If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
The endometrial scratch (ES) procedure is an IVF ‘add-on’ that is sometimes provided prior to the first IVF cycle. A 2019 systematic review concluded that there was insufficient evidence regarding whether the ES has a significant effect on pregnancy outcomes (including live birth rate, LBR) when undertaken prior to the first IVF cycle. Further evidence has been published following this review, including the Endometrial Scratch Trial (ISRCTN23800982). The objective of this review was to synthesise and critically appraise the evidence for the clinical effectiveness and safety of the ES procedure in women undergoing their first IVF cycle. Databases searched include MEDLINE, EMBASE, CINAHL and ClinicalTrials.gov. Eligible RCTs included women undergoing IVF for the first time that reported the effectiveness and/or safety of the ES procedure. Twelve studies were included. Meta-analysis showed no evidence of a significant effect of the ES on LBRs (10 trials, odds ratio [OR] of 1.17, 95% confidence interval [CI], 0.76 to 1.79), and other pregnancy outcomes. This review confirms that there is a lack of evidence that ES improves pregnancy outcomes, including LBR, for women undergoing their first IVF cycle. Clinicians are recommended not to perform this procedure in individuals undergoing their first cycle of IVF.
Endometrial scratch (ES) is a procedure that has been rapidly adopted into routine clinical practice at a rate that far exceeds the rate of production of good quality evidence (Lensen et al., 2016). While the procedure was initially adopted for women suffering from recurrent implantation failure during IVF treatment based on evidence from the initial study by Barash et al (2003), it then rapidly spread to other populations of women and other types of treatments (Barash et al., 2003).
Amongst these groups, are women undergoing their first IVF cycle, where ES had started to be offered to this group despite the lack of evidence (Lensen et al., 2016). Indeed there have since been many studies, but these were mostly subject to significant confounding factors or not designed or powered to address this particular group (Vitagliano et al., 2019).
A systematic review on this topic, focussing on randomised controlled trials (RCTs) and the effectiveness of the ES procedure in women undergoing their first IVF cycle, was published by Vitagliano et al in 2019 (Vitagliano et al., 2019). The review included seven RCTs and concluded that there was no evidence that the ES followed by IVF compared to IVF alone increased the success of treatment, with a relative risk/risk ratio (RR) of live birth (or ongoing pregnancy if life birth rate [LBR] was not reported) of 0.99 (95% CI: 0.57 to 1.73, p=0.97) (Vitagliano et al., 2019). Secondary outcomes (miscarriage, multiple pregnancy and ectopic pregnancy) were also not significantly altered by undertaking the ES. Notably, the small sample sizes of the included studies resulted in huge uncertainty around the effects of ES in women undergoing their first IVF cycle so positive effect could not be rule out. In addition, the trials included were at either a high or unclear risk of bias making it difficult to make reliable conclusions. Consequently, the authors concluded that a robust and definitive RCT is required to assess the effect of ES on the chances of success of the first IVF cycle (Vitagliano et al., 2019).
Recently we have published evidence from a large definitive multicentre RCT in the United Kingdom (the Endometrial Scratch Trial) which focussed only women undergoing first IVF cycle, with or without intracytoplasmic sperm injection (ICSI) (Metwally et al., 2021). In this trial we did not find evidence for any significant benefit from the ES. Given the large number of other studies in the literature, some with similar and some with conflicting findings and given that often a meta-analysis of all published literature rather than a single RCT is important in propagating a certain research finding and implementing change in practice, we performed this meta-analysis to synthesise the effect of ES in increasing success rates of pregnancy outcomes in women undergoing first time IVF treatment with or without ICSI.
The review was conducted, and this manuscript written, in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines (Page et al., 2021).
Only RCTs examining the clinical effect or safety of ES in women undergoing their first IVF cycle with or without ICSI, compared to treatment as usual (IVF/ICSI without the use of ES), were eligible for inclusion. Studies that included participants who were undertaking intrauterine insemination (IUI) or ovulation induction (or other treatments not classed as IVF) and/or their second or subsequent IVF cycle were excluded from this review, unless separate outcome data could be extracted for a subset of women who have undertaken their first IVF cycle. We included all forms of ES regardless of the timing of the procedure during the cycle, however, procedures defined as a mock transfer, where the aim of the procedure was not to scratch the endometrium but to test embryo transfer techniques, were excluded.
We excluded reports published as abstracts only, and/or, reports published in languages other than English, where insufficient methodological details are reported in the abstract (if written in English) to allow extraction of study characteristics.
The following clinical and safety outcome measures were considered, which were included regardless of the definition or timing of assessments:
Primary: Live birth rate
Clinical pregnancy rate
Ongoing pregnancy rate
Ectopic pregnancy rate
Pain related to the procedure
Adverse and serious adverse event rates (AEs and SAEs)
Search methods for identification of studies
Data sources and search period
The following electronic databases were searched without language restrictions on 10th January 2020 except for clinicaltrials.gov which was searched on 21st September 2020.
MEDLINE via Ovid from 1948 to present (Appendix A)
EMBASE (Ovid) from 1980 to present
Cochrane database of systematic reviews from 2005 to present
Cumulative Index to Nursing and Allied Health Literature (CINAHL) from 1981 to present
CENTRAL via The Cochrane Register of Studies Online from 1898 to present
Language restrictions were applied after the search was undertaken.
Clinicaltrials.gov was searched using combinations of keywords - "endometrial biopsy" AND "infertility", "endometrial biopsy" AND "subfertility", "endometrial hysteroscopy" AND "infertility", "endometrial hysteroscopy" AND "subfertility".
We also hand searched the reference list of all retrieved articles, relevant journals and conference proceedings. In addition, we contacted the authors seeking data clarification and to obtain additional information on missing data.
Selection of studies and data extraction
Titles, abstracts and full-text articles were screened independently by two reviewers (JH and LR). Any disagreements regarding eligibility were resolved through discussion with RC.
Data were extracted from the studies by one researcher (JH) and all data checked by RC. Data extracted included the outcomes, study characteristics (e.g., country where research was conducted, number of trial arms, description of trial arms, control condition(s), timing of ES procedure in menstrual cycle, device used for ES) and participant characteristics (e.g., average age of trial population, average duration of infertility, and egg source). Where further information was needed, the authors were contacted.
