Endometrial compaction to predict pregnancy outcomes in patients undergoing assisted reproductive technologies: a systematic review and meta-analysis.

Increasing the live birth rate (LBR) of patients undergoing ART is the ultimate goal of any healthcare professional working within reproductive medicine. For successful implantation to take place, synchrony between both an embryo and a receptive endometrium is an essential requirement ( Casper, 2020 ). The endometrium usually becomes receptive during the mid-secretory phase for a narrow period of time, commonly referred to as the ‘window of implantation’ (WOI) ( Abdallah et al. , 2012 ). During this time, there is profound architectural remodelling, alongside transcriptional and secretional alterations within the endometrium, in response to the rise in circulating serum progesterone ( Rosario et al. , 2003 ; Wang et al. , 2020 ). Endometrial receptivity has been defined as a key factor in influencing IVF success, and an abnormal or displaced WOI has been proposed as a possible cause for recurrent implantation failure (RIF) and recurrent miscarriage (RM) ( Casper, 2020 ). Pelvic ultrasound provides an accessible, cost-effective, and non-invasive method of assessing the endometrium ( Bourne et al. , 1997 ). Endometrial assessment using ultrasound technology has therefore been proposed as a possible alternative method of predicting endometrial receptivity and successful pregnancy ( Craciunas et al. , 2019 ). First described by Haas et al. in 2019, studies now describe the occurrence of endometrial compaction (EC) in a subset of patients, referring to the progesterone-induced absolute decrease in endometrial thickness (EMT) seen between the end of the proliferative or oestrogenic phase and the day of embryo transfer (ET) ( Haas et al. , 2019 ; Casper, 2020 ). As a new concept, little is known about EC, however, it is speculated to occur because of rising progesterone levels following ovulation, resulting in cessation of endometrial proliferation, and increased endometrial glandular development, immune cell proliferation, and angiogenesis, thereby increasing the density but not the volume of the endometrium ( Fleischer et al. , 1984 ; Tabibzadeh, 1990 ; Bassil, 2001 ). Ultrasound follow-up of natural menstrual cycles shows that the EMT reaches a peak just before ovulation and then either plateaus or thins, giving rise to EC ( Youngster et al. , 2022 ). Some studies looking at EC have attempted to correlate it with serum oestradiol and progesterone levels, however, poor correlation exists ( Jarrah et al. , 2021 ; Olgan et al. , 2022 ; Youngster et al. , 2022 ; Ju et al. , 2023 ). Endometrial progesterone receptor deficiency, or resistance, may explain differences in EC among different patients/cycles as serum progesterone levels are not necessarily consistent with those within endometrial tissue ( Usadi et al. , 2008 ; Lawrenz and Fatemi, 2022 ). In recent years, several studies have tried to determine if EC is linked to reproductive outcomes with inconsistent evidence published ( Haas et al. , 2019 ; Zilberberg et al. , 2020 ; Riestenberg et al. , 2021 ; Yaprak et al. , 2021 ; Shah et al. , 2022 ). Some authors report no correlation between EC and pregnancy outcomes ( Huang et al. , 2020 , 2021 ; Jarrah et al. , 2021 ; Riestenberg et al. , 2021 ; Gursu et al. , 2022 ; Shah et al. , 2022 ), whilst others have observed a positive association between EC and pregnancy rates ( Haas et al. , 2019 ; Zilberberg et al. , 2020 ; Kaye et al. , 2021 ; Yaprak et al. , 2021 ; Youngster et al. , 2022 ). If shown to be beneficial, EC could be a valuable way to help predict pregnancy outcomes in patients undergoing ART and could be an inexpensive method of guiding the timing of ET to synchronise with endometrial receptivity. The primary aim of this study is to evaluate the association between EC and LBR. Within this systematic review and meta-analysis, we also aimed to robustly review and provide an up-to-date summary of the currently available evidence on the effect of EC on reproductive outcomes and the prevalence of EC within the sub-fertile population.

