Lotte L Nieuwenhuis, Frederik Jr Hermans, A J Marjolein Bij de Vaate, Mariska Mg Leeflang, Hans Am Brölmann, Wouter Jk Hehenkamp, Ben Willem J Mol, T Justin Clark, Judith Af Huirne

Focal abnormalities most commonly acquired within the uterine cavity include endometrial polyps (arising from the endometrium) and submucous fibroids (arising from the myometrium). These benign abnormalities can cause several problems, including abnormal uterine bleeding (AUB) and subfertility. Two-dimensional saline infusion sonography (2D SIS) is a minimally invasive test that can be used to diagnose these pathologies, but it is less accurate than hysteroscopy, which is a more invasive procedure by which an endoscope allows direct visualisation of the uterine cavity. Three-dimensional (3D) SIS appears to enhance sonographic visualisation within the uterine cavity, thereby offering a potentially more accurate minimally invasive diagnostic test. Primary objectives • To evaluate the diagnostic accuracy of 3D SIS (index test 1) compared with 2D SIS for the diagnosis of focally growing lesions (presence or not) in women with AUB or subfertility, with hysteroscopy performed as the reference test. • To evaluate the diagnostic accuracy of 2D+3D SIS (index test 2) compared with 2D SIS for the diagnosis of focally growing lesions (presence or not) in women with AUB or subfertility, with hysteroscopy performed as the reference test. In this case, any abnormality on either modality was regarded as a positive result ('OR' approach). Secondary objectives • To evaluate the diagnostic accuracy of 3D SIS (index test 1) compared with 2D SIS according to type of abnormality and discrimination between uterine polyps and submucous fibroids in women with AUB or subfertility, with hysteroscopy and histology used as the reference.• To evaluate the diagnostic accuracy of 2D+3D SIS (index test 2) compared with 2D SIS according to type of abnormality and discrimination between uterine polyps and submucous fibroids in women with AUB or subfertility, with hysteroscopy and histology used as the reference. We searched the following databases: Cochrane Central Register of Studies Online (CENTRAL CRSO), MEDLINE, Embase, PubMed, Cochrane Gynaecology and Fertility Group (CGF) Specialised Register and CGFG Diagnostic Test Accuracy (DTA) Specialised Register, clinicaltrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP). Screening reference lists of appropriate studies was also performed. We screened for eligibility all studies identified from inception until March 2016. We performed searches with no date or language restrictions. The population of interest consisted of premenopausal women with AUB or subfertility and postmenopausal women with AUB. Diagnostic test accuracy studies, randomised controlled trials (RCTs) and prospective cohort studies were eligible for inclusion if they evaluated the accuracy of both 2D SIS and 3D SIS for the diagnosis of acquired intracavitary abnormalities with hysteroscopy used as the reference standard. In light of the lack of data for 3D SIS, we also included studies that evaluated the accuracy of 3D SIS alone. Two review authors read all potentially eligible references after performing a first screening by title and abstract (LLN and FJRH). They independently extracted data to construct 2×2 tables from eligible studies and assessed studies for methodological quality using the QUADAS-2 tool (revised tool for quality assessment of diagnostic accuracy studies). To describe and visually present results, we produced in RevMan forest plots showing pairs of sensitivity and specificity together with 95% confidence intervals from each study, as well as raw receiver operating characteristic (ROC) plots. We displayed paired analyses in an ROC plot by linking sensitivity-specificity pairs from each study by using a dashed line. To compare 3D SIS versus 2D SIS, we restricted analyses to studies that provided 2×2 tables for both tests and used the bivariate meta-analysis of sensitivity and specificity. Thirteen studies (1053 women) reported the accuracy of 3D SIS for focal uterine abnormalities; 11 of these (846 women) were suitable for meta-analysis, and eight reported accuracy according to the type of focal abnormality. The design of the included studies seems applicable. The main problem involving the quality of included studies is insufficient reporting of study methods, resulting in unclear risk of bias for several of the quality domains assessed. Therefore, we considered the overall quality of the evidence as low. The summary estimate (11 studies reporting absence or presence of abnormality at 3D SIS) for sensitivity was 94.5% (95% confidence interval (CI) 90.6% to 96.9%) and for specificity 99.4% (95% CI 96.2% to 99.9%). Meta-analysis of the eight studies (N = 716) directly comparing 2D SIS versus 3D SIS showed summary sensitivity of 96.9% (95% CI 91.9% to 98.8%) and summary specificity of 99.5% (95% CI 96.1% to 100%) for 3D SIS. For 2D SIS, summary sensitivity was 90.9% (95% CI 81.2% to 95.8%) and summary specificity was 96.3% (95% CI 86.1% to 99.1%). The difference in accuracy between 2D SIS and 3D SIS was non-significant (P values of 0.07 for sensitivity and 0.10 for specificity). Low-quality evidence suggests that 3D SIS may be very accurate in detecting intracavitary abnormalities. Meta-analysis revealed no statistically significant differences between 2D SIS and 3D SIS. Summary sensitivity and summary specificity are higher for 3D SIS, but margins of improvement are limited because 2D SIS is already very accurate. When the technology and appropriate expertise are available, 3D SIS offers an alternative to 2D SIS. Both 2D SIS and 3D SIS should be considered alternatives to diagnostic hysteroscopy when intracavitary pathology is suspected in subfertile women and in those with abnormal uterine bleeding.