Quality assessment strategy
The methodological quality of the included RCTs was assessed using the Cochrane Collaboration risk of bias assessment criteria at an outcome level (Higgins et al., 2019). The risk of bias was assessed for each reported outcome. The assessment was undertaken independently by two reviewers (either PK, RC, AP or JH). Discrepancies were resolved by a third reviewer who had not been involved in the previous assessments of that study (RC or JH). Studies were graded with an overall risk of bias of ‘high’, ‘low’ or ‘unclear’.
Statistical analysis was conducted according the guidelines outlined by the Cochrane Collaboration (Higgins et al., 2021). For each included RCT, summaries on the number of events and the denominator were recorded for binary outcomes and meta-analysis performed using RevMan software version 5.3 (Review Manager (RevMan). Version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). Study specific treatment effects as measured by odds ratios (ORs) and RRs were combined together to produce pooled ORs or RRs with 95% CIs, where appropriate using Mantel-Haenszel method which performs relatively well in several settings (Piaget‐Rossel and Taffé, 2019). A random effects model was used when between study heterogeneity was viewed as substantial. Otherwise, a fixed effects model was used when there was no evidence of significant heterogeneity. We assessed heterogeneity between studies using the χ2 and I2 statistics (Higgins et al., 2003; Higgins and Thompson, 2002). For example, the I2 statistic quantifies that the percentage of total variation in treatment effects estimates attributable to between study heterogeneity and a value of >50% indicates evidence of significant heterogeneity of treatment effects between studies (Deeks JJ, Higgins JPT, 2021). A subgroup analysis was conducted for two trials that reported ‘early’ miscarriages (prior to 12 weeks) (Izquierdo Rodriguez et al., 2020; Maged, 2018). Only one trial separately reported ‘late’ miscarriages (occurring between 12 and 24 weeks) (Izquierdo Rodriguez et al., 2020), with all other trials reporting miscarriages up to 24 weeks (which also included early miscarriages) – therefore, due to the heterogeneity of this outcome, the late miscarriage subgroup was not included in this analysis.
In the evidence of significant heterogeneity, a random effects model was used in addition to an exploration of the causes of heterogeneity followed by a sensitivity analysis where appropriate. Meta-analysis are presented in forest plots. Some outcomes (pain scores, adverse events) were narratively assessed due to a small number of studies reporting these outcomes, and/or heterogeneity in the definition of outcomes.
Screening and study eligibility
Searches identified a total of 1462 records. When needed, authors were contacted regarding missing data and to help assess eligibility. Of the 14 authors that were contacted, eight were not contactable. One author confirmed that the trial was not eligible for inclusion as recruitment to the trial never commenced (Checa, 2013). The authors of four trials that included participants undergoing their first IVF cycle but did not present their outcomes separately provided data and were included in the review (Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Mackens et al., 2020; Nastri et al., 2013). Polanski et al., which included women undergoing an unselected number of previous IVF cycles, could not be contacted to obtain data for first cycle participants only. However, unpublished data presented, received directly from the authors in a recent systematic review were used (Vitagliano et al., 2019). A risk of bias assessment could not be conducted for this study due to a lack of methodological details described in the abstract. After screening, 11 RCTs were eligible for inclusion in the review. We also included the results of our recently conducted Endometrial Scratch Trial, published after the searches were undertaken, thus bringing the total to 12 RCTs (Metwally et al., 2021) published between 2010 and 2021. Details of literature search and study selection can be seen in figure 1.
Characteristics of included trials
Table 1 summarises the characteristics of the 12 included RCTs comprising 3234 participants undergoing their first IVF/ICSI cycle.
Table 1Characteristics of included trials
Author (trial registration number)
Number participants randomised undergoing 1st IVF cycle (total)
Timing of ES:
Karimzade 2010 (Karimzade et al., 2010) (NCT00846183)
Single centre, Iran, June 2008 to January 2009
Endometrial scratch vs usual care
Age <38 years, BMI >19 and <30 kg/m2, day 3 FSH < 12 mIU/mL
IVF with or without ICSI
During IVF Day of oocyte retrieval
CPR, OPR, IR, AE
Yeung 2014 (Yeung et al., 2014) (NCT01977976)
Single centre, China, March 2011 to October 2013
Endometrial scratch vs usual care
IVF with or without ICSI
Prior to IVF 7 days post LH surge / day 21 of cycle preceding IVF
LBR, OPR, CPR, IR, MR, MPR, QP, AE
Mahran 2016 (Mahran et al., 2016) (ISRCTN61316186)
Multi-centre (2 centres), Egypt, June 2012 to September 2014
Endometrial scratch vs usual care
20 to 40 years, FSH ≤12 mIU/mL and ≥2 good quality embryos replaced
IVF with or without ICSI
Prior to IVF Day 21 to 24 of cycle preceding IVF
LBR, OPR, CPR, IR, MR, MPR, QP, AE
Maged 2018 (Maged et al., 2018) (NCT02660125)
Single-centre, Egypt, January 2016 to March 2017
Endometrial scratch vs usual care
Age < 40 years and FSH < 10 mIU/mL
Prior to IVF
CPR, IR, MR, MPR, AE
Liu 2017 (Liu et al., 2017) (ChiCTR-IOR-17011506)
Single-centre, China February 2012 to November 2014
Only three of the 12 studies were multicentre RCTs (Lensen et al., 2019; Mahran et al., 2016; Metwally et al., 2020), with other studies involving a single centre. All 12 studies were individually randomised and were conducted across ten countries: Iran, Hong Kong, Egypt (two studies), China, USA, Belgium, Spain, Turkey, Brazil, and the UK. One RCT was undertaken multi-nationally across five countries (Lensen et al., 2019). The total number of participants included in each trial undergoing their first IVF cycle ranged from 18 to 1048.