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) and Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines ( Stroup et al. , 2000 ; Page et al. , 2021 ). The protocol was prospectively written and registered with PROSPERO (registration number: CRD42022378464). The protocol did not require any revisions during the study period. A comprehensive literature search was conducted by two independent authors (H.A. and K.S.) for all studies published from inception to May 2023. PubMed, Ovid MEDLINE, Science Direct, Scopus, Embase, Web of Science, Cochrane Library, and Google Scholar databases were searched. The search strategy included the following Medical Subject Heading (MeSH) terms, keywords, and their combinations: ‘endometrium’ OR ‘endometrial’ AND ‘compaction’ OR ‘thickness’ AND ‘in vitro fertilisation’ OR ‘assisted reproductive techniques’. Additionally, database searches were supplemented with manual forward and backward citation chaining, and the ‘similar articles’ feature was searched on PubMed. Review articles were utilised to ensure all relevant citations were identified and included. Duplicate articles were deleted, and the remaining articles were uploaded to Rayyan, an electronic review software ( Ouzzani et al. , 2016 ) available from: https://www.rayyan.ai/ . Two independent reviewers (H.A. and K.S.) performed title and abstract screening according to the pre-determined eligibility criteria. All original randomised and non-randomised studies assessing EC in association with any pregnancy outcomes in patients undergoing ART in the form of IVF/ICSI-ET cycles, including the transfer of either fresh or frozen embryos, were included in this review. Studies were excluded if: (i) they did not report on EMT or EC and any associated pregnancy outcome, (ii) they did not include study participants that had ART in the form of IVF/ICSI-ET cycles, (iii) were not written in the English language, (iv) were not full-text articles (including abstracts and incomplete datasets), and (v) were not original research studies (including review articles, meta-analyses, case-reports and conference abstracts). The remaining articles were subject to independent full-text review by the same two authors. In case of any disagreements, a third reviewer (N.T.) was consulted for resolution and discussed between all reviewers. A standardised spreadsheet was developed and agreed upon between the authors. Selected studies were comprehensively examined, data extracted and recorded into the spreadsheet by H.A. and K.S. and then cross-checked by H.A. The data recorded included: author, year of publication, country of study, study aims, study design, sample size, experimental methods, outcome measures within the experimental group (EC) and comparator (no endometrial compaction (no EC)), study definitions of EC and outcomes parameters, and study conclusions. The primary outcome measure of this systematic review and meta-analysis was LBR. Secondary outcomes comprised of other pregnancy outcomes including positive pregnancy test (PPT), clinical pregnancy rate (CPR), ongoing pregnancy rate (OPR) miscarriage rate (MR), and endometrial compaction rate (ECR). Implantation rate (IR) and ectopic pregnancy rate (EPR) are not included as outcomes in this study due to a lack of studies reporting these outcomes. Definitions of pregnancy outcomes vary across the different studies. However, they are broadly defined as; PPT: either urinary or serum β-hCG detected at least five days following ET, IR: ratio of intrauterine gestation sac (IUGS) number over the number of embryos transferred, CPR: at least one IUGS detected with or without a foetal heartbeat (FH) at <12 weeks' gestation (where gestation was defined), OPR: viable pregnancy ≥12 weeks gestation, LBR: live birth of a fetus at least 22 weeks' gestation, MR: pregnancy loss <22 weeks' gestation (where gestation was defined), EPR: at least one gestation sac seen outside of the uterus. Further specific definitions of all pregnancy outcomes used within each study are outlined in Table 1 . Currently, there is no universally accepted definition of EC; therefore, within this study overall, we defined EC as any decrease in EMT between the end of the oestrogen phase and the day of ET, allowing us to include all available study data on EC. Where possible, sub-analysis of different levels of EC and pregnancy outcomes was performed, using studies which reported on different degrees of EMT decrease/EC, to understand the impact on the results. Sub-group analysis of pregnancy outcomes at ≥5% EC, ≥10% EC, and ≥15% EC was performed. Corresponding authors were contacted where necessary if further clarity or data was required during the data extraction stage. Pregnancy outcomes measured by all studies, including