assisted reproductive techniques.

Endometrial injury compared to control in women undergoing assisted reproductive techniques
Patient or population: women undergoing assisted reproductive techniques Setting: clinic Intervention: endometrial injury Comparison: control
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes	Risk with control	Risk with Endometrial injury	Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Live birth per woman randomised (studies at low risk of selection bias and other bias)	273 per 1,000	296 per 1,000 (269 to 324)	OR 1.12 (0.98 to 1.28)	4402 (8 RCTs)	⊕⊕⊕⊝ MODERATE ¹
Live birth per woman randomised: sensitivity analysis (no high risk)	254 per 1,000	278 per 1,000 (177 to 409)	OR 1.13 (0.63 to 2.03)	229 (1 RCT)	⊕⊕⊝⊝ LOW ^{2 3}
Live birth per woman randomised: sensitivity analysis (including all studies)	Meta‐analysis not undertaken		‐	7792 (29 RCTs)	⊕⊝⊝⊝ VERY LOW ^{4 5}
Miscarriage per woman randomised (studies at low risk of selection bias and other bias)	60 per 1,000	53 per 1,000 (42 to 68)	OR 0.88 (0.68 to 1.13)	4402 (8 RCTs)	⊕⊕⊕⊝ MODERATE ¹
Miscarriage per woman randomised: sensitivity analysis (no high risk)	53 per 1,000	61 per 1,000 (21 to 166)	OR 1.17 (0.38 to 3.58)	229 (1 RCT)	⊕⊕⊝⊝ LOW ^{2 3}
Miscarriage per woman randomised: sensitivity analysis (including all studies)	54 per 1,000	56 per 1,000 (46 to 67)	OR 1.03 (0.85 to 1.25)	8092 (30 RCTs)	⊕⊕⊝⊝ LOW ⁴
Clinical pregnancy per woman randomised (studies at low risk of selection bias and other bias)	323 per 1,000	340 per 1,000 (312 to 370)	OR 1.08 (0.95 to 1.23)	4402 (8 RCTs)	⊕⊕⊕⊝ MODERATE ¹
Clinical pregnancy per woman randomised: sensitivity analysis (no high‐risk)	307 per 1,000	339 per 1,000 (229 to 472)	OR 1.16 (0.67 to 2.02)	229 (1 RCT)	⊕⊕⊝⊝ LOW ^{2 3}
Clinical pregnancy per woman randomised: sensitivity analysis (including all studies)	Meta‐analysis not undertaken		‐	8786 (37 RCTs)	⊕⊝⊝⊝ VERY LOW ^{4 5}
*The risk in the intervention group (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; RCT: randomised controlled trial.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

Higher compared to lower degree of injury in women undergoing assisted reproductive techniques
Patient or population: women undergoing assisted reproductive techniques Setting: clinic Intervention: higher Comparison: lower degree of injury
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)
	Risk with lower degree of injury	Risk with Higher
Live birth per woman randomised: sensitivity analysis (including all studies)	56 per 1,000	70 per 1,000 (18 to 240)	OR 1.28 (0.31 to 5.37)	129 (1 RCT)	⊕⊝⊝⊝ VERY LOW ^{1 2 3}
Miscarriage per woman randomised: sensitivity analysis (including all studies)	56 per 1,000	70 per 1,000 (18 to 240)	OR 1.28 (0.31 to 5.37)	129 (1 RCT)	⊕⊝⊝⊝ VERY LOW ^{1 2 3}
Clinical pregnancy per woman randomised: sensitivity analysis (including all studies)	111 per 1,000	141 per 1,000 (54 to 318)	OR 1.31 (0.46 to 3.73)	129 (1 RCT)	⊕⊝⊝⊝ VERY LOW ^{1 2 3}
*The risk in the intervention group (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; RCT: randomised controlled trial.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate:;the true effect is likely to be substantially different from the estimate of effect.

Date	Event	Description
15 June 2020	New citation required and conclusions have changed	The addition of 24 new studies has led to a change in the conclusions of the review.
15 June 2020	New search has been performed	Twenty‐four new trials have been included (Aflatoonian 2016; Berntsen 2020; Eskew 2018; Frantz 2019; Gurgan 2019; Hilton 2019; Hur 2012; Izquierdo Rodriguez 2020; Lensen 2019; Liu 2017; Mackens 2020; Maged 2018; Mak 2017; Merriam 2017; Metwally 2020; Olesen 2019; Pecorino 2018; Shahrokh‐Tehraninejad 2016; Sherif 2018; Tang 2020; van Hoogenhuijze 2020; Wolff 2018; Xu 2015; Zygula 2016) along with full published versions of two studies that were included as abstracts or unpublished data in the previous view (Polanski 2015; TK 2017). The search was updated in February 2019 and again in June 2020.

Date	Event	Description
21 January 2015	New citation required and conclusions have changed	Comparisons have been restructured and conclusions have changed
21 January 2015	New search has been performed	9 new studies have been included (Aleyamma 2013; Baum 2012; Gibreel 2015; Guven 2014; Karim Zadeh 2008; Polanski 2014; Safdarian 2011; Shohayeb 2012; Yeung 2014), along with full published versions of 2 studies (Inal 2012; Nastri 2013). We updated the search in January 2015 and categorised 1 study as awaiting classification (Hur 2012a)

#	Query	Results
S39	S24 AND S38	49
S38	S25 OR S26 or S27 or S28 OR S29 OR S30 OR S31 OR S32 OR S33 OR S34 OR S35 OR S36 OR S37	1,606,041
S37	TX allocat* random*	13,364
S36	(MH "Quantitative Studies")	30,673
S35	(MH "Placebos")	13,740
S34	TX placebo*	71,673
S33	TX random* allocat*	13,364
S32	(MH "Random Assignment")	68,525
S31	TX randomi* control* trial*	222,758
S30	TX ( (singl* n1 blind) or (singl n1 mask) ) or TX ( (doubl n1 blind) or (doubl n1 mask) ) or TX ( (tripl n1 blind) or (tripl n1 mask) ) or TX ( (trebl n1 blind) or (trebl n1 mask*) )	1,221,300
S29	TX ( (trebl* n1 blind) or (trebl n1 mask*) )	294
S28	TX ( (trebl* n1 blind) or (trebl n1 mask*) )	294
S27	TX clinic* n1 trial*	296,110
S26	PT Clinical trial	110,938
S25	(MH "Clinical Trials+")	320,671
S24	S15 AND S23	161
S23	S16 OR S17 OR S18 OR S19 OR S20 OR S21 OR S22	12,105
S22	TX embryo* N3 transfer*	3,517
S21	TX ovar* N3 hyperstimulat*	979
S20	TX ovari* N3 stimulat*	1,168
S19	TX IVF or TX ICSI	5,780
S18	(MM "Fertilization in Vitro")	3,973
S17	TX vitro fertilization	8,054
S16	TX vitro fertilisation	8,054
S15	S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7 OR S8 OR S9 OR S10 OR S11 OR S12 OR S13 OR S14	2,659
S14	TX endometri* N3 wound*	77
S13	TX endometri* N5 harm*	21
S12	TX endometri* N3 inflammation	69
S11	TX endometri* N3 trauma*	11
S10	TX endometri* N3 damage*	38
S9	TX(endometri* N5 lesion*)	654
S8	TX (endometri* adj5 lesion*)	0
S7	TX (endometri* adj5 lesion*)	0
S6	TX(endometri* N3 insult*)	0
S5	TX (endometri* N5 sampl*)	582
S4	TX mock N3 transfer*	11
S3	TX endometri* N3 scratch*	58
S2	TX endometri* N3 biops*	768
S1	TX endometri* N3 injury	91