Eleven studies were two-arm RCTs (Eskew et al., 2019; Izquierdo Rodriguez et al., 2020; Karimzade et al., 2010; Lensen et al., 2019; Mackens et al., 2020; Maged et al., 2018; Mahran et al., 2016; Metwally et al., 2020; Nastri et al., 2013; Polanski et al., 2014; Yeung et al., 2014) and one was a four-arm RCT (Liu et al., 2017). Nine of the two-arm RCTs compared the ES procedure to usual care and two trials included a comparator involving a sham procedure (Eskew et al., 2019; Nastri et al., 2013). The four-arm trial compared ES at two different time points with a sham procedure undertaken at the same two different time points in the menstrual cycle – proliferative and luteal (Liu et al., 2017).
Recruitment to three of the trials was prematurely ended due to an unplanned futility analysis showing no differences in clinical pregnancy rates between intervention and control groups in one trial (Eskew et al., 2019), a planned interim analysis identifying higher miscarriage rates in the ES arm in another trial (Mackens et al., 2020), and identifying a significant benefit of the ES during a planned interim analysis (Nastri et al., 2013).
Characterisation of ES procedure and timing
The method of undertaking ES was largely similar across studies. Most used a pipelle sampler to invoke injury, except for one trial that used an embryo transfer catheter (Izquierdo Rodriguez et al., 2020), one that used a Novak Curette (Karimzade et al., 2010), and another that used either a pipelle or Wallace endometrial sampler (Polanski et al., 2014). However, there was substantial variation in the timing of when ES was performed across trials. Two trials undertook ES during the IVF cycle, either on the day of egg collection (Karimzade et al., 2010), or during ovarian stimulation (Mackens et al., 2020). Ten trials undertook ES in the menstrual cycle prior to IVF, with seven within the luteal phase (Eskew et al., 2019; Izquierdo Rodriguez et al., 2020; Maged et al., 2018; Mahran et al., 2016; Metwally et al., 2021; Polanski et al., 2014; Yeung et al., 2014), and one during the early or mid-luteal phase (Nastri et al., 2013). Lensen et al. undertook ES at any point between day three of the menstrual cycle prior to ES (Lensen et al., 2019), and day three of the cycle in which IVF was being undertaken. However, this trial reported that the median time (interquartile range, IQR) between ES and embryo transfer was 35 days (22 to 39), and therefore, it is likely that most women received ES in the menstrual cycle prior to IVF. Liu et al. undertook ES either the proliferative or luteal phases of the menstrual cycle (Liu et al., 2017), therefore in this review the two time points of delivery of ES, or the sham procedure, were combined, so that, for each outcome, there was one rate for the ES arm (both proliferative and luteal), and another for the sham arm (both proliferative and luteal). Two trials provided hysteroscopy to all trial participants, prior to IVF (Mahran et al., 2016; Yeung et al., 2014).
Trials used different participant eligibility criteria. Ten trials had age restrictions with an upper limit of between 35 and 49 years (Eskew et al., 2019; Karimzade et al., 2010; Lensen et al., 2019; Liu et al., 2017; Mackens et al., 2020; Maged et al., 2018; Mahran et al., 2016; Metwally et al., 2021; Nastri et al., 2013; Polanski et al., 2014). Three trials restricted the body mass index (BMI) to an upper limit ranging from 30 to 35 kg/m2 (Karimzade et al., 2010; Maged et al., 2018; Metwally et al., 2021). Five trials selected women that were deemed to have a good ovarian reserve, by allowing only those with a certain level of follicle stimulating hormone (FSH) to participate, with a maximum level of 10 IU/mL and 12 IU/mL in two (Maged et al., 2018; Metwally et al., 2021), and three studies (Karimzade et al., 2010; Liu et al., 2017; Mahran et al., 2016), respectively. At eligibility screening, two studies set requirements for the number of oocytes collected or embryos replaced: ≥2 replaced in one study (Mahran et al., 2016), and four to 14 oocytes collected in another (Karimzade et al., 2010). Only one trial stipulated that the embryos replaced had to be of a certain quality, with Mahran et al. (Mahran et al., 2016) stating that the two or more embryos replaced had to be ‘good’ quality. However, the exact method of grading embryo or defining a good quality embryo was not reported. Two trials excluded women receiving donor eggs (Mackens et al., 2020; Yeung et al., 2014), whilst one study only included those receiving donor eggs (Izquierdo Rodriguez et al., 2020).
Ten trials reported LBRs (Eskew et al., 2019; Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Liu et al., 2017; Mackens et al., 2020; Mahran et al., 2016; Metwally et al., 2021; Nastri et al., 2013; Polanski et al., 2014; Yeung et al., 2014). Clinical pregnancy rates were reported in all trials. However, this was defined inconsistently, with marked variation in the time point at which this outcome was assessed: at four weeks post embryo transfer in two trials (Maged et al., 2018; Mahran et al., 2016); five weeks in one trial (Karimzade et al., 2010); six weeks in four trials (Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Liu et al., 2017; Yeung et al., 2014); seven (Mackens et al., 2020) and eight weeks (Metwally et al., 2021); and not reported in three trials (Eskew et al., 2019; Nastri et al., 2013; Polanski et al., 2014). Ongoing pregnancy rates were reported in four trials, which was assessed at 12 (Izquierdo Rodriguez et al., 2020; Karimzade et al., 2010; Lensen et al., 2019) and 20 (Yeung et al., 2014) weeks post embryo transfer.