study definition of outcome. ET, embryo transfer; FH, fetal heartbeat; FET, frozen embryo transfer; GS, gestation sac; IUGS, Intrauterine gestation sac; IUP, intrauterine pregnancy; LB, live birth; US, ultrasound; β-hCG, beta human chorionic gonadotrophin. Results were statistically analysed with the aid of Review Manager (RevMan), Version 5.4, The Cochrane Collaboration, 2020. Relevant findings were summarised and discussed between all the authors allowing overall themes and conclusions to be drawn. Random effect models were used for meta-analyses unless the Higgins I 2 statistic was <50%, indicating more homogenous datasets, in which case a fixed effect model was used. Meta-analyses are presented as forest plots. The Mantel–Haenszel method was applied for pooling of dichotomous data, and results were presented as odds ratio (OR) with 95% CIs ( Mantel and Haenszel, 1959 ). An OR >1 for any outcome indicated a result in favour of the EC group and conversely, an OR <1 indicated a result in favour of no EC. Pooled weighted proportions of EC versus no EC across the studies were calculated using a random-effects model and the ‘metaprop’ command in Stata, version 14, StataCorp ( Nyaga et al. , 2014 ). Summary proportions are presented with their corresponding 95% CIs. An overall effect P -value <0.05 was considered as statistically significant. Sensitivity analysis was performed by using the leave-one-out method for the primary outcome (LBR). In order to assess the quality of all studies included, a risk of bias assessment was performed using the Newcastle-Ottawa scale (NOS) ( Wells et al. , 2000 ). Each study was scored independently by two reviewers (H.A. and K.S.) between 0 and 9 stars based on three main areas: selection, comparability, and outcome, resulting in an overall quality assessment of ‘good’ (3–4 stars in the selection domain and 1–2 stars in comparability domain and 2–3 stars in the outcome domain), ‘fair’ (2 stars in the selection domain and 1–2 stars in comparability domain and 2–3 stars in the outcome domain), or ‘poor’ (0–1 stars in the selection domain or 0 stars in the comparability domain or 0–1 stars in the outcome domain). Any discrepancies were discussed and resolved by a third author (N.T.).

A total of 6192 records were identified from the database searches and six by citation searching ( Fig. 1 , PRISMA diagram). Following the removal of duplicate records, 4024 records were eligible for title and abstract screening. An additional six studies were identified using forward and backward chaining. A total of 33 papers underwent full-text screening. Three studies were excluded as they had a discordant theme and two were excluded as they were conference abstracts only. One further study was excluded from the final analysis following data extraction, as they did not report an overall cohort size for EC and no EC, and we were unable to obtain this information when contacting the authors ( Li et al. , 2022 ). PRISMA flowchart demonstrating the selection of publications identified in the systematic review and meta-analysis. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analyses. Twenty-one studies were included within the final analysis, with a total of 27 857 patients. The studies comprised of seven prospective and 14 retrospective observational studies, with no randomised controlled trials (RCTs) available on the topic. All studies were published over a 3-year period between 2019 and 2023. A summary of the study characteristics is provided in Table 2 . Further details regarding pregnancy outcomes and study methods are provided in Tables 1 and 3 . Summary of the study characteristics. FET, frozen embryo transfer. Details of study design. CPR, clinical pregnancy rate; EC, endometrial compaction; EMT, endometrial thickness; ET, embryo transfer; FET, frozen embryo transfer; GnRHa, gonadotropin-releasing hormone agonist; LBR, live birth rate; OD, oocyte donor; OPR, ongoing pregnancy rate; PGT-A, pre-implantation genetic testing for aneuploidy; PGT-M, pre-implantation genetic testing for monogenic disorders; PGT, pre-implantation genetic testing; RIF, recurrent implantation failure; RM, recurrent miscarriage; TVS, transvaginal ultrasound; US, ultrasound. The NOS was used to perform a quality assessment on all studies included in this meta-analysis ( Table 4 ). All studies were deemed to be of good quality following assessment, indicating reliable data. Risk-of-bias assessment for cohort studies using the Newcastle-Ottawa scale (NOS). Risk-of-bias assessment for cohort studies using the NOS. Selection: (1) representative of exposed cohort, (2) selection of non-exposed cohort, (3) ascertainment of exposure, (4) demonstration that outcome of interest was not present at the start of the study. Comparability: (5) study controls for age, (6) study controls for any additional factors. Outcome: (7) assessment of outcome (8) was follow-up long enough for outcomes to occur (9) adequacy of follow-up of cohorts. Ten studies reported on LBR (2812 in EC group; 8898 in no EC group), showing no significant difference in LBR between the two groups (OR 0.95; 95% CI 0.87–1.04; P  =   0.27) ( Fig. 2A ). When sub-analysing this data based on level of EC, this also showed no significant difference in LBR between ≥5% EC and no EC (OR 0.58; 95% CI 0.29–1.19; P  =   0.14), ≥10% EC and no EC (OR 0.60; 95% CI 0.35–1.02; P  =   0.06) or ≥15% EC and no EC (OR 0.40; 95% CI 0.07–2.15; P  =   0.29). The same result was observed when sub-analysing the four studies that included euploid embryos only (OR 0.95; 95% CI 0.72–1.25; P  =   0.71) ( Fig. 2B ). Sub-analysis of fresh ET (three studies) versus FET cycles (seven studies) also showed no significant differences in LBR regardless of ET protocol (fresh ET cycles: OR 0.93; 95% CI 0.77–1.11; P  =   0.41 ( Fig. 2C ); FET cycles: OR 1.12; 95% CI 0.78–1.62; P  =   0.54) ( Fig. 2D ). Further sub-analysis of FET cycles only, based on whether they were natural cycle (NC) FET or HRT FET cycles, also showed no significant differences in LBR between EC and no EC for either cycle type (NC FET cycles: OR 0.90; 95% CI 0.74–1.10; P  =   0.31; HRT FET cycles: OR 0.98; 95% CI 0.87–1.11; P  =   0.75). Sensitivity analysis using the leave-one-out approach presented similar results, with no single paper found to alter the results significantly. Forest plots to show meta-analysis of LBR between EC versus no EC groups. ( A ) LBR between EC versus no EC. ( B ) LBR in PGT studies only. ( C ) LBR in fresh cycles only. ( D ) LBR in FET cycles only. LBR, live birth rate; EC, endometrial compaction; PGT, pre-implantation genetic testing; FET, frozen embryo transfer. Seven of the included studies reported on PPT (631 in EC group; 2540 in no EC group). No significant difference was demonstrated between EC and no EC groups (OR 1.54; 95% CI 0.97–2.45; P  =   0.07) ( Supplementary Fig. S1A ). Sub-analysis of FET cycles only included five studies (324 in EC group; 1617 in no EC group) which showed no significant difference between the groups (OR 2.27; 95% CI 0.78–6.61; P  =   0.13) ( Supplementary Fig. S1B ). Four studies were further sub-analysed as they included medicated HRT FET cycles only, providing a more homogeneous cohort. HRT FET cycles showed a trend towards a higher PPT in the EC group although still not passing the level of statistical significance (OR 3.06, 95% CI 1.01–9.23; P  =   0.05) ( Supplementary Fig. S1C ). Sixteen studies reported on CPR (8189 in EC group; 16 042 in no EC group). No statistically significant difference in CPR was found between the two groups (OR 1.19; 95% CI 0.99–1.43; P  =   0.06) ( Supplementary Fig. S2A ). No significant difference was found in the sub-group analysis of this data for CPR between ≥5% EC and no EC (OR 1.19; 95% CI 0.84–1.70; P  =   0.33), ≥10% EC and no EC (OR 1.11; 95% CI 0.85–1.43; P  =   0.45) or ≥15% EC and no EC (OR 1.37; 95% CI 0.69–2.70; P  =   0.37). Definitions of CPR varied significantly across the studies. One study did not define CPR within the manuscript ( Bu et al. , 2019 ). A list of the definitions used within the remaining 15 studies is provided in Table 1 . To further increase homogeneity, additional sub-analysis of 14 FET cycle-only studies was performed (3968 in EC group; 7181 in no EC group), showing significantly higher CPR in the EC group compared to the no EC group (OR 1.29; 95% CI 1.04–1.61; P  =   0.02) ( Supplementary Fig. S2B ). Further analysis of FET cycles showed that CPR was significantly greater within the EC group on the hormone replacement therapy (HRT)-FET cycle protocol (OR 1.33; 95% CI 1.06–1.67; P  =   0.01) but not in those with EC having natural cycle (NC) FETs (OR 1.31; 95% CIs 0.58–2.92; P  =   0.52). To ensure differences in CPR were not being altered due to heterogeneity within this group, sub-analysis was performed on a more homogenous cohort that defined CPR as a viable intrauterine pregnancy with FH seen on ultrasound scan at less than 12 weeks’ gestation, where gestational age was defined. Eight studies were included in this cohort, and CPR was still found to be statistically in favour of EC (OR 1.83; 95% CI 1.15–2.92; P  =   0.01) ( Supplementary Fig. S2C ). Additionally, sub-group analysis of seven FET cycle-only studies with this same definition of CPR (425 in EC group; 690 in no EC group) also showed a statistically significantly CPR in favour of the EC group (OR 2.08; 95% CI 1.28–3.39; P  =   0.003) ( Supplementary Fig. S2D ). Again, this significant improvement in CPR within the EC FET group seemed to be driven