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Live birth per woman randomised (studies at low risk of selection bias and other bias)	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.12 [0.98, 1.28]
1.2 Live birth per woman randomised: sensitivity analysis (no high risk)	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.13 [0.63, 2.03]
1.3 Live birth per woman randomised: sensitivity analysis (including all studies)	29		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.4 Live birth per woman randomised: subgrouping by control exposure to endometrial manipulation	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.12 [0.98, 1.28]
1.4.1 No control exposure	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.12 [0.98, 1.28]
1.5 Live birth per woman randomised: subgrouping by RIF	8		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.5.1 Not recurrent or previous implantation failure	6		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.5.2 Recurrent or previous implantation failure	2		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.5.3 Unselected/unclear number of prior transfers	3		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.6 Live birth per woman randomised: subgrouping by timing of endometrial injury	7	3038	Odds Ratio (M‐H, Fixed, 95% CI)	1.18 [1.00, 1.38]
1.6.1 Follicular phase prior cycle	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.42 [0.67, 3.01]
1.6.2 Luteal phase prior cycle	6	2809	Odds Ratio (M‐H, Fixed, 95% CI)	1.16 [0.99, 1.37]
1.7 Live birth per woman randomised: subgrouping by intensity of endometrial injury	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.12 [0.98, 1.28]
1.7.1 Moderate intensity	7	4173	Odds Ratio (M‐H, Fixed, 95% CI)	1.11 [0.97, 1.27]
1.7.2 High intensity	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.42 [0.67, 3.01]
1.8 Live birth per woman randomised: subgrouping by timing and intensity	7	3038	Odds Ratio (M‐H, Fixed, 95% CI)	1.18 [1.00, 1.38]
1.8.1 Follicular phase prior cycle and high intensity	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.42 [0.67, 3.01]
1.8.2 Luteal phase prior cycle and moderate intensity	6	2809	Odds Ratio (M‐H, Fixed, 95% CI)	1.16 [0.99, 1.37]
1.9 Miscarriage per woman randomised (studies at low risk of selection bias and other bias)	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	0.88 [0.68, 1.13]
1.10 Miscarriage per woman randomised: sensitivity analysis (no high‐risk)	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.17 [0.38, 3.58]
1.11 Miscarriage per woman randomised: sensitivity analysis (including all studies)	30	8092	Odds Ratio (M‐H, Fixed, 95% CI)	1.03 [0.85, 1.25]
1.12 Clinical pregnancy per woman randomised (studies at low risk of selection bias and other bias)	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.08 [0.95, 1.23]
1.13 Clinical pregnancy per woman randomised: sensitivity analysis (no high‐risk)	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.16 [0.67, 2.02]
1.14 Clinical pregnancy per woman randomised: sensitivity analysis (including all studies)	37		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.15 Clinical pregnancy per woman randomised:subgrouping by control exposure to endometrial manipulation	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.08 [0.95, 1.23]
1.15.1 No control exposure	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.08 [0.95, 1.23]
1.16 Clinical pregnancy per woman randomised:subgrouping by RIF	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.07 [0.94, 1.21]
1.16.1 Not recurrent or previous implantation failure	6	3389	Odds Ratio (M‐H, Fixed, 95% CI)	1.10 [0.95, 1.27]
1.16.2 Recurrent or previous implantation failure	2	533	Odds Ratio (M‐H, Fixed, 95% CI)	0.91 [0.63, 1.31]
1.16.3 Unselected/unclear number of prior transfers	3	480	Odds Ratio (M‐H, Fixed, 95% CI)	1.00 [0.66, 1.51]
1.17 Clinical pregnancy per woman randomised:subgrouping by timing of endometrial injury	7	3038	Odds Ratio (M‐H, Fixed, 95% CI)	1.11 [0.95, 1.29]
1.17.1 Follicular phase prior cycle	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.28 [0.62, 2.62]
1.17.2 Luteal phase prior cycle	6	2809	Odds Ratio (M‐H, Fixed, 95% CI)	1.10 [0.94, 1.29]
1.18 Clinical pregnancy per woman randomised:subgrouping by intensity of endometrial injury	8	4402	Odds Ratio (M‐H, Fixed, 95% CI)	1.08 [0.95, 1.23]
1.18.1 Moderate intensity	7	4173	Odds Ratio (M‐H, Fixed, 95% CI)	1.07 [0.94, 1.22]
1.18.2 High intensity	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.28 [0.62, 2.62]
1.19 Clinical pregnancy per woman randomised:subgrouping by timing and intensity	7	3038	Odds Ratio (M‐H, Fixed, 95% CI)	1.11 [0.95, 1.29]
1.19.1 Follicular phase prior cycle and high intensity	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	1.28 [0.62, 2.62]
1.19.2 Luteal phase prior cycle and moderate intensity	6	2809	Odds Ratio (M‐H, Fixed, 95% CI)	1.10 [0.94, 1.29]
1.20 Multiple pregnancy per woman randomised (studies at low risk of selection bias and other bias)	5	3074	Odds Ratio (M‐H, Fixed, 95% CI)	1.25 [0.80, 1.96]
1.21 Multiple pregnancy per woman randomised: sensitivity analysis (no high risk)	1	229	Odds Ratio (M‐H, Fixed, 95% CI)	0.99 [0.24, 4.06]
1.22 Multiple pregnancy per woman randomised: sensitivity analysis (including all studies)	20	5978	Odds Ratio (M‐H, Fixed, 95% CI)	1.33 [1.05, 1.68]
1.23 Pain (visual analogue scale): sensitivity analysis (including all studies)	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Live birth rate per woman randomised: sensitivity analysis (including all studies)	1	129	Odds Ratio (M‐H, Fixed, 95% CI)	1.28 [0.31, 5.37]
2.2 Miscarriage: sensitivity analysis (including all studies)	1	129	Odds Ratio (M‐H, Fixed, 95% CI)	1.28 [0.31, 5.37]
2.3 Clinical pregnancy per woman randomised: sensitivity analysis (including all studies)	1	129	Odds Ratio (M‐H, Fixed, 95% CI)	1.31 [0.46, 3.73]
2.4 Pain (visual analogue scale): sensitivity analysis (including all studies)	1	129	Mean Difference (IV, Fixed, 95% CI)	1.10 [0.29, 1.91]

*Study characteristics*
Methods	RCT, 2 arms, 100 randomised Setting: Iran, research/clinical centre, one centre Study period: March 2015 to January 2016 (recruitment period unclear)
Participants	Criteria related to previous IVF failure: none Inclusion criteria: women indicated for frozen embryo transfer treatment, had one or more frozen embryo(s) and had a normal uterine cavity (confirmed by vaginal ultrasonography). Exclusion criteria: <40 years (assume error > 40), history of endocrine disorders (hypothyroidism, diabetes mellitus), intrauterine abnormality (uterine polyp, sub‐mucosal fibroma, intrauterine adhesion) and severe endometriosis diagnosed by laparoscopy or endometrioma in ultrasound scanning.
Interventions	Study group: pipelle procedure once between days 21‐23 of the cycle prior to the embryo transfer cycle Control group: no procedure
Outcomes	Reported in paper: ongoing pregnancy, clinical pregnancy, miscarriage, multiple pregnancy Obtained by author correspondence: ‐
Notes	Trial registration: IRCT2015101324512N1 (registered November 2015, retrospective registration) Additional concerns and comments: recruitment rate was high at approximately 14 patients per month. The first author is also the first author of an RCT which has been retracted for methodological issues, which may have been misrepresented (e.g. not a true RCT) (Aflatoonian 2013). The first author has published a large number of trials as first author recently (total of five RCTs as first author in the last 10 years). Additionally, the implantation rates are not consistent with the clinical pregnancy and multiple pregnancy rates; which raises some concern about how the validity of the results. Funding: the financial supporter was Research and Clinical Center for Infertility, Shahid Sadoughi University of Medical Sciences, Yazd, Iran. Author correspondence: Attempted, however no response received Publication: full‐text
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Paper states quote: "a computer‐generated randomization table was created" but also that "93 consecutive subjects" were recruited, therefore unclear whether truly randomised
Allocation concealment (selection bias)	Unclear risk	No description
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only reporting objective outcomes
Incomplete outcome data (attrition bias) All outcomes	Low risk	Appears to be no attrition
Selective reporting (reporting bias)	High risk	Retrospective trial registration and not reporting live birth or adverse events such as pain/bleeding
Other bias	High risk	Trial registered retrospectively. Additionally, the implantation rates are not consistent with the clinical pregnancy and multiple pregnancy rates; which raises some concern about how the validity of the results.

*Study characteristics*
Methods	RCT, 2 arms, 36 randomised Setting: Israel, academic research clinic, one centre Recruitment period: July 2006‐June 2009
Participants	Criteria relating to previous IVF failure: yes, diagnosis of RIF (3 or more unsuccessful cycles of IVF‐ET with good ovarian response in previous cycles) Inclusion criteria: age 18‐41; scheduled for IVF with fresh embryo transfer on the next cycle Exclusion criteria: uterine malformation; presence of endometrioma; ultrasound evidence of hydrosalpinx
Interventions	Study group: pipelle procedure undertaken twice, on days 9–12 and 21–24 of the spontaneous menstrual cycle preceding the IVF treatment cycle Control group: cervical pipelle was done by introducing the biopsy catheter into the cervix without scraping or taking a biopsy specimen, on the same days as above
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage Obtained by author correspondence: ‐
Notes	Trial registration: NCT00411021 (registered Dec 2006, retrospective registration however was initiated post 2010) Additional concerns and comments: none Funding: none stated Author correspondence: attempted but no response received Publication: full‐text
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was performed quote:"using a table of random numbers"
Allocation concealment (selection bias)	Unclear risk	No description
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants were blinded by use of a cervical pipelle procedure.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only reporting objective outcomes
Incomplete outcome data (attrition bias) All outcomes	Low risk	Appears to be no attrition
Selective reporting (reporting bias)	High risk	Trial was registered 6 months into recruitment, however no outcomes listed. Live birth reported, however adverse events such as pain not reported.
Other bias	High risk	On the trial registration page the study authors planned to include 70 women in a cross‐over trial; however in the published study, only 36 women were included and only 1 phase of the study was described. Reasons for the early stop are not stated. Additionally, the trial was registered late.