Implantation rates were reported in seven studies (Izquierdo Rodriguez et al., 2020; Karimzade et al., 2010; Liu et al., 2017; Maged et al., 2018; Mahran et al., 2016; Metwally et al., 2021; Yeung et al., 2014). Six studies defined this similarly as the number of gestational sacs, divided by the number of embryos transferred, whilst in Metwally et al.,(Metwally et al., 2021) this was defined as the number gestational sacs divided by the number of participants randomised to each arm (under intention to treat principles). In order to include our trial in this meta-analysis we therefore recalculated this outcome using the number as embryos transferred as the denominator. Miscarriage rates per clinical pregnancy were reported in 11 trials (Eskew et al., 2019; Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Liu et al., 2017; Mackens et al., 2020; Maged et al., 2018; Mahran et al., 2016; Metwally et al., 2021; Nastri et al., 2013; Polanski et al., 2014; Yeung et al., 2014), with the time point of data collection differing between 12 to 24 weeks of gestation, but unclear in two trials (Lensen et al., 2019; Mackens et al., 2020). A subjective assessment of pain of the ES procedure on a numerical rating scale was reported in three trials (Lensen et al., 2019; Metwally et al., 2021; Nastri et al., 2013), with four studies providing qualitative reports of pain (Liu et al., 2017; Mackens et al., 2020; Mahran et al., 2016; Yeung et al., 2014). Eight trials reported adverse events and/or complications in the participating women (Karimzade et al., 2010; Lensen et al., 2019; Liu et al., 2017; Mackens et al., 2020; Maged et al., 2018; Mahran et al., 2016; Metwally et al., 2021; Yeung et al., 2014), and four trials reported such events in the baby or neonate (Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Metwally et al., 2021; Nastri et al., 2013).
Risk of bias assessment
It was not possible to conduct a risk of bias analysis for Polanski et al. (Polanski et al., 2014) as the author did not respond to our request for essential missing information. The only information available was from a recent review, which used a previous version of the risk of bias tool and therefore the authors assessments could not be considered in this review (Polanski et al., 2014). For other included studies, Figure 2 summarises our assessment of the risk of bias.
Domains one to three were consistent across all outcomes per study. Four trials were assessed as ‘some concerns’ in domain one (allocation concealment) (Izquierdo Rodriguez et al., 2020; Karimzade et al., 2010; Liu et al., 2017; Maged et al., 2018). Nine trials used a computerised system to undertake randomization (Eskew et al., 2019; Izquierdo Rodriguez et al., 2020; Karimzade et al., 2010; Lensen et al., 2019; Mackens et al., 2020; Maged et al., 2018; Metwally et al., 2021; Nastri et al., 2013; Yeung et al., 2014), one trial used sealed envelopes (Mahran et al., 2016), whilst another used a table of random numbers (Liu et al., 2017).
Domain two considered the risk of bias due to deviations from the intended interventions. Three trials were considered to have some concerns of bias for this domain (Izquierdo Rodriguez et al., 2020; Karimzade et al., 2010; Mackens et al., 2020). The differing dropout rate between the ES group (8.5%) and the control group (2.2%) in Izquierdo Rodriguez et al. resulted in this assessment (Izquierdo Rodriguez et al., 2020). Similarly in Karimzade et al., there were four patients excluded from the analysis in the ES arm only (Karimzade et al., 2010).
Domain three considered the risk of bias due to missing outcome data. The rate of missing data was low across all trials, therefore all were considered to be at low risk of bias for this domain. All studies were judged to be at low risk of bias for domain four, ‘measurement of the outcome’.
Given the nature of the outcomes being assessed, patient knowledge of the intervention was unlikely to have affected the analysis. Therefore, even though only three of the twelve included studies involved some form of blinding (Eskew et al., 2019; Liu et al., 2017; Nastri et al., 2013), this is not considered to affect the patient outcomes. There was also no blinding in most of the included trials, however it is unlikely that participant's or clinician's knowledge of the intervention could have biased outcome assessment due to the included trials using objective outcome measures unlikely to be influenced by placebo effect.
Domain five assessed the risk of bias in selection of the reported result. For all outcomes, studies where the outcome was not specified prior to the start of the trial (no protocol, trials registry, or a retrospectively added trials registry), the timing of the outcome was not specified, or the outcome specification in the paper did not match the protocol/registry we considered to have some concerns.
Only Lensen et al. and Nastri et al. were considered to have a low risk of bias across all assessments (Lensen et al., 2019; Nastri et al., 2013). Metwally et al. was denoted to have ‘some concerns’ for the implantation rate outcome only, as this was originally reported under intention to treat principle using the number of randomised women as the denominator in each arm – this was recalculated using the number of embryos transferred for the purposes of this meta-analysis (Metwally et al., 2021). All other studies had at least one outcome considered to have some bias concerns.
Live birth rate (LBR)
Pooled analysis of the ten trials that reported LBR showed no evidence for a significant effect for ES on the LBR; OR of 1.17 (figure 3, 95% CI: 0.76 to 1.79, p-value=0.48) (Eskew et al., 2019; Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Liu et al., 2017; Mackens et al., 2020; Mahran et al., 2016; Metwally et al., 2021; Nastri et al., 2013; Polanski et al., 2014; Yeung et al., 2014). There was a substantial between study heterogeneity in treatment effects (I2 = 83%).
Clinical pregnancy rate (CPR)
Pooled analysis of the 12 trials that reported CPR showed no evidence for a significant effect for ES on the CPR.; OR of 1.18 (figure 4, 95% CI: 0.82 to 1.72, p-value=0.38) (Eskew et al., 2019; Izquierdo Rodriguez et al., 2020; Karimzade et al., 2010; Lensen et al., 2019; Liu et al., 2017; Mackens et al., 2020; Maged et al., 2018; Metwally et al., 2021; Nastri et al., 2013; Polanski et al., 2014; Yeung et al., 2014). Results showed evidence for significant heterogeneity between the included studies (I2 = 82%).
Ongoing pregnancy rate (OPR)
Pooled analysis of the four studies that reported on OPR (Izquierdo Rodriguez et al., 2020; Karimzade et al., 2010; Lensen et al., 2019; Yeung et al., 2014) showed no evidence for a significant effect for ES on the OPR; OR of 0.86 (figure 5, 95% CI: 0.49 to 1.48, p-value=0.58). Results showed evidence for significant heterogeneity between the included studies (I2 = 71%).
Implantation rate (IR)
Two trials (Karimzade et al., 2010; Mahran et al., 2016) where the implantation rate was only reported as a percentage without the absolute numbers were excluded from this analysis. Consequently, five trials (Izquierdo Rodriguez et al., 2020; Liu et al., 2017; Maged et al., 2018; Metwally et al., 2021; Yeung et al., 2014) were included and the overall effect showed evidence to support a significant improvement in the IR attributed to the use of ES; OR of 1.14 (figure 6, 95% CI: 1.02 to 1.27, p-value=0.02), with moderate evidence of heterogeneity (I2 = 23%).