by those on HRT-FET cycles (OR 1.73; 95% CI 1.03–2.92; P  =   0.04) and not those having NC-FETs (OR 3.35; 95% CI 0.82–13.76; P  =   0.09); however, only two studies could be included in the NC-FET meta-analysis, therefore, this finding should be viewed with caution. Sub-set analysis was performed on four studies which used pre-implantation genetic testing (PGT) for aneuploidy (PGT-A). No significant difference was seen in the CPR between the two groups (OR 0.83; 95% CI 0.63–1.10; P  =   0.2). These studies had varying definitions of CPR ( Supplementary Fig. S2E ). Eight studies reported OPR as an outcome (1333 In EC group; 5118 in no EC group). This outcome was significantly in favour of EC (OR 1.61; 95% CI 1.09–2.38; P  =   0.02) ( Supplementary Fig. S3A ). When sub-analysing this data based on level of EC, similarly, a significant difference was found in OPR between ≥5% EC and no EC (OR 1.87; 95% CI 1.05–3.34; P  =   0.03) and ≥15% EC and no EC (OR 3.52; 95% CI 2.36–5.23; P  <   0.00001), both in favour of EC. OPR was found to be in favour of EC when sub-analysing EC at a level of ≥10%, however, this was not found to reach the level of significance (OR 1.50; 95% CI 0.95–2.36; P  =   0.08). Sub-set analysis of FET cycles included seven studies and confirmed significantly higher OPR in the EC group (OR 1.79; 95% CI 1.06–3.02; P  =   0.03) ( Supplementary Fig. S3B ), particularly within the HRT-FET cycles (OR 1.87; 95% CI 1.38–2.55; P  <   0.0001). Sub-set analysis of euploid embryos only was not possible as only one study that reported OPR used PGT-A ( Zilberberg et al. , 2020 ). Twelve papers reported on MR as an outcome. MR was not significantly different between the EC and no EC groups (OR 1.06; 95% CI 0.89–1.25; P  =   0.53) ( Supplementary Fig. S4A ). This result remained when embryos were tested to be euploid prior to transfer (OR 0.63; CI 0.38–1.07; P  =   0.09) ( Supplementary Fig. S4B ). Sub-analysis based on fresh ET (2 studies) and FET cycles (10 studies) confirmed no significant differences in MR (fresh ET cycles: OR 0.78; 95% CI 0.78–1.62; P  =   0.54 ( Supplementary Fig. S4C ); FET cycles: OR 1.04; 95% CI 0.86–1.26; P  =   0.7) ( Supplementary Fig. S4D ). Across all 21 studies, the pooled weighted prevalence of EC was 32% (9283 patients, 95% CI 26–38%) versus no EC at 68% (18 574 patients). This suggests that within the general population having IVF/ICSI treatment, EC occurs in a third of patients ( Fig. 3 ). Forest plot presenting a meta-analysis of the prevalence of EC across all included studies. EC, endometrial compaction.

This meta-analysis includes a total of 27 857 patients undergoing IVF/ICSI treatments across 21 recent research studies and demonstrated that EC is associated with a significant improvement in CPR and OPR but not LBR. Our analysis has also shown the population prevalence of EC to be 32% of patients having IVF/ICSI treatment. However, the positive association of EC for some early pregnancy outcomes did not translate to increased LBR. Therefore, based on current available evidence, stratification of women according to EC cannot yet be justified within clinical practice. Reassuringly, data has shown that EC is not associated with a higher MR or lower PPT or LBR and therefore, is not a negative predictor for favourable pregnancy outcomes. However, our data highlights the assessment of EC as an important and promising area of focus for future studies. Our findings have shown that LBR is similar between cycles demonstrating EC and cycles without evidence of EC. These results were found to be consistent when sub-analysing for different definitions of EC based on the degree of EMT decrease, with no significant differences between ≥5%, ≥10, or ≥15% EC and no EC. Additionally, no significant difference in LBR was found within PGT-A euploid ET cycles, fresh ET, or frozen (natural or HRT) ET cycles. In keeping with some studies, we also report significantly increased CPR ( Jarrah et al. , 2021 ; Yaprak et al. , 2021 ; Youngster et al. , 2022 ; Ju et al. , 2023 ) and OPR with EC ( Haas et al. , 2019 ; Zilberberg et al. , 2020 ; Kaye et al. , 2021 ; Youngster et al. , 2022 ). Of the 21 studies we included, only two reported on both OPR and LBR ( Huang et al. , 2020 , 2021 ) and only four reported on both CPR (defined as a viable intrauterine pregnancy at less than 12 weeks’ gestation) and LBR ( Riestenberg et al. , 2021 ; Yaprak et al. , 2021 ; Gursu et al. , 2022 ; Shah et al. , 2022 ), thus, CPR/OPR and LBR results cannot be directly compared due to the large discrepancy in studies included in the meta-analyses and the amount of heterogeneity between the study protocols. However, additional factors to consider that may account for the discrepancy between CPR/OPR and LBR results are