Diagnostic accuracy of sliding sign for detecting pouch of Douglas obliteration and bowel involvement in women with suspected endometriosis: systematic review and meta‐analysis

The systematic review and meta‐analysis was performed according to preferred reporting items for systematic reviews and meta‐analyses (PRISMA) and synthesizing evidence from diagnostic accuracy tests (SEDATE) guidelines 14 , 15 . Inclusion and exclusion criteria and methods for data extraction and quality assessment were specified in advance. The protocol was registered with PROSPERO (CRD42021290671) and is available in Appendix  S1 . No amendment was made after registration. Institutional review board approval was waived owing to the nature and design of the study. Studies published between January 2000 and October 2021 were identified by two authors (E.T. and C.M.) using PubMed/MEDLINE, Web of Science, CINAHL, The Cochrane Library, ClinicalTrials.gov and SCOPUS databases to identify potentially eligible studies. The search terms were as follows: ‘endometriosis’, ‘pouch of Douglas’, ‘bowel’, ‘recto‐sigmoid’, ‘rectal’ and ‘sliding’. Language restriction in the search was set to English, French and Spanish. Three authors (P.M.E., P.F. and J.L.A.) screened the titles and abstracts of the identified studies to exclude articles that were not relevant to the topic under review, such as those focusing on magnetic resonance imaging instead of ultrasound as the diagnostic method, as well as reviews, letters and case reports. Full‐text articles were obtained to identify eligible studies, and reviewers applied independently the following inclusion criteria: (1) prospective cohort design with at least 20 women included (sample size was set arbitrarily); (2) premenopausal women with a clinical suspicion of endometriosis included as participants; (3) TVS performed by an expert or trained gynecologist used as the index test; (4) laparoscopy (visual inspection with or without histological diagnosis) used as the reference standard; (5) sufficient data reported to construct a 2 × 2 table of diagnostic performance. The ‘snowball strategy’ was used to identify relevant papers by reviewing the reference lists of the papers selected for full‐text review. In the case of missing relevant data, we sought to contact the authors to ask for more information. For studies by the same research group, the time period of patient recruitment was examined. If we detected at least two studies from the same group with a clear overlap or a potential risk of overlap of patients, the most recent study was selected for analysis. The Patients, Intervention, Comparator, Outcomes and Setting (PICOS) criteria were used to describe the included studies (Table  1 ). Characteristics of studies included in systematic review and meta‐analysis, according to Patients, Intervention, Comparator, Outcomes and Setting (PICOS) criteria Only first author of each study is given. LPS, laparoscopy; NS, not stated; PoD, pouch of Douglas; Prosp, prospective; TVS, transvaginal ultrasound. Diagnostic accuracy results and additional useful information about patients and procedures were retrieved from selected primary studies independently by three authors (P.M.E., P.F. and J.L.A). Any disagreement regarding study selection and data collection was resolved by consensus among the three authors. Quality assessment was conducted using the Quality Assessment of Diagnostic Accuracy Studies‐2 (QUADAS‐2) tool adapted to this systematic review. The QUADAS‐2 tool includes four domains: patient selection, index test, reference standard and flow and timing. For each domain, the risk of bias and concerns regarding applicability were classified as high, low or unclear. The results of the quality assessment were used to evaluate the overall quality of included studies and investigate potential sources of heterogeneity. Three authors (P.M.E., P.F. and J.L.A) studied independently the methodological quality using a standard form with quality assessment criteria and a flow diagram. Disagreements were resolved by discussion among the three authors until a consensus was reached. The risk of bias in the patient‐selection domain was determined based on the description of inclusion and exclusion criteria of the studies. Patient selection was considered to be at high risk of bias if studies included a non‐consecutive or non‐random series of patients and performed inappropriate exclusions (for example, excluding patients with poor imaging). The index‐test domain