*Study characteristics*
Methods	RCT, 2 arms, 229 randomised Setting: Denmark, two public fertility clinics (only participants from one included in the analysis) Recruitment period: 2013 ‐ 2018 (months not provided, recruitment period stated as 5 years)
Participants	Criteria related to previous IVF failure: yes, minimum one previous failed cycle Inclusion criteria: women were eligible if they had minimum one previous failed cycle and were planned to undergo their next IVF or ICSI cycle with fresh embryo transfer. Women between 18 and 40 years of age (both ages included) were included. Exclusion criteria: freeze‐all cycles and frozen embryo transfers (FET) were excluded. Further, women were excluded if they had (i) BMI 35, (ii) known intrauterine pathology as cause of infertility, (iii) significant systemic disorders, (iv) ongoing infection (reproductive tract or systemic), (v) intrauterine abnormalities diagnosed during the trial hysteroscopy, or if they became spontaneously pregnant during the trial
Interventions	Study group: hysteroscopy and endometrial biopsy in the follicular phase of the preceding cycle, performed using 7 F forceps Control group: no procedure
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage (defined as loss of a clinical pregnancy before 24 weeks, authors confirmed clinical pregnancy was defined as a positive pregnancy test) Obtained by author correspondence: the authors confirmed they did not measure pain or bleeding following the procedure.
Notes	Trial registration: NCT01743391 (registered Dec 2012, prospective registration) Additional concerns and comments: none Funding: Paper states quote: "No specific funding was sought for this study" Author correspondence: yes, undertaken with kristine.juul.hare.01@regionh.dk and sine.berntsen.01@regionh.dk. Publication: full‐text. This poster was first discovered at the NFOG conference in Odense, 2018 (Poster ID ES27‐0205), however there was no useable data in available in the poster.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The paper states quote: "participants were computer randomised using an online, third‐party randomisation system and assigned...Randomisation was performed as a simple randomisation in a 1:1 ratio without using block randomisation or stratification...One physician was responsible for computer randomising all participants."
Allocation concealment (selection bias)	Low risk	A substantial imbalance in recruitment to each arm was observed (122 versus 107). No description provided in the paper however the authors informed us that quote:"The randomisation was performed in the program SAS. We used a logarithm where a physician (Dr Hare) typed the date of birth of the patient, this together with exact time – date, hour, minute and second, generated a number <1. All <0.5 were in group 1 (ESI group) all >0.5 were controls. All participants were only recruited/randomised once." Although it appears possible that the physician could have randomised women a second time to retrieve a different allocation, the authors also confirmed that the randomisation registered the patients Danish social security number and assigned a number based on the chronology of randomisation ‐ at the end of the study they checked that each patients social security number was only assigned a single randomisation event.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Trial was not blinded quote:"This was a randomised controlled trial (RCT) with no blinding of participants, investigators or health care personnel”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No subjective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	High risk	It is stated that a number of women were excluded post‐randomisation quote:"Freeze‐all cycles and frozen embryo transfers (FET) were excluded. Further, women were excluded if they had... intrauterine abnormalities diagnosed during the trial hysteroscopy, or if they became spontaneously pregnant during the trial“ The authors confirmed that these exclusions are all those documented within Figure 1. In total, there were 14 exclusions in the scratch arm and 3 in the control arm. This imbalance in exclusions may have impacted the analyses. Additionally, data from one of the two included centres was completely omitted due to clinic closure quote:"Data reported and analysed in this article are exclusively from the main centre at Copenhagen University Hospital, Hvidovre as the other fertility clinic (Holbaek) closed down during the study and the patients were lost to follow‐up" and these are the 28 women (15 and 13) described as excluded in Figure 1.
Selective reporting (reporting bias)	Low risk	The trial was registered prospectively, and the listed outcomes are reported.
Other bias	Low risk	The trial was terminated early due to lack of funding, however this is not considered to cause bias

*Study characteristics*
Methods	RCT, 2 arms, 100 randomised Setting: Washington University, St. Louis, USA, one centre Recruitment period: September 2013 ‐ July 2017
Participants	Criteria related to previous IVF failure: no Inclusion criteria: women aged 18 or older undergoing a fresh or frozen embryo transfer. Exclusion criteria: third party reproduction, women undergoing a poor responder protocol, or women with a history of an abnormal uterine cavity
Interventions	Study group: endometrial biopsy with an endometrial pipelle in the luteal phase of the cycle prior to embryo transfer Control group: a sham biopsy (placement of pipelle in posterior fornix, withdraw plunger and “scratch” behind cervix 4 times‐ did not insert pipelle into cervix or uterus) during the luteal phase of the cycle prior to embryo transfer If patients were on oral contraceptive pills for their IVF cycle, the procedure was scheduled anytime during the last 7 days or up until 1 day after their pills were discontinued of the cycle prior to embryo transfer. If patients were not on oral contraceptive pills, patients were instructed to check for a luteinising hormone surge and the procedure was scheduled for 7–13 days following in the cycle prior to embryo transfer.
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage Obtained by author correspondence: ‐
Notes	Trial registration: authors confirmed the study was not registered Additional concerns and comments: none Funding: NIH funding (5T32HD055172‐09, UL1 TR002345) Author correspondence: yes, with Ashley Eskew (eskewa@wustl.edu) We noticed that the numbers of pregnancies and associated odds ratios were inconsistent in the text of the paper compared to the outcomes table. After contacting the authors we were reassured that the results in the outcomes table were correct; and the authors are contacting the journal to arrange a revision. Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote:"randomization was performed using a computer‐generated model with varied block lengths"
Allocation concealment (selection bias)	Low risk	Women were randomised using quote:"consecutively numbered sealed opaque envelopes"
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Sham procedure implemented quote:"Both the patient and the clinician performing the embryo transfer were blinded to the randomization arm"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only reporting objective outcomes
Incomplete outcome data (attrition bias) All outcomes	Low risk	No attrition: quote:"No patients were lost to follow up"
Selective reporting (reporting bias)	High risk	Authors confirmed trial was not registered, and adverse outcomes such as pain were not reported
Other bias	High risk	Trial stopped early for futility, only after observing challenges with recruitment. Stopping on the basis of O'Brien and Fleming 1979 rule means that the data available for meta‐analysis give a biased effect estimate. Additionally the trial not registered.

*Study characteristics*
Methods	RCT, 2 arms, 191 randomised Setting: France, three centres Recruitment period: February 2010 ‐ July 2014
Participants	Criteria related to previous IVF failure: yes, first or second IVF attempt Inclusion criteria: 18‐38 years of age; primary or secondary infertility; regular menstrual cycles; follicle‐stimulating hormone (FSH) ≤ 12 IU/L; signed informed consent Exclusion criteria: had participated to in oocyte donation program, presented a BMI > 35, had hydrosalpinx, uterine deformations, uterine fibroids (≥4 and the largest > 5 cm), abnormal gynaecological bleeding of unknown origin, or ongoing vaginal infection, had been pre‐treated with oestrogen–progesterone or estradiol
Interventions	Study group: the biopsy is realised with a Pipelle de Cornier, moving the pipelle in and out while twisting, twisting the pipelle to cover an angle of 360° and making several "in and out" cycles to collect a complete sample of the endometrium. Performed between Day 20 and Day 24 of the cycle preceding ovarian stimulation Control group: no intervention
Outcomes	Reported in paper: ongoing pregnancy, clinical pregnancy, miscarriage, multiple pregnancy, pain, bleeding Obtained by author correspondence: ‐
Notes	Trial registration: NCT01064193 (registered Feb 2010, registered prospectively) Additional concerns and comments: none. Funding: Ministère de la Santé Français (Programme Hospitalier de Recherche Clinique 2009) Author correspondence: yes, brief correspondence with sandrine.frantz@chu‐bordeaux.fr Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Qquote:"The randomization sequence was generated using SAS Software...with a 1:1 allocation using random block sizes of 4 and 6" therefore adequate sequence generation
Allocation concealment (selection bias)	Low risk	Quote:"Randomization was...performed using a centralized web‐based service"
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was used. Additionally, there were a substantial number of women in the intervention arm who did not undergo the endometrial scratch (n = 25 of 98).
Blinding of outcome assessment (detection bias) All outcomes	High risk	The outcome of pain and adverse events were self‐reported by participants who were not blind to their treatment allocation
Incomplete outcome data (attrition bias) All outcomes	Low risk	Appears only one women was lost to follow‐up
Selective reporting (reporting bias)	Low risk	Trial was registered prospectively and all outcomes reported
Other bias	High risk	The study was quote:"stopped prematurely in July 2014 after an unplanned interim analysis" and "This analysis was prompted by the tendency towards lower pregnancy rates observed in the ES arm"