obstetric complications that may lead to loss later in pregnancy, and the adequacy of the study power calculation. Due to the natural cumulative loss of pregnancies between PPT through to LBR, a larger sample size is required to achieve a comparable number of live births to clinical/ongoing pregnancies. Within this meta-analysis, we see that a total of 11 149 clinical pregnancies (defined as a viable IUP <12 weeks) were included within the analysis and only 4690 live births, which may also contribute to the discrepancy in the translation of results. The lack of consistency with outcome definitions, cycle protocols, and definitions of EC makes it difficult to draw generalised conclusions from this limited data, and attempting to do so, could potentially lead to spurious results. Only Zilberberg et al. , (2020) reported on OPR in PGT-A euploid embryos, precluding a sub-group analysis for this outcome. Haas and Kaye et al. included a mix of tested and untested embryos and therefore could not be included in PGT-A sub-group analysis of OPR ( Haas et al. , 2019 ; Kaye et al. , 2021 ). Analysis of FET cycles showed significantly better OPR and CPR within the EC versus no EC group, regardless of the definition used for CPR. Sub-group analysis of HRT-FET cycles and NC-FET cycles showed that this significance was likely being driven from the HRT-FET cycles. HRT and natural FET cycles are very different, with natural cycles relying on endogenous production of oestradiol and progesterone (with/without additional luteal supplementation dependent on cycle protocol) and HRT cycles relying on complete artificial endometrial preparation. This difference may be attributed to a difference in serum oestradiol/progesterone levels or because of the comparatively few studies which include NC-FET cycles. Overall, the studies do consistently show that EC is not a negative predictor for pregnancy outcomes. However, according to our findings, it cannot yet be used as a prognostic tool for LBR. Within this systematic review and meta-analysis, a robust search strategy was implemented using multiple databases and search strategies to ensure the inclusion of all available literature. Where data was unclear or lacking within the manuscripts, authors were contacted directly to supply further clarification. Our protocol was prospectively registered with PROSPERO and performed in accordance with both PRISMA and MOOSE guidelines, ensuring a rigorous study design and high-quality data output. Multiple meta-analyses including subgroup analysis were performed, to ensure the homogenisation of the data analysis as much as possible. An important limitation of this study is the wide range of heterogeneity identified between the study protocols, including fresh (IVF/ICSI/autologous versus oocyte donor) versus FET cycles (either medicated or natural), different inclusion/exclusion criteria, variation between chosen pregnancy outcomes reported and the definitions used for each pregnancy outcome, study definition of EC, luteal support regimes, method of measuring the EMT (transabdominal (TA) ultrasound versus transvaginal (TV) ultrasound), use of PGT-A, day of ET, and number of embryos transferred. Studies reporting on different pregnancy outcomes as their endpoints, and using different medical protocols, prevented/precluded drawing reliable conclusions from direct comparisons of pooled data. Additionally, 14 out of the 21 studies included were performed as retrospective analyses. Currently, there is no universally agreed definition for EC. Out of the 21 studies, four defined EC as any decrease in EMT, seven defined EC as > or ≥5% decrease in EMT, four defined EC as ≥10% decrease in EMT, one defined EC as ≥15% decrease in EMT and five did not define EC and therefore it is likely that they used any decrease in EMT as evidence of EC. Due to the lack of definition and studies on the topic, we felt it was important to include all available data on EC within this study, therefore, we used a definition of any decrease in EMT as EC. Whilst we recognise that this increases the heterogeneity in the no EC group, we believe that any significant results obtained would only be further enhanced by greater levels of EC. Additionally, by analysing all available data, this will help plan future studies and develop a definition of EC. Where possible, we sub-analysed the results into different definitions of EC, including ≥5% EC, ≥10% EC, and ≥15% EC, which did not appear to significantly alter the results, and found no change to the OPR or LBR results. When categorising EMT change within the studies, some authors separated the groups into EC and no EC, whilst others separated the no EC group further into unchanged EMT and increased EMT. For the purpose of this study, we compared