was assessed based on the description of the technique of the sliding‐sign assessment. The risk of bias was considered low when the sliding‐sign technique was described in detail. The reference‐standard domain was evaluated based on the method used in the study to diagnose obliteration of the pouch of Douglas and/or bowel involvement. The correct reference standard was considered to be laparoscopic surgical and/or histological findings. A lack of blinding of surgeons to ultrasound findings was not considered to indicate a high risk of bias. For the flow‐and‐timing domain, a description of the time elapsed from the index test to the reference‐standard assessment was evaluated. An interval of more than 3 months was considered to indicate a high risk of bias. Data on the diagnostic performance of the sliding‐sign test performed during TVS were extracted or derived. A positive test (negative sliding sign) was defined as the absence of sliding between the anterior rectum and the serosa on the posterior surface of the cervix, posterior vaginal wall (for an anteverted uterus) or uterine fundus (for a retroverted uterus); the test was considered negative (positive sliding sign) when those structures were completely free of one another. The reference standard was obliteration of the pouch of Douglas and/or bowel involvement demonstrated on laparoscopy, either by visual inspection or histological confirmation. The primary outcome was pooled sensitivity, specificity, positive likelihood ratio (LR+), negative likelihood ratio (LR–) and diagnostic odds ratio (OR) of the sliding sign. The numbers of true‐positive, true‐negative, false‐positive and false‐negative cases were obtained from each included study. Post‐test probabilities were calculated and plotted on Fagan nomograms, using the mean prevalence of pouch of Douglas obliteration and bowel involvement as the pretest probability. The presence of heterogeneity for sensitivity and specificity was assessed graphically by drawing forest plots of sensitivity and specificity, and then using Cochran's Q and the I 2 statistic. A test for heterogeneity examines the null hypothesis that all studies are evaluating the same effect; P  < 0.1 was considered to indicate heterogeneity. According to Higgins et al . 16 , I 2 values of 25%, 50% and 75% are considered to indicate low, moderate and high heterogeneity, respectively. In cases of moderate or high heterogeneity, meta‐regression was used. The covariates analyzed in meta‐regression were year of publication, sample size and prevalence of pouch of Douglas obliteration or bowel involvement. Summary receiver‐operating‐characteristics (sROC) curves for each condition were plotted to illustrate the relationship between sensitivity and specificity, and the area under the curve was calculated. Analyses were performed using Meta‐analytical Integration of Diagnostic Accuracy Studies (MIDAS) and METANDI commands in STATA version 12 for Windows (Stata Corp., College Station, TX, USA); P  < 0.05 was considered to indicate statistical significance. The Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology was used to assess the quality of the retrieved evidence 17 . The online GRADE tool was adopted ( http://GRADEPro.org , accessed in December 2021). The assessment was performed by three authors (J.L.A., S.G., M.A.P.) by consensus.

The electronic search provided a total of 334 citations. After removal of 162 duplicate records, 172 citations remained. Of these 172 citations, 123 were excluded after screening by title and abstract, including reviews ( n  = 6), case reports ( n  = 3), letters to the editor ( n  = 1), opinions ( n  = 1) and studies not related to the addressed topic ( n  = 112). We reviewed the full text of the remaining 49 articles. Forty‐three studies were excluded for the following reasons: study not relevant to the topic addressed ( n  = 8), data for 2 × 2 table not available ( n  = 6), retrospective study design ( n  = 2), surgery not used as the reference standard ( n  = 4), overlapping series ( n  = 6), no specific evaluation of the sliding sign ( n  = 5), review/opinion paper ( n  = 7) and reproducibility/learning curve rather than diagnostic type of study ( n  = 5) (Appendix  S2 ). Two additional relevant studies were found in the reference lists of the studies included in the review (snowball technique). A flowchart summarizing study identification and selection is presented in Figure  1 . There was no need to contact the authors, as all relevant data