*Study characteristics*
Methods	RCT, 2 arms, 387 women randomised Setting: Egypt, academic and private clinics, 3 centres Recruitment period: January 2011 ‐ June 2012 (provided by author correspondence)
Participants	Criteria relating to previous IVF failure: yes, at least one previous IVF failure Inclusion criteria: younger than 40 years of age with Exclusion criteria: women who were described as poor responders after previous IVF treatment (produced fewer than 3 oocytes or had their cycles cancelled because of poor follicular growth); women with known endocrinopathy; women undergoing tubal disconnection for hydrosalpinx; history of endometrial curettage within 3 months of the study; fibroids and other uterine factors (polyps, adhesions)
Interventions	Study group: pipelle procedure twice between days 21 and 26 of the cycle before the IVF index cycle and after initiation of the GnRHa in long agonist protocols Control group: placebo procedure using the uterine sound inserted into the cervix until the internal os on the same days of the cycle, as in women in the intervention group When the pipelle or the sound could not be introduced, a hysteroscopy was performed at the second appointment, in both groups
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage, multiple pregnancy Obtained by author correspondence: ‐
Notes	Trial registration: NCT01245309 (registered Nov 2010, registered prospectively) Additional concerns and comments: recruitment rate is approximately 22 participants per month, however across three centres this is not deemed to be particularly high. Funding: university funding Author correspondence: yes, Ahmed Gibreel is a co‐author of this Cochrane Review and the included trial Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Paper states quote:"Randomization was performed through a computer generated tables of random numbers"
Allocation concealment (selection bias)	High risk	Quote:"Women were asked to pick an opaque sealed envelope on the day of start of pituitary down‐regulation" as the envelopes were not numbered this is graded as high risk
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Paper states quote:"Women were blinded to their allocation in the trial while physicians were not blinded". Women in the control arm underwent an endometrial sound procedure.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only reporting objective outcomes
Incomplete outcome data (attrition bias) All outcomes	Low risk	5 dropouts in total, 3 in one arm and 2 in the other arm
Selective reporting (reporting bias)	Low risk	Trial was registered prospectively and all outcomes reported
Other bias	Low risk	‐

*Study characteristics*
Methods	RCT, 2 arms, 305 randomised Setting: Turkey, private fertility centre, one centre Recruitment period: May 2015 ‐ July 2017 (confirmed by author correspondence)
Participants	Criteria related to previous IVF failure: yes, the failure to achieve a clinical pregnancy after the transfer of at least four good‐quality embryos in a minimum of three fresh or frozen cycles to a woman under the age of 40 years Inclusion criteria: under the age of 40 and follicle‐stimulating hormone (FSH) levels ≤15 IU/mL, and meeting the above RIF condition Exclusion criteria: congenital uterine anomalies, patients with Asherman’s syndrome, patients with uterine cavity distorted by myoma or endometrial polyps, patients with confirmed endometriosis or endometrioma and patients with BMI of <18.5 and >29.9, endometrium thickness of less than 7 mm in the cycle before the cycle
Interventions	Study group: Endometrial injury on the 10‐12^th day of late follicular phase in the preceding cycle through office hysteroscopy. Endometrial injury was performed without energy modality (i.e. with scissors). Endometrial injury was performed first on the fundus by cutting into the endometrium (without injuring the myometrium) transversally. Later, three or four vertical incisions were performed 0.5 cm apart each other, on the anterior and posterior walls of the uterus, 1‐1.5 cm away from the fundus and one cut for each lateral wall. Also: standard gynaecological surgical procedures were used to treat recognized pathology Control group: no procedure
Outcomes	Reported in paper: live birth, clinical pregnancy, multiple pregnancy, miscarriage Obtained by author correspondence: ‐ Authors confirmed pain and bleeding were not collected.
Notes	Trial registration: NCT03748238 (registered Nov 2018, registered retrospectively) Additional concerns and comments: one of the trial authors (Makrigiannakis) is the first author of a retracted study, which appears to have been retracted for duplication (Makrigiannakis 2010); the implication of this is unclear. Funding: authors confirmed the study did not receive any specific funding Author correspondence yes, undertaken with muberranamli@hotmail.com Publication: full‐text. A conference abstract was initially identified and the authors provided a report of the trial pending the full publication, and the full‐text paper has now been published.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote:"randomized using a computer‐generated random number sequence (1:1 simple randomization)"
Allocation concealment (selection bias)	Unclear risk	No description. Correspondence undertaken with the authors and it appears that no allocation concealment was in place quote: "Our computer program was very basic. Our numbers were given by turns 0 and 1. So it wasn't hidden and it was possible to see the next allocation if we check computer at that time"
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	High risk	Three women in the injury group and 9 women in the control group were excluded for refusing their allocation. There was also an imbalance in the number of women with cancelled or failed IVF cycles: 13 in the injury group and 28 in the control group ‐ which could lead to a higher pregnancy rate in the injury group under intention‐to‐treat. Additionally, 13 women in the injury group were excluded post‐randomisation if uterine pathology was detected during the hysteroscopy quote:"Standard gynaecological surgical procedures were used to treat recognised pathology and any patients with surgical intervention other that endometrial injury were excluded from the study."
Selective reporting (reporting bias)	High risk	Registered retrospectively and only listing biochemical pregnancy as an outcome. Live birth reported, however adverse events such as pain not reported. The manuscript states "No hysteroscopy‐related adverse events were reported" but it is unclear whether they measured pain and bleeding.
Other bias	High risk	Trial registered retrospectively

*Study characteristics*
Methods	RCT, 2 arms, 124 randomised Setting: Turkey, academic clinic, one centre Recruitment period: September 2010 ‐ April 2011
Participants	Criteria related to previous IVF failure: yes, women were undergoing their first IVF cycle (as per author correspondence) Inclusion criteria: age < 35; history of primary infertility; normal responder (antral follicle count of 5 to 10 in 1 ovary at early follicular phase); grade I or II embryos for transfer Exclusion criteria: endocrinopathies; any systemic disease; history of neoplasm; high risk for or history of ovarian hyperstimulation syndrome; use of any concurrent medication; failure to proceed to follicle retrieval; severe male infertility requiring testicular sperm aspiration; Mullerian tract anomalies; history of endometrial instrumentation or surgery within 1 month of the study; fibroids and other uterine factors (polyps, adhesions); lack of agreement to undergo endometrial biopsy during the stimulation cycle
Interventions	Study group: endometrial injury was performed on day 3 of the menstrual cycle following down regulation; the scratching was done in 2 defined (anterior and posterior) portions of the uterine cavity under sterile conditions with the use of a biopsy catheter (Gynetics 4164 Probet Pipella, HD Aksu Medical, Ankara, Turkey) Control group: n o intervention
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage Obtained by author correspondence: ‐
Notes	Trial registration: NCT01851876 (registered May 2013, registered retrospectively) Additional concerns and comments: trial recruited 124 women in 8 months from 1 centre (16 per month), which is a high recruitment rate for such a short recruiting window. Funding: unclear Author correspondence: yes, undertaken with the first author drsuleymanguven@yahoo.com Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Trial registration describes study as non‐randomised. Described in paper as quote: "prospective case–control study" and also as an RCT "Women were allocated at random (sealed envelopes) to the intervention group or the control group" author correspondence confirmed the trial was an RCT with computer‐assisted randomisation
Allocation concealment (selection bias)	Unclear risk	Sealed envelopes were used, unclear if SNOSE
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	High risk	The eligibility criteria include having grade I or II embryos to transfer and no 'failure to proceed to follicle retrieval' which are post‐randomisation exclusions as the women in the intervention arm underwent endometrial injury prior to follicle retrieval. It is unclear how many women were excluded for these reasons.
Selective reporting (reporting bias)	High risk	Trial registered retrospectively and adverse outcomes such as pain not reported
Other bias	High risk	Trial registered retrospectively.

*Study characteristics*
Methods	RCT, 2 arms, 51 participants Setting: Canada, three centres Recruitment period: May 2013 to May 2015
Participants	Criteria related to previous IVF failure: yes, first or second IVF cycle Inclusion criteria: with or without ICSI; age 18–39 years; BMI 18–35 kg/m2 ; evaluation of uterine cavity (hysterosalpingogram, sonohysterogram, hysteroscopy) performed in the preceding 24 months; early follicular phase (day 2 or 3) serum FSH evaluated in the preceding 6 months; use of a long gonadotropin‐releasing hormone agonist or antagonist protocol; and documented LH surge 9–11 days before enrolment for patients not pretreated with the oral contraceptive pill or use of the pill for ≥ 10 days at the time of enrolment. Exclusion criteria: previously enrolled in this study; had prior early follicular phase follicle‐stimulating hormone level ≥ 12 IU/L; previous poor ovarian response (defined as prior IVF cycle cancelled for poor response or≤ 4 oocytes retrieved); IVF for preimplantation genetic diagnosis or fertility preservation, diabetes mellitus or uncontrolled thyroid disease; abnormal uterine cavity; untreated hydrosalpinx; any contraindication to endometrial biopsy, or if they had office hysteroscopy or other uterine procedure planned or performed during the cycle preceding IVF stimulation; or planned on using surgically retrieved sperm.
Interventions	Study group: a single endometrial biopsy performed 5‐10 days prior to the start of gonadotropins in a standard IVF cycle Control group: no intervention
Outcomes	Reported in paper: [ive birth, clinical pregnancy, miscarriage Obtained by author correspondence: ‐
Notes	Trial registration: NCT01983423 (registered Nov 2013, registered retrospectively by only 6 months) Additional concerns and comments: none Funding: Unrestricted Educational Grant from Ferring Inc. (Canada) Author correspondence: Undertaken with JHavelock@pacificfertility.ca and Kimberly.Liu@sinaihealthsystem.ca Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Authors described quote: "The SAS System for Windows was used generate randomization numbers that were accessed electronically using an encrypted web‐based randomization system"
Allocation concealment (selection bias)	Low risk	As per the above description, allocation was concealed until the point of randomisation
Blinding of participants and personnel (performance bias) All outcomes	High risk	Study described as quote:"open‐label"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Appears to be no attrition, this was confirmed by study authors
Selective reporting (reporting bias)	Low risk	Trial registered six months after initiating recruitment and all listed outcomes reported
Other bias	Low risk	Trial stopped early when only 51/332 intended participants recruited. Recruitment was closed due to difficulty recruiting, which is not itself considered a risk of bias.