two groups, EC (defined as any decrease in EMT) and no EC (including those who had no change in EMT and those who had an increase in EMT). Another important limitation to consider is the method of ultrasound used, the ultrasound technician, and the day of the ultrasound scan. Nine studies used TV ultrasound to measure EMT at all points throughout the cycle. Nine studies used TV ultrasound to measure EMT at the end of the oestrogenic phase and then TA ultrasound to measure EMT around the time of ET. Gursu et al . (2022) used TA ultrasound to measure EMT on the first day of progesterone administration and again on the day of ET. Two studies did not specify which method of ultrasound was used. The sensitivity of TV ultrasound is recognised to be superior to TA ultrasound and changes between ultrasound methodology may have introduced intra-observer variability even when performed by the same sonographer, however, this more accurately represents real-world clinical scenarios where it is very common for cycle tracking to be performed under TV ultrasound and TA ultrasound to be used on the day of ET. In some studies, different ultrasound technicians were used for the different scans, increasing the probability of inter-observer variability. Some studies mitigated this by having the EMTs checked by independent practitioners. Due to the well-recognised margin of error that exists when measuring EMT, for future studies, we would advise that having an EC change of at least ≥5% EMT would mitigate this. Outcome reporting in reproductive medicine studies is a wider and long-standing area of debate ( Clarke et al. , 2010 ; Barnhart, 2014 ; Braakhekke et al. , 2014 ; Gadalla et al. , 2018 ). Different research groups have varying opinions on what pregnancy outcome is the most meaningful within a trial setting ( Clarke et al. , 2010 ; Barnhart, 2014 ; Braakhekke et al. , 2014 ; Gadalla et al. , 2018 ). In 2003, ESHRE recommended that the outcome measure for ART and non-ART should be ‘singleton live birth rate’ ( Land and Evers, 2003 ). Some argue that OPR serves as a better primary outcome, as this eliminates confounding factors such as second-trimester loss, stillbirth, multiple pregnancies, and the number of embryos transferred, hence why occasionally some studies report on the IR as an outcome ( Dickey et al. , 2004 ). Additional arguments against LBR include the need for a larger sample size, as LBR is lower than other pregnancy outcomes; the need for longer trials, increasing the risk of patients being lost to follow-up, increasing costs and delaying results and difficulties due to fragmentation between gynaecological and obstetric care ( Braakhekke et al. , 2014 ). However, due to the inconsistencies in definitions, even other pregnancy outcomes such as biochemical pregnancy rate, CPR or OPR may fall into different categories relevant to different studies, depending on the specific gestational cut-off used. Therefore, until outcomes are reported homogeneously, all pooled data should be viewed with an element of caution ( Barnhart, 2014 ). Having a live birth is the reason why sub-fertile patients seek ART, and therefore LBR should be the outcome of focus and should be reported on in every trial performed within reproductive medicine. Standardising outcome definitions is essential to ensure that we are producing reliable, meaningful, and translatable data for use in clinical practice and, thus, ultimately, we can provide better care to our patients.

EC may significantly improve CPR and OPR, although this is not yet seen to translate to LBR. This may be masked due to the heterogeneity between the studies or due to a lack of currently available data. However, for patients, LBR is understandably the most important pregnancy outcome, and therefore currently, EC should not be used to stratify patients in clinical practice. Nevertheless, EC may hold promise for the future as a non-invasive predictor of positive pregnancy outcomes and, therefore, should be a focus for a future clinical trial, including LBR as the primary outcome. We recommend a large prospective multicentre clinical trial with a robust protocol, ensuring minimal cycle variability, clear definitions for EC and pregnancy outcomes, and TV sonography to be performed throughout by highly trained ultrasound technicians and reviewed by independent technicians according to pre-determined criteria. If shown to be of benefit, EC has many advantages over currently available invasive and costly methods of testing for endometrial receptivity, which have little robust evidence to support them. Additionally, in patients with RIF, where many cases remain unexplained, EC could act as a useful tool to aid appropriate counselling and management of patients for a successful pregnancy.