Miscarriage rate (MR)
Analysis of ten trials reporting on this outcome (Eskew et al., 2019; Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Liu et al., 2017; Mackens et al., 2020; Maged et al., 2018; Mahran et al., 2016; Metwally et al., 2021; Nastri et al., 2013; Yeung et al., 2014) which did not show evidence for a significant effect of ES on reducing the MR; OR of 0.96 (figure 7, 95% CI: 0.57 to 1.63, p-value = 0.89). Results showed evidence for moderately significant heterogeneity (I2 = 47%).
We performed a subgroup analysis on ‘early’ miscarriages (before 12 weeks), which was reported by two studies (Izquierdo Rodriguez et al., 2020; Maged, 2018). Results remained unchanged as there was no evidence for a significant effect for ES on the early miscarriage rate, with no evidence of significant heterogeneity between the included studies (I2=0%), but with significant uncertainty in this result (early: OR of 0.67 (figure 8, 95% CI: 0.31 to 1.43, p-value = 0.29).
Ectopic pregnancy rate (EPR)
Pooled analysis of the six studies that reported ectopic pregnancy (Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Liu et al., 2017; Mackens et al., 2020; Mahran et al., 2016; Metwally et al., 2021), showed no evidence for a significant effect for ES on this outcome, with high uncertainty; OR of 0.57, (figure 9, 95% CI: 0.24 to 1.35, p=0.20). There was no evidence of significant heterogeneity between studies (I2 = 0%), however, there was huge uncertainty in this estimate.
It was not possible to perform a meta-analysis on the effect of ES procedure on the pain experienced by participants as only three trials reported a quantitative outcomes used to assess pain (Lensen et al., 2019; Metwally et al., 2021; Nastri et al., 2013). Two of the trials reported similar mean (SD) pain levels (on a rating scale of 0 to 10) directly post procedure (4.1 (2.4)) and during the procedure (4.2 (2.5)), whereas the third reported slightly higher pain levels (6.42 (2.35)) directly after the procedure (Nastri et al., 2013). Other trials reported pain qualitatively, with three trials reporting a low proportion of participants that reported severe pain (0/150 in Yeung et al., 3/209 (1.4%) in Mahran et al., 0/70 in Liu et al., 2/70 (2.9%) in Mackens et al.) (Liu et al., 2017; Mackens et al., 2020; Mahran et al., 2016; Yeung et al., 2014).
Of the eight trials that reported AEs, seven appeared to have collected AEs in the ES arm only, although information was often not presented in the reports. Four trials reported that there were “no complications” recorded, (Karimzade et al., 2010; Liu et al., 2017; Maged et al., 2018; Yeung et al., 2014) and one trial reported “minimal” spotting for a few days post ES (number of participants not reported) (Mahran et al., 2016).
One trial reported that 4 (1.3%) participants experiencing fainting, 2 (0.7%) reported excessive pain and 1 (0.3%) reported excessive bleeding (Lensen et al., 2019). Another trial reported that 3 participants (5%) experienced bleeding (Mackens et al., 2020). Metwally et al. was the only study to report safety events in both trial arms, and found comparable/similar incidences of AEs between groups (Metwally et al., 2021).
Four trials reported AEs in the baby or neonate (Izquierdo Rodriguez et al., 2020; Lensen et al., 2019; Metwally et al., 2021; Nastri et al., 2013). The incidences of AEs in participants undergoing the first or subsequent cycle of IVF in these trials was low, with no “major” fetal malformations reported in one trial (Nastri et al., 2013), rare congenital abnormalities (ES arm: 3 (1.6%), TAU arm: 0 (0%)), fetal growth restriction (ES: 10 (5.2%), TAU: 8 (4.3%)) and neonatal death (ES: 0 (0%), TAU: 1, (0.5%)) in another trial (Lensen et al., 2019) and rare placentation abnormalities (ES: 1 (1.0%), TAU: 0 (0%)) and intrauterine growth restriction (ES: 0 (0%), TAU: 1 (1%)) in the final trial(Izquierdo Rodriguez et al., 2020). Metwally et al. was the only trial which presented accessible data separately for those babies born to participants undergoing their first IVF cycle, reporting no neonatal deaths in both arms, no severe congenital abnormalities in the ES arm and low numbers of congenital abnormalities in both trial arms (Metwally et al., 2021).
A sensitivity analysis was performed after exclusion of studies that were thought to be contributing to the heterogeneity to explore the impact on the conclusions. Four such studies were identified for exclusion from the analyses (Karimzade et al., 2010; Mackens et al., 2020; Mahran et al., 2016; Polanski et al., 2014).
Mahran et al., appeared to be a clear outlier both statistically and methodologically (Mahran et al., 2016). Firstly, this is the only trial that required participants to have two good quality embryos transferred to be eligible to participate in the trial – including such a stipulation prior to randomisation may have resulted in a participant population not comparable to other trials included in this review. Secondly, all participants received hysteroscopy prior to IVF, a procedure which in theory could have an effect similar to ES, thus exposing the endometrium to two rather than one event of controlled endometrial trauma. Finally, an average of three embryos were transferred, whilst all other trials included in this review averaged one or two embryos transferred.
Polanski et al. included women who had undergone an unselected number of previous IVF cycles and could not be contacted to obtain data for first cycle participants only. However, unpublished data pertaining to the participants that had undergone their first IVF cycle presented in a recent systematic review were used (Vitagliano et al., 2019). A risk of bias assessment therefore could not be conducted for this study due to a lack of methodological details described in the abstract. Mackens et al. and Karemizade et al. were the only studies that undertook ES during the IVF cycle while the other trials undertook ES in the menstrual cycle prior to ES (Karimzade et al., 2010; Mackens et al., 2020).
In summary, exclusion of these four studies eliminated or resulted in significant reduction in heterogeneity, but conclusions on the effect of ES on LBR (Supplementary Figure 1, OR of 1.01; 95% CI: 0.85 to 1.21), CPR (Supplementary Figure 2, OR of 1.11; 95% CI: 0.95 to 1.31), OPR (Supplementary Figure 3, OR of 1.11; 95% CI: 0.95 to 1.31), MR (Supplementary Figure 4, OR of 0.84; 95% CI: 0.56 to 1.27), and EPR (Supplementary Figure 5, OR of 0.55; 95% CI: 0.21 to 1.43) remained consistent and unchanged .