needed to perform the meta‐analysis were available. Flowchart summarizing inclusion in systematic review and meta‐analysis of studies evaluating the diagnostic accuracy of the sliding sign on transvaginal ultrasound for pouch of Douglas obliteration and/or bowel involvement in women with suspected endometriosis. Eight studies published between October 2013 and January 2020 were included in the final analysis 8 , 18 , 19 , 20 , 21 , 22 , 23 , 24 . Five studies analyzed the accuracy of the preoperative sliding sign for the prediction of both pouch of Douglas obliteration and bowel involvement in women with suspected deep infiltrating endometriosis 18 , 19 , 20 , 21 , 24 . One study 23 analyzed the accuracy of the preoperative sliding sign for the prediction of pouch of Douglas obliteration only, and two studies 8 , 22 analyzed the accuracy of the preoperative sliding sign for detecting bowel involvement only. Two studies from the same research group were included 21 , 22 . The study by Menakaya et al . 21 provided data on diagnostic accuracy of the sliding sign to detect pouch of Douglas obliteration and bowel involvement, whereas the study by Reid et al . 22 reported data on bowel involvement only. It was concluded that patients with bowel involvement in the study by Menakaya et al . 21 were also included in the study by Reid et al . 22 . Therefore, data on bowel involvement from Menakaya et al . were excluded from the analysis. Thus, we analyzed data from six studies to assess the diagnostic performance of the sliding sign for detecting pouch of Douglas obliteration 18 , 19 , 20 , 21 , 23 , 24 and from six studies to assess the diagnostic performance of the sliding sign for detecting bowel involvement 8 , 18 , 19 , 20 , 22 , 24 . Nine hundred and thirty‐eight women were assessed for detecting obliteration of the pouch of Douglas. Of these 938 patients, 293 had pouch of Douglas obliteration on laparoscopy. The mean prevalence of pouch of Douglas obliteration was 37%, ranging from 15% to 65%. Nine hundred and sixty‐three women were assessed for detecting bowel involvement. Of these 963 patients, 200 had bowel involvement on laparoscopy. The mean prevalence of bowel involvement was 23%, ranging from 5% to 32%. The mean age of patients was reported in five of the eight included studies 8 , 18 , 19 , 21 , 23 . All studies were observational prospective studies. Three of them were multicenter studies 18 , 21 , 22 . Only three studies specified that patient recruitment was consecutive 18 , 21 , 22 ; in one study, recruitment was non‐consecutive 19 . Four studies did not specify the type of recruitment 8 , 20 , 23 , 24 . In all studies, TVS was performed by an expert or trained examiner. Most studies did not report whether the sonographer was blinded to the patient's medical history. In all studies, surgery was performed by an expert surgeon. One study reported that the surgeon was blinded to TVS findings 18 , and one reported that the surgeon was not blinded to TVS findings 22 . In the remaining studies, this information was not provided. The interval between TVS and surgery was not specified in four studies 19 , 20 , 23 , 24 . In two studies 21 , 22 , surgery was performed within 6 months after TVS, and in two other studies 8 , 18 , the interval elapsed between TVS and surgery was less than 3 months. Table  1 shows PICOS characteristics of the included studies. Evaluation of the risk of bias and concerns regarding applicability of the selected studies is shown in Table  2 . Quality assessment of studies included in systematic review and meta‐analysis, according to Quality Assessment of Diagnostic Accuracy Studies‐2 (QUADAS‐2) tool Only first author of each study is given. For the patient‐selection domain, all studies included patients with clinical suspicion of endometriosis. Most studies were considered to be at low risk of bias for patient selection, as there was a clear explanation of inclusion and exclusion criteria. One study was considered to be high risk because it used non‐consecutive recruitment and included patients with previous pelvic surgery 19 . For the index‐test domain, all studies were considered to be low risk because they provided an adequate description of the method for the sliding‐sign assessment on TVS, as well as how it was interpreted. For the reference‐standard domain, all but one studies were likely to classify correctly the target condition using the reference standard, and one study did not describe in detail the surgical