Methods
This was a prospective observational study of women attending the Early Pregnancy Assessment Unit at University College London Hospital between April 2017 and March 2019. Only clinically stable women who presented with mild to moderate vaginal bleeding and/or pelvic pain were included into the study. They were all seen by a single operator (E.B) who had extensive experience in scanning of women with endometriosis in a tertiary Endometriosis Centre. A single operator was used in order to reduce the inter-observer variability associated with ultrasound diagnosis of endometriosis. We recorded women’s demographic data, indications for attendance and a detailed medical history. This included their age, ethnicity, body mass index (kg/m2), smoking status, gravidity and parity. All pregnant women that had a transvaginal ultrasound scan were included in the study. All scans were performed in a standard fashion using a 7.5-Mhz probe (Voluson E8, GE Medical Systems, Milwaukee, WI, USA). The uterus was examined for the presence of congenital uterine anomalies, adenomyosis and fibroids. Congenital uterine anomalies were described using the revised ASRM classification.17 The diagnosis of adenomyosis and fibroids was made in accordance with previously described diagnostic criteria.13,18 The ovaries and adnexae were then examined. Women were considered to have a diagnosis of endometriosis if there was evidence of endometriotic ovarian cysts and/or deep endometriotic nodules. Ovarian endometriomas were defined as thick-walled and well circumscribed cysts with low-level internal echoes (‘ground glass’).19 Endometriotic nodules present as hypoechoic or isoechogenic solid lesions with irregular borders that were fixed to surrounding pelvic structures.10,20Statistical analysis was performed to establish the prevalence of endometriosis in addition to possible associations between demographic and clinical variables using logistic regression and multivariate analysis. Analysis was performed using SPSS software (SPSS Inc., Chicago, IL, USA). All clinical findings were stored in a clinical database that facilitated data entry and retrieval (PIA‐Fetal Database, Viewpoint Bildverabeitung GmbH, Wessling, Germany). The study was registered on Research Registry (Unique identifying number: researchregistry4569). A retrospective audit of records of 750 women who attended our unit during the first trimester of pregnancy (4-14 weeks’ gestation), showed a 3.2% prevalence of ovarian and deep endometriosis. Based on this information we calculated that the required sample size for our study was 1176 to achieve a precision of 1% with 95% confidence interval.