This systematic review and meta-analysis of RCTs in women undergoing the ES procedure prior to their first cycle of IVF has identified that there is no evidence of a significant effect on live birth rate, clinical pregnancy rate, ongoing pregnancy rate, ectopic pregnancy rate, or miscarriage rate. These conclusions were consistent following sensitivity analyses that excluded four studies that were contributing to significant heterogeneity. Uncertainty was high for some analyses, specifically miscarriage and ectopic pregnancy rates. A meta-analysis of implantation rates across five studies showed a significant positive effect of undertaking ES. However, due to the sensitivity of implantation rate to the number of embryos transferred, and artificial inflation of a sample size when more than one embryo is transferred per women, these results are unreliable and should be interpreted with extreme caution (Griesinger, 2016). Only five trials reported ectopic pregnancy rates, and for those trials that did, there were very few events reported, and surprisingly smaller trials reported more events than larger trials. This is difficult to explain, but may be attributable to trials using different definitions for this outcome. Few of the trials reported pain post ES (n=7); four of these trials reported a low proportion of participants experiencing “severe” pain post ES. Three trials reported similar moderate post ES pain ratings. The procedure seemed to be safe, with studies reporting very rare complications post ES. None of the included trials were deemed to be at a high risk of bias.
Numerous systematic reviews have been undertaken to assess the clinical effectiveness of the ES procedure in the first IVF cycle. However, previous reviews have been unable to conclude the effect of ES on the first cycle of IVF, due to a lack of definitive, high-quality evidence (Nastri et al., 2012; Potdar et al., 2012; Vitagliano et al., 2019). Following the completion of the Endometrial Scratch Trial, the results of which we include in this review, we can now conclude that the ES should not be undertaken prior to the first cycle of IVF.
The ease and simplicity of the ES procedure has perhaps been the main reason for its rapid adoption into routine clinical practice initially without the support of good quality research. The main problem that has afflicted previous studies is the inclusion of different and heterogeneous populations of participants and clinical practices. This has led to heterogeneity and a lack of reliability of evidence when addressing any one specific population. The current study focuses only on the population of women undergoing their first IVF cycle with or without ICSI.
There have been several key studies published over the last 7 years that are relevant to women undergoing first time IVF treatment. These include the studies by Yeung et al., Lensen et al. and the most recent study from our group (Lensen et al., 2019; Metwally et al., 2021; Yeung et al., 2014). The first of these studies (Yeung et al.) was conducted in an unselected population of women undergoing IVF, of whom nearly 70% were having their first IVF cycle and subgroup analysis of this group (N=209 of N=300 individuals included in this trial) and similarly found no difference in ongoing pregnancy rates between groups (Yeung et al., 2014). However in this study a mixture of protocols was used and there were no restrictions regarding age or day of embryo transfer with most patients receiving two embryo transfers. Similarly the study by Lensen et al, combined a mixture of patients with different prognostic potential, with two main subgroups, the first is women with recurrent implantation failure and the second is women who have had a maximum of one previous cycle (Lensen et al., 2019). The latter group although provided some useful reflections regarding women having endometrial scratch in the absence of recurrent implantation failure, cannot be used a substitute for a well-designed study powered to just the initial IVF cycle. This was what we aimed for in our recently published study (Metwally et al., 2021), which was the first large multicentre RCT that was powered only to a population of potential good responders due to undergoing their first IVF cycle, and accounted for many of the sources of heterogeneity that may have affected previous studies. Results showed that in this particular population, ES had no clinical benefit although it was tolerable and safe. However, we cannot ignore that the ease of use of endometrial scratch combined with the promises made by results of other studies regarding the ability of this procedure to revolutionise success rates, coupled with the numerous plausible physiological hypotheses that have been put forward regarding how this procedure may act to improve implantation, will lead to a difficulty in changing the current pattern of practice. We believe a meta-analysis that combined our study with all relevant previous studies would help conclusively settle the debate regarding this procedure and at the same time identify the main problems with the previous literature that can be highlighted in a systematic and robust way.
The main strength of this study is identifying the largest sources of bias and heterogeneity in the literature. This was particularly useful when examining outcomes that were associated with a significant level of statistical heterogeneity. Consequently after exclusion of the main studies that we identified as potential sources of heterogeneity (Karimzade et al., 2010; Mackens et al., 2020; Mahran et al., 2016; Polanski et al., 2014), the analysis was repeated. This approach successfully identified the studies that were causing marked heterogeneity in treatment effects across studies as evidenced by the significant decrease in statistical markers of heterogeneity after exclusion of these studies. However, the sensitivity analysis did not alter the overall findings and there remained no evidence for a significant improvement in fertility outcomes with the use of ES.
Some outcomes were difficult to assess, in particular ectopic pregnancy rates, where few studies reported this outcome, with smaller trials reporting more events. Pain scores were also difficult to assess – again, few studies reported this, and those that did reported this differently. Adverse events were often not recorded in the control arms of studies, limiting the comparisons that can be made to ‘usual’ IVF treatment. Due to time constraints, we were unable to contact all authors to collect information to aid the assessment of the risk of bias. For one trial included in the review (Polanski et al., 2014) we were unable to obtain the full-text article or correspond with the author in order to obtain data, and therefore, data for this trial were extracted from a recent systematic review (Vitagliano et al., 2019). As a result, a risk of bias assessment for this trial could not be undertaken. Key outcomes (e.g. miscarriage rate) were variably defined across the included trials – consensus regarding the definition of key fertility outcomes should be reached in order to ensure the results of future trials can be combined within meta-analyses.
The findings of this systematic review and meta-analysis conclusively confirm that there is no evidence that induced endometrial trauma improves IVF outcomes, including live birth and pregnancy rates, for women undergoing their first IVF cycle. We therefore recommend that the ES procedure is not undertaken in this population of women undergoing their first IVF cycle. Despite uncertainty in the effect of the ES on miscarriage and ectopic pregnancy outcomes, we recommend that further research is not undertaken, due to the ES not having a significant effect on live birth rates.