*Study characteristics*
Methods	RCT, 2 arms, 59 randomised Setting: South Korea, Maria Fertility Hospital, one centre Recruitment period: no description
Participants	Criteria related to previous IVF failure: yes, first IVF cycle only Inclusion criteria: all patients were under 35 years old, day 3 follicle‐stimulating hormone level were below 10 IU/L. Exclusion criteria: infertility cause of uterine factor, severe male factor, and severe endometriosis were excluded.
Interventions	Study group: endometrial biopsy was done with a biopsy catheter (pipelle de cornier, France), performed only one time on starting day of stimulation. Control group: no procedure
Outcomes	Reported in paper: clinical pregnancy Obtained by author correspondence: ‐
Notes	Trial registration: authors confirmed trial was not registered Additional concerns and comments: none Funding: no description Author correspondence: yes, correspondence undertaken with Dr Yeonhee Ka (miriuh@daum.net) and Dr Hur (cyhur68@gmail.com). Authors provided the poster that was presented at the conference. Publication: abstract only (and poster)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Authors confirmed quote:"patient allocation was made by computerized randomization"
Allocation concealment (selection bias)	Unclear risk	No description
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was implemented
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only reported objective outcomes
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Only abstract available therefore difficult to assess level of attrition
Selective reporting (reporting bias)	High risk	Trial was not registered and important outcomes such as live birth and adverse events not reported
Other bias	High risk	Trial published as abstract only with limited detail, and further information not obtained from author correspondence. Trial not registered.

*Study characteristics*
Methods	RCT, 2 arms, 100 randomised Setting: academic setting, Turkey Study period: January 2008‐March 2009 (unclear if recruitment period or entire study period)
Participants	Criteria related to previous IVF failure: yes, failed to conceive during 1 or more cycles of IVF and embryo transfer (ET) Inclusion criteria: women considered to be good responders to hormonal stimulation; age between 25 and 36 years Exclusion criteria: hydrosalpinx; thrombophilia; submucous myoma and factors found to have a negative impact on implantation
Interventions	Study group: 2 consecutive endometrial biopsies at 1‐week intervals during the luteal phase of the non‐transfer cycle, when on Gonadotropin‐releasing hormone agonist for downregulation. Endometrial biopsy was performed with a biopsy catheter (Pipelle de Cornier, Prodimed, Neuilly‐en‐Thelle, France) introduced through the cervical os and rotated within the uterine cavity 3‐4 times after withdrawal of the piston. Antibiotics were administered after the procedure Control group: no intervention
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage Obtained from author correspondence: ‐
Notes	Trial registration: believed to be: NCT01340560 (registered Apr 2011, registered retrospectively) Additional concerns and comments: Implantation and clinical pregnancy rates appear inconsistent; similar implantation rates reported in each arm (31% versus 35%) however clinical pregnancy rates significantly different (34% versus 60%); no obvious explanation. Funding: quote:"This study has no financial support" Author correspondence: yes, undertaken Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Paper states quote:"The randomization was based on a computer generated random numbers"
Allocation concealment (selection bias)	Unclear risk	No description provided
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Appears to be no attrition
Selective reporting (reporting bias)	High risk	Unclear if registered, and important outcomes such as adverse events not reported
Other bias	High risk	Trial registered retrospectively. There is an apparent inconsistency in the implantation and pregnancy rates reported, which are similar per arm for implantation rate but significantly different for clinical pregnancy rates.

*Study characteristics*
Methods	RCT, 2 arms, 352 randomised Setting: Spain, one private IVF centre (ProcreaTec Fertility Clinic) Recruitment period: January 2017 to October 2018
Participants	Criteria relating to previous IVF failure: no Inclusion criteria: women undergoing egg donation, Included patients’ age ranged between 18 and 50 years, and all had normal uterine cavity, assessed by 2D transvaginal ultrasound. Patients with endometrial polyps were only included if polypectomy was performed at least 2 months before the treatment cycle. Exclusion criteria: severe male factor (less than 2 million sperm per mL) or if they presented any factor interfering with embryo implantation, such as uterine abnormalities (uterine fibroids classified as 0–2 FIGO stage, Müllerian malformations, or severe adenomyosis) or unilateral or bilateral hydrosalpinx; BMI over 35 kg/m2, previous endometrial scratch or hysteroscopy (less than a month before the randomisation)
Interventions	Study group: Scratch performed 5 to 10 days before their period started and the endometrial preparation began. The scratch was carried out in an out‐patient setting under abdominal ultrasound guidance. A speculum was inserted into the vagina and after cervix disinfection with an iodine solution, an endometrial biopsy catheter (Pipelle de Cornier,Laboratoire CCD, France) was inserted through the cervix into the uterine cavity. Once in the cavity, the catheter piston was partially removed to create a suction effect and the catheter was then moved back and forth and rotated 360° in order to scratch the four walls. Control group: no procedure
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage, multiple pregnancy Obtained via author correspondence: the authors provided data subgrouped by women with/without recurrent implantation failure, which we have used in the subgroup analyses. The authors stated that quote:"3 patients were scratch was difficult, 6 patients who referred some mild pain and 9 patients that suffered from some spotting some days after the technique and before their period started" however that "there was not a specific questionnaire or scale to report pain or bleedings and not all patients were specifically asked just after the scratch. Since some data were obtained after the treatment, it was possible that this information was not 100% accurate, so that is why it was not included in our paper" and also we have not included it in this review for the same reasons.The authors also provided the results subgrouped by RIF nonRIF.
Notes	Trial registration: NCT03108157 (registered Apr 2017, registered 3 months retrospectively) Additional concerns and comments: recruitment rate is approximately 16 participants per month, which may be considered high for a trial recruiting over a period of only 22 months, however cannot on its own be considered a serious risk of bias. The authors suggested the recruitment rate may be higher than other studies owing to the wider inclusion criteria used in the study. The authors provided the live birth data among women with recurrent implantation failure, and the overall live birth rate in this group was 57% (81/143) which is relatively high for a recurrent implantation failure population, and was also higher than the women in the study without implantation failure, however the women in this study were receiving donated eggs. Funding: this work was fully supported by ProcreaTec Fertility Center. This research received no specific grant from any funding agency in the public, commercial, or not‐for‐profit sectors Author correspondence: yes, with Alexandra Izquierdo, izquierdo.alexandra@yahoo.es Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The paper states that quote:“The randomization chart was obtained by a web‐based randomization program (randomization.com) using simple randomization. Patients were allocated to a treatment arm (group A—intervention arm, group B—control arm) and then received the instructions for their treatment protocol.”
Allocation concealment (selection bias)	High risk	There is no description of the randomisation process, however the authors confirmed that they quote:"used the software to obtain a long list of allocations and patients were included in consecutive order. If patients accepted the study, they were assigned to each group at the moment they accepted the treatment and dates, according to the list" and specifically that "we could see the next allocation" therefore allocations were not concealed
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was used as the paper states quote:Blinding was not possible since patients in the study group received an intervention and those in control group did not (no placebo intervention was performed).”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No subjective outcomes reported.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Although in the paper it was not clear, we confirmed with the authors that there were no women lost to follow‐up or withdrawn from the trial. The 5 women that appear to have been lost in Figure 1 had missing data for pregnancy complications and newborn information only.
Selective reporting (reporting bias)	Low risk	Trial was registered late, but only by 3 months, and all outcomes were collected. therefore rated as low‐risk
Other bias	Low risk	‐

*Study characteristics*
Methods	RCT, 2 arms, 156 randomised Setting: Iran in an academic setting Recruitment period: June 2008 to January 2009
Participants	Criteria related to previous IVF failure: yes, participants were undergoing their first IVF cycle Inclusion criteria: age < 38 years; BMI > 19 kg/m² and < 30 kg/m²; day 3 follicle‐stimulating hormone < 12 mIU/ mL; triple‐layer endometrium with thickness > 8 mm on the day of human chorionic gonadotrophin administration; normal ovarian response to stimulation (estradiol on the day of trigger between 500 and 3000 pg/mL and between 4 and 14 retrieved oocytes Exclusion criteria: any uterine anomaly such as myoma and endometrial polyp on USTV; endometrioma with a diameter > 3 cm; visible hydrosalpinges
Interventions	Study group: one endometrial injury procedure using Novak curette on the day of oocyte retrieval Control group: no intervention
Outcomes	Reported in paper: ongoing pregnancy, clinical pregnancy. Miscarriage calculated as the difference between clinical and ongoing pregnancies. Obtained by author correspondence: ‐
Notes	Trial registration: NCT00846183 (registered Feb 2009, registered retrospectively) Additional concerns and comments: over a recruitment period of 8 months the researchers recruited 156 women (19.5 per month). This is a high recruitment rate for one centre. Three of the four authors are also authors of an RCT which has been retracted for methodological issues, which may have been misrepresented (e.g. not a true RCT) (Aflatoonian 2013). Funding: no description Author correspondence: attempted but no response received Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "computer generated randomization method"
Allocation concealment (selection bias)	Unclear risk	No description
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Four women were excluded from the analysis because they had bleeding on the day of planned embryo transfer and therefore embryo transfer was cancelled. It is a reasonable assumption that these four women did not conceive.
Selective reporting (reporting bias)	High risk	Trial was registered retrospectively and live birth not reported
Other bias	High risk	Quote:"because of significant lower pregnancy rates in the experimental group, the study was stopped sooner" it is not clear whether adequate stopping rules were used. Trial registered retrospectively. Three authors are also co‐authors of a retracted paper.

