Learning curve for the detection of pouch of Douglas obliteration and deep endometriosis of the rectum in gynaecological sonology trainees

Objectives Transvaginal ultrasound (TVS) for deep endometriosis (DE) has a high diagnostic accuracy, allowing for preoperative disease staging. We aimed to evaluate the learning curve of gynaecological sonology trainees (observers) to correctly classify pouch of Douglas (POD) obliteration state and identify rectal DE. Methods This was a prospective study from December 2017-December 2018. Three observers were recruited to participate. Each observer read the International Deep Endometriosis Analysis (IDEA) group consensus statement and primary publication on the real-time dynamic sliding sign test. All observers observed the unit's supervisor, who was defined as the study reference standard, for five DE TVS examinations. A pre-determined 150 examinations were planned. A DE TVS was performed by the study reference standard as per the IDEA group consensus statement. Subsequently, the observer performed a focused component to assess the rectum and POD state, having been blinded to the patient's clinical history and reference standard findings. Immediate feedback and hands-on teaching were provided after the observer's official classification. In order to evaluate the number of scans needed to gain proficiency, the learning curve using the cumulative sum (LC-CUSUM) was utilized. The acceptable performance rate (proficiency) was set at 15% failure, the unacceptable performance rate was set at 30%, and the equivalence zone was set at 5%. Results One hundred and fifty learning curve TVS examinations were performed. Twenty-six (17.3%) patients had rectal DE and 34 (22.7%) had a negative sliding sign by the reference standard. Each observer performed 50 scans supervised by the reference standard. The overall accuracy of presence or absence of rectal DE was 90%, ranging from 82–94% amongst the three observers. The accuracy of POD state classification was 92.7%, ranging from 90–96% amongst the three observers. LC-CUSUM for rectal DE demonstrated observer 1 did not reach proficiency, whereas observers 2 and 3 required 32 and 20 TVS examinations, respectively. For POD obliteration state classification, observer 1 did not reach proficiency, whereas observers 2 and 3 required 46 and 30 TVS examinations, respectively. Conclusion For two of the three gynaecological sonology trainees, proficiency in diagnosing rectal DE and POD obliteration was attainable in less than the 50 planned supervised ultrasounds. Contrary to other studies on the topic, this study suggests that not all trainees can reach proficiency during a program based upon a pre-defined number of scans. Overall, we demonstrate that a one-size-fits-all approach may not be appropriate for teaching trainees these advanced ultrasound skills.