The data supporting this systematic review and meta-analysis are from previously reported studies and datasets, which have been cited. The processed data are available upon reasonable request to the corresponding author.
We would like to thank the authors that kindly provided data on request.
Thanks to Louisa Robinson for assisting with the screening of journal articles, and Pavithra Kumar and Anna Packham for assisting with risk of bias assessments.
MM, RC and DW conceived the systematic review and designed the study. RC and JH undertook screening and data extraction. MM wrote the first draft of the manuscript and undertook the meta-analysis. SW provided guidance. All authors approved of the final version.
This study is funded by the National Institute for Health Research (NIHR) [Health Technology Assessment Programme (project reference 14/08/45)]. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care. The funder did not have any role in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.
Disclosure of Interests
The authors declare no conflict of interest.
Details of Ethics Approval
Appendix A - MEDLINE Search Strategy
1 exp embryo transfer/ or exp fertilization in vitro/ or exp sperm injections, intracytoplasmic/
2 embryo transfer$.tw.
3 in vitro fertili?ation.tw.
5 (ivf or et).tw.
7 intracytoplasmic sperm injection$.tw.
8 (blastocyst adj2 transfer$).tw.
9 exp reproductive techniques, assisted/ or insemination, artificial/ or exp insemination, artificial, heterologous/ or exp insemination, artificial, homologous/
10 assisted reproducti$.tw.
13 (endometri$ adj5 scratch$).tw.
14 (endometri$ adj5 injur$).tw.
15 (endometri$ adj5 trauma$).tw.
16 (endometri$ adj5 biop$).tw.
17 (endometri$ adj5 harm$).tw.
18 (endometri$ adj5 damag$).tw.
19 (endometri$ adj5 inflammation).tw.
20 (endometri$ adj5 wound$).tw.
21 (endometri$ adj5 lesion$).tw.
22 (endometri$ adj5 insult$).tw.
24 12 and 23
26 randomized controlled trial.pt.
27 controlled clinical trial.pt.
30 clinical trials as topic.sh.
33 (crossover or cross-over or cross over).tw.
35 exp animals/ not humans.sh.
36 34 not 35
37 36 and 24
Barash, A., Dekel, N., Fieldust, S., Segal, I., Schechtman, E., Granot, I., 2003. Local injury to the endometrium doubles the incidence of successful pregnancies in patients undergoing in vitro fertilization. Fertil. Steril. 79, 1317–1322. https://doi.org/10.1016/S0015-0282(03)00345-5
Checa, M.A., 2013. Endometrial Scratching in In Vitro Fertilization Cycles With Oocyte Donation [WWW Document]. URL https://clinicaltrials.gov/show/NCT01842178 (accessed 3.12.21).
Deeks JJ, Higgins JPT, A.D., 2021. Chapter 10: Analysing data and undertaking meta-analyses. In: Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (editors) [WWW Document]. Cochrane Handb. Syst. Rev. Interv. version 6.2. URL https://training.cochrane.org/handbook/current/chapter-10 (accessed 8.24.21).
Eskew, A.M., Reschke, L.D., Woolfolk, C., Schulte, M.B., Boots, C.E., Broughton, D.E., Jimenez, P.T., Omurtag, K.R., Keller, S.L., Ratts, V.S., Odem, R.R., Jungheim, E.S., 2019. Effect of endometrial mechanical stimulation in an unselected population undergoing in vitro fertilization: futility analysis of a double-blind randomized controlled trial. J. Assist. Reprod. Genet. 36, 299–305. https://doi.org/https://dx.doi.org/10.1007/s10815-018-1356-5
Griesinger, G., 2016. Beware of the ’implantation rate’! Why the outcome parameter “implantation rate” should be abandoned from infertility research. Hum. Reprod. https://doi.org/10.1093/humrep/dev322
Higgins, J., Thomas, J., Chandler, J., Cumpston, M., Li, T., Page, M., Welch, V., 2021. Cochrane Handbook for Systematic Reviews of Interventions version 6.2 (updated February 2021) [WWW Document]. Cochrane. URL https://training.cochrane.org/handbook (accessed 7.5.21).
Higgins, J.P.T., Savović, J., Page, M.J., Sterne, J.A.C., 2019. RoB 2: A revised Cochrane risk-of-bias tool for randomized trials. Bmj.