*Study characteristics*
Methods	RCT, 2 arms, 160 randomised Setting: Iran in an academic setting Recruitment period: not described
Participants	Criteria related to previous IVF failure: yes, at least 2 prior implantation failures Inclusion criteria: at least 2 prior implantation failures Exclusion criteria: ‐
Interventions	Study group: Novak endometrial suction curettage during the secretory phase in a non‐medicated cycle before IVF/ICSI Control group: no intervention
Outcomes	Reported in paper: Clinical pregnancy Obtained by author correspondence: ‐
Notes	Trial registration: does not appear to be registered Additional concerns and comments: one of the authors is an author on a paper which has been retracted for duplication, as it was very similar to another paper by the same author group in a different journal (Mohsenzadeh 2018). Funding: no description Author correspondence: ‐ Publication: abstract only
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of randomisation was not described
Allocation concealment (selection bias)	Unclear risk	No description
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	It is unclear whether there was an attrition or dropout
Selective reporting (reporting bias)	High risk	Trial not registered and important outcomes such as live birth and adverse events not reported
Other bias	High risk	Published only as a conference abstract with limited detail; no further information could be retrieved from study authors. Trial not registered. Authors are also co‐authors of a retracted paper.

PERMALINK

Endometrial injury in women undergoing in vitro fertilisation (IVF)

Sarah F Lensen

Sarah Armstrong

Ahmed Gibreel

Carolina O Nastri

Nick Raine-Fenning

Wellington P Martins

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

Effectiveness

Adverse events

Secondary outcomes

Effectiveness

Adverse events

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

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

Summary of findings and assessment of the certainty of the evidence

Results

Description of studies

Results of the search

1.

Included studies

Study design and setting

Participants

Interventions

Outcomes

Excluded studies

Ongoing studies

Studies awaiting classification

Risk of bias in included studies

2.

Allocation

Random sequence generation

Allocation concealment

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Summary of findings 1. Endometrial injury compared to control in women undergoing assisted reproductive techniques.

Summary of findings 2. Higher compared to lower degree of injury in women undergoing assisted reproductive techniques.

1. Endometrial injury versus control

Primary outcome

1.1 Live birth per woman randomised

1.1. Analysis.

*Study characteristics*
Methods	RCT, 2 arms, 115 randomised Setting: Iran in an academic setting Recruitment period: Unclear
Participants	Criteria related to previous IVF failure: yes, recurrent implantation failure (defined as 2 to 6 unsuccessful cycles of IVF‐embryo transfer with previous transfer of at least 10 high‐grade embryos without achievement of clinical pregnancy) Inclusion criteria: age between 20 and 40 years; no history of blood disease; Exclusion criteria: age > 40 years; poor response in previous cycles (defined as day 3 follicle‐stimulating hormone > 10 IU mL or < 4 follicles on the day of trigger in previous cycle); uterine malformation; presence of endometrioma; ultrasound evidence of hydrosalpinx
Interventions	Study group: 1 endometrial injury procedure using Pipelle de Cornier on days 21‐26 of spontaneous cycle Control group: no intervention
Outcomes	Reported in paper: clinical pregnancy Obtained by author correspondence: ‐
Notes	Trial registration: does not appear to be registered Additional concerns and comments: one of the authors is an author on a paper which has been retracted for duplication, as it was very similar to another paper by the same author group in a different journal (Mohsenzadeh 2018). Funding: Research and Clinical Center for Infertility, Shahid Sadoughi University of Medical Sciences. Author correspondence: attempted but no response received Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Random selection to either group was performed by drawing a piece of paper from the bag containing equal number of printed paper for each method"
Allocation concealment (selection bias)	High risk	No description of any safeguards in place to ensure allocation concealment and prevent someone from replacing the paper and selecting another
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Nine women lost to follow‐up in each arm (4 and 5)
Selective reporting (reporting bias)	High risk	Trial does not appear to have been registered and important outcomes such as live birth and adverse events not reported
Other bias	High risk	Trial not registered. Authors are also co‐authors of a retracted paper.

*Study characteristics*
Methods	RCT, 2 arms, 1364 randomised Setting: 13 academic/public hospital centres across five countries (New Zealand, Australia, UK, Belgium, Sweden) Recruitment period: June 2014 ‐ June 2017
Participants	Criteria related to previous IVF failure: no Inclusion criteria: women planning IVF with their own oocytes (stimulated IVF cycle with planned fresh transfer, or frozen embryo transfer using stored embryos) Exclusion criteria: not planning an embryo transfer (e.g. fertility preservation or planned freeze‐all cycles), had any contraindication to pipelle biopsy (e.g. vaginismus), or had undergone any disruptive intrauterine procedures within three months prior to commencing IVF, specifically: hysteroscopy, sonohysterogram, hysterosalpingogram, laparoscopy, surgically managed miscarriage or endometrial biopsy
Interventions	Intervention: single endometrial pipelle biopsy performed between day 3 of the menstrual cycle preceding the embryo transfer cycle and day 3 of the menstrual cycle for which embryo transfer is planned Control: no procedure
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage, multiple pregnancy, pain, bleeding Obtained by author correspondence: ‐
Notes	Trial registration: ACTRN12614000626662 (registered June 2014, registered prospectively) Additional concerns and comments: none Funding: University of Auckland, New Zealand; the A+ Trust, Auckland District Health Board, New Zealand; the Nurture Foundation, New Zealand; and the Maurice & Phyllis Paykel Trust, New Zealand Author correspondence: yes, one of the investigators is an author on this Cochrane Review (s.lensen@auckland.ac.nz) Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "An online, third‐party, data collection and randomisation system was used... participant allocations were concealed within the system until the patient was randomised. Participants were randomised in a 1:1 ratio using block randomisation of two different sizes between 6 and 16 repeating in random order, stratified by recruiting site and by whether a fresh or frozen embryo transfer was planned"
Allocation concealment (selection bias)	Low risk	As above
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was used. A higher proportion of women in the endometrial injury arm underwent embryo transfer compared to women in the control arm; however there was no impact on the results after adjusting for this observation
Blinding of outcome assessment (detection bias) All outcomes	High risk	The outcome of pain during the procedure and bleeding were self‐reported by participants who were not blind to their treatment allocation
Incomplete outcome data (attrition bias) All outcomes	Low risk	Minimal attrition: four women withdrew from the trial and the pregnancy outcome of one further women is unknown
Selective reporting (reporting bias)	Low risk	Trial was registered prospectively and the protocol published; all outcomes reported
Other bias	Low risk	‐

*Study characteristics*
Methods	RCT, 4 arms, 142 randomised Setting: Beijing Obstetrics and Gynecology Hospital, China ‐ one centre Recruitment period: February 2012 to November 2014
Participants	Criteria related to previous IVF failure: yes, patients were undergoing first IVF cycle Inclusion criteria: infertile women indicated for IVF treatment; ≤40 years of age; a normal uterine cavity demonstrated by saline infusion sonogram; basal follicle stimulating hormone (bFSH) <12 IU/L Exclusion criteria: endometrium with polyp or fibroid; hydrosalpinx; endometriosis.
Interventions	Study group: for patients in proliferative phase group, endometrial injury was performed between cycle day 10–12. For patients in luteal phase group, endometrial injury was performed 7–9 days after ovulation. The endometrial injury procedure was performed in a standard approach using a pipelle Control group: pipelle catheter inserted through the cervix but no injury was performed to the endometrium (on either cycle days 10‐12 or 7‐9 after ovulation)
Outcomes	Reported in paper: live birth, clinical pregnancy, multiple pregnancy, miscarriage (provided in paper but numbers do not match the difference between live birth and clinical pregnancy, therefore miscarriage calculated based on this difference instead), pain, bleeding Obtained by author correspondence: ‐
Notes	Trial registration: ChiCTR‐IOR‐17011506 (registered May 2017, registered retrospectively) Additional concerns and comments: the outcomes of all clinical pregnancies are not known, for example there are 29 clinical pregnancies in the intervention group and 1 miscarriage, 1 ectopic and 25 live births; there is no description of the fate of the other two pregnancies. Additionally, there are two ectopics in total reported in Table 2 but only one in Table 4 ‐ however only small numbers are involved. Funding: the study was supported by the National Natural Science Foundation of China (81471520), Beijing Natural Science Foundation Project (5122015), and Project Training High‐Level Medical Technical Personnel in the Health System in Beijing (2014–3‐075). Author correspondence: attempted but no response received (yingliubj@hotmail.com) Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "table of random numbers" was used
Allocation concealment (selection bias)	Unclear risk	No description
Blinding of participants and personnel (performance bias) All outcomes	Low risk	A sham procedure was implemented which is considered likely to blind participants
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Subjective outcome of pain reported and participants were blind to their trial allocation
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	There appears to be no attrition, however the paper states that all randomised women underwent embryo transfer, which seems unlikely for all 142 women. It is possible that some women were excluded.
Selective reporting (reporting bias)	Low risk	Retrospective registration however live birth and adverse events reported
Other bias	High risk	Trial registered retrospectively. Additionally, the outcome of some pregnancies are not accounted for which questions the accuracy of the data.