procedure performed 24 . For the flow‐and‐timing domain, the time elapsed between the index test and reference standard indicated a low risk of bias in two studies 8 , 18 and a high risk in two studies 21 , 22 . The risk for this domain was unclear in four studies 19 , 20 , 23 , 24 . In terms of applicability, all studies were deemed to include patients who were relevant to the review question. For the index‐test and reference‐standard domains, all studies presented low concerns regarding applicability. Pooled sensitivity, specificity, LR+, LR– and OR of the sliding sign on TVS for detecting pouch of Douglas obliteration were 88% (95% CI, 81–93%), 94% (95% CI, 91–96%), 15.3 (95% CI, 10.2–22.9), 0.12 (95% CI, 0.07–0.21) and 123 (95% CI, 62–244), respectively. Heterogeneity was moderate for sensitivity (Cochran's Q, 16.20; P  = 0.01, I 2  = 69.1%) and low for specificity (Cochran's Q, 8.96; P  = 0.11, I 2  = 44.2%) (Figure  2a ). As heterogeneity was moderate, metaregression was performed. We observed that the differences in the prevalence of pouch of Douglas obliteration across studies could explain this heterogeneity ( P  < 0.01). Forest plots showing pooled sensitivity and specificity of the sliding sign on transvaginal ultrasound in the detection of pouch of Douglas obliteration (a) and bowel involvement (b) in women with suspected endometriosis. Only first author of each study is given. The area under the sROC curve was 0.97 (95% CI, 0.95–0.98) (Figure  3a ). As shown in the Fagan nomogram (Figure  4a ), a positive test on TVS (negative sliding sign) in women with suspected deep endometriosis significantly increased the pretest probability of pouch of Douglas obliteration on laparoscopy, from 37% to 90%, while a negative test (positive sliding sign) significantly decreased the pretest probability, from 37% to 7%. No publication bias was observed ( P  = 0.64). Summary receiver‐operating‐characteristics curves ( ) showing performance of the sliding sign on transvaginal ultrasound in the detection of pouch of Douglas obliteration (a) and bowel involvement (b) in women with suspected endometriosis. , Study estimate; , summary point; , 95% confidence region; , 95% prediction region. Fagan nomograms for detecting pouch of Douglas obliteration (a) and bowel involvement (b) based on negative ( ) and positive ( ) sliding sign on transvaginal ultrasound in women with suspected endometriosis. LR−, negative likelihood ratio; LR+, positive likelihood ratio; prob, probability. Pooled sensitivity, specificity, LR+, LR– and OR of the sliding sign on TVS for detecting bowel involvement were 81% (95% CI, 64–91%), 95% (95% CI, 91–97%), 16.0 (95% CI, 9.0–28.6), 0.20 (95% CI, 0.10–0.40) and 81 (95% CI, 34–191), respectively. Heterogeneity was high for both sensitivity (Cochran's Q, 23.40; P  < 0.01, I 2  = 78.6%) and specificity (Cochran's Q, 20.96; P  < 0.01, I 2  = 76.1%) (Figure  2b ). As heterogeneity was high, meta‐regression was performed. We observed that the differences in the prevalence of bowel involvement across studies could explain this heterogeneity ( P  < 0.01). The area under the sROC curve was 0.96 (95% CI, 0.94–0.98) (Figure  3b ). As shown in the Fagan nomogram (Figure  4b ), a positive test on TVS (negative sliding sign) in women with suspected deep endometriosis significantly increased the pretest probability of bowel involvement on laparoscopy, from 23% to 83%, while a negative test (positive sliding sign) significantly decreased the pretest probability, from 23% to 6%. No publication bias was observed ( P  = 0.14). Regarding GRADE assessment, evidence of high quality showed that the sliding sign as assessed by TVS has a high accuracy for detecting pouch of Douglas obliteration and bowel involvement in women with endometriosis (Tables  3 and 4 ). This assessment should be recommended for all women evaluated for this clinical entity. Grading of Recommendations Assessment, Development and Evaluation (GRADE) assessment of the quality of evidence regarding the diagnostic accuracy of the sliding sign on transvaginal ultrasound for pouch of Douglas obliteration in women with suspected endometriosis Pretest probability of 37%. CoE, class of evidence. Grading of Recommendations Assessment, Development and Evaluation (GRADE) assessment of the quality of evidence regarding the diagnostic accuracy of the sliding sign on transvaginal ultrasound for bowel involvement in women with suspected endometriosis Pretest probability of 23%. CoE, class of evidence.