Higgins, J.P.T., Thompson, S.G., 2002. Quantifying heterogeneity in a meta-analysis. Stat. Med. 21, 1539–1558. https://doi.org/10.1002/sim.1186
Higgins, J.P.T., Thompson, S.G., Deeks, J.J., Altman, D.G., 2003. Measuring inconsistency in meta-analyses. Br. Med. J. https://doi.org/10.1136/bmj.327.7414.557
Izquierdo Rodriguez, A., de la Fuente Bitaine, L., Spies, K., Lora, D., Galindo, A., 2020. Endometrial Scratching Effect on Clinical Pregnancy Rates in Patients Undergoing Egg Donor In Vitro Fertilization Cycles: the ENDOSCRATCH Randomized Clinical Trial (NCT03108157). Reprod. Sci. 1–10. https://doi.org/10.1007/s43032-020-00204-8
Karimzade, M.A., Oskouian, H., Ahmadi, S., Oskouian, L., 2010. Local injury to the endometrium on the day of oocyte retrieval has a negative impact on implantation in assisted reproductive cycles: A randomized controlled trial. Arch. Gynecol. Obstet. 281, 499–503. https://doi.org/10.1007/s00404-009-1166-1
Lensen, S., Osavlyuk, D., Armstrong, S., Stadelmann, C., Hennes, A., Napier, E., Wilkinson, J., Sadler, L., Gupta, D., Strandell, A., Bergh, C., Vigneswaran, K., Teh, W.T., Hamoda, H., Webber, L., Wakeman, S.A., Searle, L., Bhide, P., McDowell, S., Peeraer, K., Khalaf, Y., Farquhar, C., 2019. A Randomized trial of endometrial scratching before in vitro fertilization. N. Engl. J. Med. https://doi.org/10.1056/NEJMoa1808737
Lensen, S., Sadler, L., Farquhar, C., 2016. Endometrial scratching for subfertility: Everyone's doing it. Hum. Reprod. 31, 1241–1244. https://doi.org/10.1093/humrep/dew053
Liu, W., Tal, R., Chao, H., Liu, M., Liu, Y., 2017. Effect of local endometrial injury in proliferative vs. luteal phase on IVF outcomes in unselected subfertile women undergoing in vitro fertilization. Reprod. Biol. Endocrinol. 15, 75. https://doi.org/https://dx.doi.org/10.1186/s12958-017-0296-8
Mackens, S., Racca, A., Van de Velde, H., Drakopoulos, P., Tournaye, H., Stoop, D., Blockeel, C., Santos-Ribeiro, S., 2020. Follicular-phase endometrial scratching: a truncated randomized controlled trial. Hum. Reprod. 35, 1090–1098. https://doi.org/10.1093/humrep/deaa018
Maged, A.M., Rashwan, H., AbdelAziz, S., Ramadan, W., Mostafa, W.A.I., Metwally, A.A., Katta, M., 2018. Randomized controlled trial of the effect of endometrial injury on implantation and clinical pregnancy rates during the first ICSI cycle. Int. J. Gynecol. Obstet. 140, 211–216. https://doi.org/10.1002/ijgo.12355
Mahran, A., Ibrahim, M., Bahaa, H., 2016. The effect of endometrial injury on first cycle IVF/ICSI outcome: A randomized controlled trial. Int. J. Reprod. Biomed. (Yazd, Iran) 14, 193–8.
Metwally, M., Chatters, R., Dimairo, M., Walters, S., Pye, C., White, D., Bhide, P., Chater, T., Cheong, Y., Choudhary, M., Child, T., Drakeley, A., Evbuomwan, I., Gelbaya, T., Grace, J., Harris, P., Laird, S., da Silva, S.M., Mohiyiddeen, L., Pemberton, K., Raine-Fenning, N., Rajkhowa, M., Young, T., Cohen, J., 2021. A randomised controlled trial to assess the clinical effectiveness and safety of the endometrial scratch procedure prior to first-time IVF, with or without ICSI. Hum. Reprod. 36, 1841–1853. https://doi.org/10.1093/humrep/deab041
Metwally, M., Chatters, R., Pye, C., Dimairo, M., Walters., S., 2020. A Multicentre Randomised Controlled Trial of Induced Endometrial Scratch in Women Undergoing First Time In Vitro Fertilisation (IVF) - protocol [WWW Document]. https://doi.org/https://doi.org/10.15131/shef.data.12886997.v1
Nastri, C.O., Ferriani, R.A., Raine-Fenning, N., Martins, W.P., 2013. Endometrial scratching performed in the non-transfer cycle and outcome of assisted reproduction: A randomized controlled trial. Ultrasound Obstet. Gynecol. 42, 375–382. https://doi.org/10.1002/uog.12539
Nastri, C.O., Gibreel, A., Raine-Fenning, N., Maheshwari, A., Ferriani, R. a, Bhattacharya, S., Martins, W.P., 2012. Endometrial injury in women undergoing assisted reproductive techniques. Cochrane database Syst. Rev. 7, CD009517. https://doi.org/10.1002/14651858.CD009517.pub2
Page, M.J., McKenzie, J.E., Bossuyt, P.M., Boutron, I., Hoffmann, T.C., Mulrow, C.D., Shamseer, L., Tetzlaff, J.M., Akl, E.A., Brennan, S.E., Chou, R., Glanville, J., Grimshaw, J.M., Hróbjartsson, A., Lalu, M.M., Li, T., Loder, E.W., Mayo-Wilson, E., McDonald, S., McGuinness, L.A., Stewart, L.A., Thomas, J., Tricco, A.C., Welch, V.A., Whiting, P., Moher, D., 2021. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ. https://doi.org/10.1136/bmj.n71
Piaget‐Rossel, R., Taffé, P., 2019. Meta‐analysis of rare events under the assumption of a homogeneous treatment effect. Biometrical J. 61, 1557–1574. https://doi.org/10.1002/bimj.201800381
Polanski, L., Gov, N.R.-F.-C., 2014, U., 2014. A study to assess if scratching the lining of the womb prior to IVF treatment increases the chances of pregnancy [WWW Document].
Potdar, N., Gelbaya, T., Nardo, L.G., 2012. Endometrial injury to overcome recurrent embryo implantation failure: a systematic review and meta-analysis. Reprod. Biomed. Online 25, 561–571. https://doi.org/10.1016/j.rbmo.2012.08.005
Vitagliano, A., Andrisani, A., Alviggi, C., Vitale, S.G., Valenti, G., Sapia, F., Favilli, A., Martins, W.P., Raine-Ferring, N., Polanski, L., Ambrosini, G., 2019. Endometrial scratching for infertile women undergoing a first embryo transfer: a systematic review and meta-analysis of published and unpublished data from randomized controlled trials. Fertil. Steril. https://doi.org/10.1016/j.fertnstert.2018.12.008
Yeung, T.W.Y., Chai, J., Li, R.H.W., Lee, V.C.Y., Ho, P.C., Ng, E.H.Y., 2014. The effect of endometrial injury on ongoing pregnancy rate in unselected subfertile women undergoing in vitro fertilization: A randomized controlled trial. Hum. Reprod. 29, 2474–2481. https://doi.org/10.1093/humrep/deu213
Mr Metwally is a Consultant Gynaecologist in the NHS. His main areas of interest are reproductive medicine, fertility treatment and reproductive surgery. He was the Chief Investigator of the recently completed Endometrial Scratch (ES) Trial - an NIHR funded trial to assess the effectiveness of the ES prior to first cycle IVF.
There is no evidence that the ES procedure, undertaken prior to the first cycle of IVF, improves pregnancy outcomes, including LBR. Clinicians are recommended not to perform this procedure in individuals undergoing their first cycle of IVF.