*Study characteristics*
Methods	RCT, 2 arms, 200 randomised Setting: Belgium, one University Hospital. Recruitment period: April 2014 ‐ October 2017 (confirmed with authors)
Participants	Criteria related to previous IVF failure: no Inclusion criteria: 18‐40 years of age; fresh IVF/ICSI cycle; antagonist downregulation; signed informed consent Exclusion criteria: other known reasons for impaired implantation (i.e. hydrosalpinx, fibroid distorting the endometrial cavity, Asherman’s syndrome, thrombophilia or endometrial tuberculosis); oocyte donation acceptors; frozen egg transfers; embryos planned to undergo embryo biopsy; BMI >35 or <18; women already recruited for another trial on medically assisted procreation during the same cycle; women who have previously enrolled in the trial; those unable to comprehend the investigational nature of the proposed study
Interventions	Study group: endometrial biopsy was performed on Days 6 to 8 of ovarian stimulation with a Pipelle de Cornier® (Laboratoire CCD, France). The device was introduced into the uterus until slight resistance from the fundus was felt after which the piston was withdrawn and the device rotated through 360◦ as it was moved up and down for four times Control group: no intervention
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage, bleeding, pain Obtained by author correspondence: ‐
Notes	Trial registration: NCT02061228 (registered Feb 2014, registered prospectively) Additional concerns and comments: the trial authors conduct a large number of RCTs run at this single academic centre. However, all RCTs were found to be staggered and conducted in non‐overlapping populations. Funding: ‘Fonds Wetenschappelijk Onderzoek’ (FWO, Flanders, Belgium, 11M9415N, 1524417N). Author correspondence: yes, with Shari Mackens Shari.Mackens@uzbrussel.be and Samuel dos Santos Ribeiro, samueldsribeiro@gmail.com Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	As per paper quote: "randomization sequence and allocation was appointed using a computer‐generated randomization list with a 1:1 allocation”
Allocation concealment (selection bias)	Low risk	Quote:“concealment was ensured with sequentially numbered, opaque, sealed envelopes”
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding
Blinding of outcome assessment (detection bias) All outcomes	High risk	No blinding and subjective outcomes reported: pain and bleeding
Incomplete outcome data (attrition bias) All outcomes	Low risk	Only 1 women withdrew in the control group
Selective reporting (reporting bias)	Low risk	Trial was registered prospectively and reports all review outcomes originally planned. Some additional outcomes such as premature birth and low birth weight not reported, but not review outcomes
Other bias	High risk	The quote:"trial was terminated prematurely following the second interim analysis after the recruitment of 200 patients due to safety concerns (specifically, a potentially increased risk of miscarriage in the intervention arm)" and there does not appear to be any statistical adjustment made for these multiple interim analyses

*Study characteristics*
Methods	RCT, 2 arms, 300 randomised Setting: Kasr Al Ainy IVF unit in Cairo, Egypt ‐ one centre Study period: 1 January 2016 to 31 March 2017 (unclear if this is recruitment period)
Participants	Criteria related to previous IVF failure: yes, first time ICSI Inclusion criteria: candidates for ICSI, were to undergo ICSI for the first time, and met the following inclusion criteria: aged younger than 40 years, the follicle‐stimulating hormone (FSH) level measured on the third day of a natural cycle was less than 10 IU/L, the serum prolactin level was normal, and uterine cavity abnormality was excluded by hysteroscopy or hysterosalpingography. Exclusion criteria: abnormal endocrine function (e.g. abnormal thyroid or adrenal function), ovarian cysts, hydrosalpinx, endometrial polyps, male partner with azoospermia, and ICSI performed for preimplantation genetic diagnosis
Interventions	Study group: endometrial injury was induced by performing endometrial aspiration in the mid luteal phase of the cycle immediately preceding the scheduled IVF treatment, using a Pipelle catheter Control group: no procedure
Outcomes	Reported in paper: clinical pregnancy, miscarriage, multiple pregnancy Obtained by author correspondence: ‐ Although the authors report clinical pregnancies and miscarriages before 12 weeks, it was not appropriate to calculate Ongoing pregnancy as the difference between these two numbers as it was unclear whether all participants 12 week status was confirmed as either Ongoing or not.
Notes	Trial registration: NCT02660125 (registered 17 Jan 2016, appears to be registered in the same month as recruitment initiated however recruitment period unclear) Additional concerns and comments: the recruitment rate is high; the study is stated to have been conducted within 15 months (Jan 2016‐March 2017) and was submitted to the journal in May 2017. Women were followed to the stage of clinical pregnancy (6‐8 weeks following endometrial scratch), therefore 300 women were recruited within 13 months = 20‐23 women per month. It appears the trial may have been registered a few weeks late. Authors were contacted to request ethics approval letter and were unable to produce this. We wrote to the ethics office and received no response. The first author has co‐authored a total of 20 RCTs within 10 years, 17 of which as first author. This might be considered an impressive and possibly improbable rate of RCT publications. The study reports an unusually high multiple pregnancy rate; the average number of embryos replaced was approximately 1.4 per woman (428 embryos transferred to 300 women), yet of the clinical pregnancies 51% were multiples ‐ this is much higher than other studies included in this review that have similar or higher numbers of embryos replaced. The clinical pregnancy rate was determined on ultrasound 4 weeks after embryo transfer, however the implantation rate is reported as being captured by ultrasound at 14 days after embryo transfer; at such an early gestation it is unlikely that gestational sacs would be visible. Lastly, the methods state that the antagonist protocol was used for women with a previous history of OHSS, however all women were undergoing their first ICSI cycle. Funding: authors confirmed no funding used Author correspondence: yes, with Ahmed Maged (prof.ahmedmaged@gmail.com) Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote:"automated web‐based randomization system"
Allocation concealment (selection bias)	Low risk	No description in paper, however authors confirmed during correspondence that the randomisation was tied to the participant trial ID and subversion of allocation was not possible
Blinding of participants and personnel (performance bias) All outcomes	High risk	paper states quote:"neither the participants nor the clinicians were masked to group assignment"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Appears to be no attrition
Selective reporting (reporting bias)	Low risk	Trial registered prospectively and all outcomes are reported
Other bias	High risk	There are a number of concerns about this trial, including the high recruitment rate, large number of other RCTs published recently by the first author, and an unexpectedly high multiple pregnancy rate.

*Study characteristics*
Methods	RCT, 2 arms, 229 randomised Setting: Hong‐Kong, research/academic institution, one centre Study period: March 2013 to April 2016 (unclear if recruitment period or whole study period)
Participants	Criteria related to previous IVF failure :no Inclusion criteria: scheduled for frozen embryo transfer (FET) cycles using non‐donor oocytes, normal ovulation and were deemed suitable for natural‐cycle FET Exclusion criteria: any uterine anomaly or pathology such as endometrial polyps, endometriomas larger than 4 cm or hydrosalpinx.
Interventions	Study group: pipelle procedure performed once in the mid‐luteal phase of the cycle preceding the embryo transfer cycleControl group: endocervical manipulation with pipelle, performed once in the mid‐luteal phase of the cycle preceding the embryo transfer cycle
Outcomes	Reported in paper: live birth/ongoing pregnancy, clinical pregnancy, miscarriage, multiple pregnancy Obtained by author correspondence: ‐
Notes	Trial registration: ChiCTRTRC‐12002389 (registered Aug 2012, registered prospectively) Additional concerns and comments: None Funding: None stated Author correspondence: minimal correspondence with jennifermak@cuhk.edu.hk Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote:"computer generated random numbers"
Allocation concealment (selection bias)	Unclear risk	The random numbers were quote:"concealed in opaque envelopes" unclear if SNOSE
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Sham procedure used.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only reporting objective outcomes
Incomplete outcome data (attrition bias) All outcomes	Low risk	8/115 and 5/114 women withdrew from the trial (small numbers)
Selective reporting (reporting bias)	Low risk	Registered prospectively and all outcomes reported
Other bias	Low risk	‐