In this meta‐analysis, we observed that the diagnostic performance of the TVS sliding sign for detecting pouch of Douglas obliteration and bowel involvement in women with suspected endometriosis was high, with pooled sensitivity of 88% and 81% and pooled specificity of 94% and 95%, respectively. The studies included had mostly low risk of bias and low concerns regarding applicability. However, it should be borne in mind for the flow‐and‐timing domain that the time elapsed from TVS to laparoscopy was not reported in four studies. We cannot assume that the time elapsed between the two procedures was long; however, we believe that, if the time elapsed was long, it could be a potential confounding factor because the condition of the pelvis could become worse. Our findings demonstrate that the sliding sign is an excellent ultrasound sign for detecting pouch of Douglas obliteration and bowel involvement in women undergoing surgery for suspected endometriosis. The heterogeneity observed across the studies analyzing pouch of Douglas obliteration was low for specificity and moderate for sensitivity, demonstrating comparability of the studies. However, the heterogeneity of the studies assessing bowel involvement was high. Therefore, in the latter case, our findings should be interpreted with caution. These findings might be clinically relevant, as pouch of Douglas obliteration may increase the duration and complexity of surgery 25 . Having this information prior to surgery may be helpful to surgeons, as it may influence the choice of surgical technique, lead to involvement of a multidisciplinary surgical team and allow referral to the most appropriate practice 26 . Additionally, a negative sliding sign alone may be useful for identifying women with clinical suspicion of deep endometriosis who require further evaluation, for example, an examination by an expert sonologist to assess for the presence of classic signs of rectal infiltration. Furthermore, a negative sliding sign may be associated with sigmoid involvement. This may also constitute a reason for referring the patient for expert examination. However, five of the six studies on bowel involvement included in this meta‐analysis did not provide separate information on rectal and sigmoid involvement; therefore, a subgroup analysis could not be performed. Some studies have shown that assessment of the sliding sign may have a short learning curve and be reproducible among expert examiners 4 , 27 , 28 , 29 . However, it is important to bear in mind that diagnostic performance depends on expertise and that not all trainees may reach competence 28 , 29 , 30 . Reproducibility should be tested in larger prospective studies. We should also consider the fact that a negative sliding sign may be produced by inflammatory changes, for example due to pelvic surgery or pelvic inflammatory disease. This might be a confounding factor. In patients with such a medical history, it would be difficult to ascertain whether a negative sliding sign is related to endometriosis or to postsurgical/disease‐related inflammatory processes. Some limitations of this meta‐analysis should be considered. We believe that the main limitation is the small number of studies and patients included. Additionally, the reported prevalence of pouch of Douglas obliteration on laparoscopy may vary depending on the surgeon's skills. Most of the included studies did not provide details regarding this aspect, and we should not assume that inspection of the abdominal cavity was made by properly trained experienced surgeons. In fact, one study did not describe at all the surgical procedures performed 24 . However, we do not consider that this had a significant effect on the results of the quantitative synthesis. The strengths of our study are that, to the best of our knowledge, it is the first meta‐analysis to analyze specifically the performance of the TVS sliding sign in detecting bowel involvement and that it provides up‐to‐date evidence on the diagnostic performance of this sign for obliteration of the pouch of Douglas in women with clinical suspicion of endometriosis. Two previous meta‐analyses have assessed the diagnostic performance of TVS for detecting pouch of Douglas obliteration 31 , 32 . However, neither of them assessed specifically the sliding sign for detecting bowel involvement. Nisenblat et al . 31 assessed six studies that evaluated the diagnostic performance of TVS for detecting pouch of Douglas obliteration. Two of those studies have been included in our meta‐analysis 19 , 20 , but four of them were not 9,33–35 . We did not include these four studies because they reported data from series that overlapped with more recent studies from the same group 9 , 35 or did not report specifically on the sliding sign as a marker for diagnosing pouch of Douglas obliteration 33 , 34 . The pooled sensitivity and specificity reported by Nisenblat et al . (83% and 97%, respectively) were similar to those in our study. Noventa et al . 32 reported data from eight studies that assessed the role of TVS in detecting pouch of Douglas obliteration. Two of these studies have been included in our meta‐analysis 18 , 19 . Six studies were excluded for one of the following reasons: overlapping data with more recent studies from the same group 9 , 34 , not describing the sliding sign 36 , 37 , data for constructing 2 × 2 table could not be extracted 38 or the study focused on transrectal ultrasound 39 . The pooled sensitivity and specificity reported by Noventa et al . were 80% and 95%, respectively. Our meta‐analysis reports data from more recent studies and represents a larger series. In conclusion, the TVS sliding sign seems to be an accurate method for the diagnosis of pouch of Douglas obliteration and bowel involvement in women with a clinical suspicion of pelvic endometriosis who undergo surgery when expert examiners perform the ultrasound examination. It remains to be seen whether the TVS sliding sign test performs equally well and is reproducible in the hands of less experienced examiners. The findings of this meta‐analysis confirm the fundamental role of TVS as a diagnostic tool in women with suspected endometriosis, suggested by previous studies 40 , 41 , 42 .

Endometriosis is a gynecological disease, defined as the presence of endometrial‐like tissue outside the uterus, that affects up to 5–10% of premenopausal women, being more frequent in women with symptoms such as dysmenorrhea, chronic pelvic pain, dyspareunia, dyschezia and infertility 1 . The diagnosis of endometriosis can be difficult and is often delayed 2 . Transvaginal ultrasound (TVS) has been shown to be a highly accurate and reproducible tool for detecting endometriosis. It has been proposed as the primary imaging modality in patients with pelvic pain 3 , 4 , 5 and has shown a high correlation with laparoscopy 6 . The International Deep Endometriosis Analysis (IDEA) group proposed a systematic scanning technique for sonographic evaluation of the pelvis when a patient is suspected to have endometriosis 7 . This technique is based on four steps: evaluation of the uterus and the adnexa to identify and describe signs of adenomyosis and examine for the presence of endometrioma; assessment of ‘soft markers’, such as ‘kissing’ ovaries; assessment of the ‘sliding sign’; and identification of deep endometriotic nodules. The sliding‐sign diagnostic test, which involves determining whether the anterior rectum glides freely across the posterior aspect of the cervix, posterior vaginal wall (for an anteverted uterus) or uterine fundus (for a retroverted uterus) 7 , has been associated with bowel involvement, namely rectal or sigmoid anterior wall infiltration by endometriotic nodules 8 , and pouch of Douglas obliteration 9 . Pouch of Douglas obliteration is considered to be a sign of severe endometriosis and could result in marked anatomical distortion of the pelvis. Women with pouch of Douglas obliteration are three times more likely to have bowel endometriosis and bowel surgery than are patients with a non‐obliterated pouch of Douglas 10 . A negative sliding sign is considered a ‘hard marker’ for rectal/sigmoid infiltration by deep endometriosis, which may make surgery more complex 11 . Pouch of Douglas obliteration or bowel involvement during surgery may increase the duration of the procedure and necessitate advanced surgical skills. Consequently, in addition to improving our understanding of pelvic pain symptoms, the ability to evaluate the sliding sign preoperatively may help surgery planning, prompt colorectal surgeon support and allow proper informed consent to be obtained 12 , 13 . The aim of this systematic review and meta‐analysis was to evaluate the diagnostic performance of the sliding sign assessed by TVS for detecting pouch of Douglas obliteration and bowel involvement in patients with suspected endometriosis, using laparoscopy as the reference standard.

Appendix S1 Protocol for the systematic review and meta‐analysis Click here for additional data file. Appendix S2 Studies excluded after full‐text review Click here for additional data file.