*Study characteristics*
Methods	RCT, 2 arms, 129 randomised (only including each women's first randomisation) Setting: USA, one centre Recruitment period: Jan 2014 ‐ Dec 2017 (provided by authors)
Participants	Criteria related to previous IVF failure: no Inclusion criteria: all patients undergoing embryo transfer who are in the cycle prior to their embryo transfer Exclusion criteria: patients not undergoing embryo transfer, Known pregnancy, Active pelvic infection, Known endometrial hyperplasia or cancer, Inability to tolerate endometrial catheter placement, Severe cervical stenosis, patients who will receive operative hysteroscopy in the cycle prior to embryo transfer
Interventions	Study group 1: endometrial injury performed with pipelle on days 21‐27 in the cycle prior to the IVF cycle Study group 2: endometrial injury performed by a Shepard insemination catheter on days 21‐27 in the cycle prior to the IVF cycle
Outcomes	Reported in paper: live birth, pain Obtained by author correspondence: live birth, clinical pregnancy and pain data provided for the 129 women randomised. Miscarriage calculated as the difference between clinical pregnancy and live birth.
Notes	Trial registration: NCT04363879 (registered April 2020, registered retrospectively) Additional concerns and comments: none Funding: The protocol states quote: "This study does not have a budget. There will be no charge to the patient for the endometrial activation because the physicians have agreed to donate their time, and the procedures will be done in‐office" Author correspondence: yes, undertaken with Brad.Hurst@atriumhealth.org and goldrick@uthscsa.edu Publication: abstract only This study was presented as a conference abstract in 2017. The authors have since registered the trial and updated the 'Study Results' tab with information and a Statistical Analysis Plan. Information from all of these sources, including correspondence with the authors, has been used for the purposes of this review. During author correspondence it was discovered that the study had permitted re‐randomising of women. Therefore we have used data in this review from only each women's first randomisation ‐ and a total of 129 women were randomised.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The authors confirmed that a quote:"1:1 ratio of codes went into envelopes which were then shuffled and chosen at random to randomize patients"
Allocation concealment (selection bias)	High risk	The authors confirmed that quote:"Envelopes were plain envelopes without writing on them", these do not meet the SNOSE criteria and are therefore considered high‐risk
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was used.
Blinding of outcome assessment (detection bias) All outcomes	High risk	The outcome of pain during the procedure was self‐reported by participants who were not blind to their treatment allocation
Incomplete outcome data (attrition bias) All outcomes	High risk	Women were excluded for not undergoing the endometrial scratch or an embryo transfer. Of 129 women, 13 were excluded (8 in the pipelle arm, 5 in the Shepard catheter arm) which is 10% of the 129 women recruited, therefore this is considered high‐risk
Selective reporting (reporting bias)	High risk	The trial was registered retrospectively and important outcomes such as bleeding and miscarriage were not reported
Other bias	High risk	Only an abstract is available. Although the authors provided additional data and clarification via email, it remains possible that important methodological detail that may result in bias has not been disclosed. Additionally, the trial was registered retrospectively.

*Study characteristics*
Methods	RCT, 2 arms, 1048 randomised Setting: UK, 16 IVF clinics Study period: June 2016 – Oct 2019 (unclear if recruitment period or entire study period)
Participants	Criteria relating to previous IVF failure: yes, women undergoing their first IVF cycle Inclusion criteria: (extracted from protocol) 1. Women expected to be aged between 18 and 37 years (inclusive) at time of egg collection. 2. First‐time IVF with or without ICSI treatment using the antagonist or long protocol only. 3. Expected to receive treatment using fresh embryos. 4. Expected good responders to treatment, with: a. Ovulatory menstrual cycle b. Normal uterine cavity c. Expected good ovarian reserve Exclusion criteria: (extracted from protocol) 1. Previous trauma/surgery to the endometrium (e.g. resection of submucous fibroid, intrauterine adhesions). 2. Body mass index (BMI) of 35 kg/m2 or greater. 3. Known grade 4 (severe) endometriosis. 4. Currently participating in any other fertility study involving medical/surgical intervention. 5. Expected to receive protocols other than antagonist or long (e.g. ultra long protocol). 6. An endometrial scratch (or similar procedure, e.g. endometrial biopsy for the collection of natural killer cells) is planned. 7. Previously randomised into this trial.
Interventions	Study group: Endometrial Scratch (ES) performed in the mid luteal phase prior to IVF/ICSI. The procedure is performed with a pipelle or similar. Control group: no procedure
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage. Information regarding pain was provided in a box plot during the oral presentation, however it was not possible to determine the mean and SD from this presentation. Obtained from author correspondence: ‐
Notes	Trial registration: ISRCTN23800982 (registered May 2016, registered prospectively) Additional concerns and comments: none Funding: NIHR (UK) Author correspondence: yes, limited correspondence undertaken with Robin Chatters (r.chatters@sheffield.ac.uk) Data reported in the trial abstract was slightly different to that reported in the oral presentation; however the authors did not answer our questions, we therefore used the abstract data. For the same reason, the methodological information relating to the trial was extracted from the supplied study documents (protocol, SAP etc) as the authors did not answer our questions. Publication: abstract only (and oral presentation viewed at ESHRE 2020)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The protocol states quote: "The randomisation schedule will be generated by Sheffield CTRU prior to the start of the trial; access to the schedule will be limited only to the trial statistician. The randomisation sequence will be computer generated and stratified by site and protocol (antagonist or long protocol). Random permuted blocks of variable size will be used to ensure enough participants are allocated evenly to each arm of the trial at each site"
Allocation concealment (selection bias)	Low risk	The protocol states "Research staff at recruiting centres will be unable to access the randomisation sequence and will use a web‐based computer system with restricted access rights to enter participant details; randomisation outcome will then be revealed"
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was used. Also it was observed that a higher proportion of women in the endometrial injury arm underwent double embryo transfer compared to women in the control arm.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The outcome of pain during the procedure were self‐reported by participants who were not blind to their treatment allocation, however this data was not available for inclusion in this review.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	No information available about the numbers of women lost to follow‐up or withdrawn
Selective reporting (reporting bias)	Unclear risk	Although the study protocol was registered prospectively, only an abstract is available and therefore it is not possible to assess whether all registered outcomes have been/will be reported
Other bias	Low risk	Only an abstract is available and the trial team refused to answer our questions by email, however sufficient information was available and therefore this is not seen to constitute a high risk of bias.

*Study characteristics*
Methods	RCT, 2 arms, 100 randomised Setting: India, private fertility clinic, one centre Recruitment period: May 2007 to July 2008
Participants	Criteria relating to previous IVF failure: yes, women with at least 1 previous failure Inclusion criteria: good responders in the previous IVF cycle (development of at least 4 good‐quality embryos); ≤ 37 years Exclusion criteria: endometrial tuberculosis in the past; intramural fibroid distorting the endometrial cavity/submucous myoma; Asherman's syndrome; evidence of hydrosalpinx
Interventions	Study group: pipelle procedure conducted twice: once at the time of hysteroscopy on days 7‐10, and again on days 24‐25 (of the cycle before IVF) Control group: hysteroscopy on days 7‐10 of the cycle before IVF
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage, multiple pregnancy Obtained from author correspondence: ‐
Notes	Trial registration: trial does not appear to be registered Additional concerns and comments: none Funding: none stated Author correspondence: yes, undertaken with corresponding author (sachnar@rediff.com) Publication: full‐text. Numbers of previous attempts were 2.3 ± 0.52 and 2.5 ± 0.7, so the study was classified in the subgroup 'Two or more previous failures'
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random numbers were used
Allocation concealment (selection bias)	Low risk	Sealed and consecutively numbered opaque envelopes were used
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding was employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Only objective outcomes reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	It does not appear there was any attrition or loss to follow‐up
Selective reporting (reporting bias)	High risk	The trial does not appear to have been registered, and important outcomes such as adverse events not reported
Other bias	High risk	Trial not registered.

*Study characteristics*
Methods	RCT, 2 arms, 158 randomised Setting: Brazil, academic research centre, one centre Recruitment period: June 2010 to March 2012
Participants	Criteria relating to previous IVF failure: no Inclusion criteria: all women undergoing ART with planned fresh embryo transfer aged < 38 years Exclusion criteria: ‐
Interventions	Study group: pipelle procedure performed once 7 to 14 days before the start of ovarian stimulation Control group: sham procedure performed at the same time. This procedure comprised drying of the cervix with no insertion of any instrument into the cervix or womb
Outcomes	Reported in paper: live birth, clinical pregnancy, miscarriage, multiple pregnancy, pain, bleeding Obtained by author correspondence: ‐
Notes	Trial registration: NCT01132144 (registered May 2010, registered prospectively) Additional concerns and comments: none Funding: this study was funded by two Brazilian official government research foundations: CNPq (direct funding (process number 473475/2010‐3) and research scholarship) and CAPES (PhD scholarship). Author correspondence: yes, two of the authors are authors on this review (CO, WPM) Publication: full‐text.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Paper states quote:"computer generated random sequence of numbers in blocks of 30 (each block having 15 numbers assigned to intervention and 15 to control"
Allocation concealment (selection bias)	Low risk	Paper states quote:"The allocation was sealed in consecutively numbered opaque envelopes and an envelope was assigned as the participant entered the study; however, sealed envelopes were only opened just before the procedure, to ensure allocation concealment"
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Sham procedure employed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Sham procedure implemented therefore outcome assessors were blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	There was no attrition or loss to follow‐up
Selective reporting (reporting bias)	Low risk	Trial was registered prospectively and all outcomes reported
Other bias	High risk	The trial was terminated early for a positive effect