Learning curve for ultrasonographic diagnosis of deep infiltrating endometriosis using structured offline training program

Objective: To assess the learning curves of trainees during a structured offline/hands-on training program for the ultrasonographic diagnosis of deep infiltrating endometriosis (DIE). Methods: Four trainees (all Ob/Gyn postgraduates with at least 5 years' experience in ultrasonography in obstetrics and gynecology, but with no experience of sonographic examination of DIE) participated in the study. They underwent a 2-week training program with a single trainer. Day 1 was devoted to theoretical issues and guided offline analysis of 10 three-dimensional ultrasound volumes. During the following days, four sessions of real-time sonographic examinations were performed in a DIE referral center ultrasound unit. In between these sessions, the trainees analyzed four datasets offline, each containing 25 volumes. At the end of each set, misinterpreted volumes were reassessed with the trainer. Presence or absence of DIE at surgery was considered the gold standard. The trainees' learning process was evaluated by learning-curve cumulative summation (LC-CUSUM) and the deviations of the trainees' level of performance at the control stage was assessed by CUSUM (standard CUSUM), for different locations of DIE. Results: The trainees reached competence after an average of 17 (range, 14–21) evaluations for bladder, 40 (range, 30–60) for rectosigmoid, 25 (range, 14–34) for forniceal, 44 (range, 25–66) for uterosacral ligament (USL) and 21 (range, 14–43) for rectovaginal septum (RVS) locations of DIE, and then kept the process under control, with error levels of less than 4.5% until the end of the test. The overall accuracy for each trainee in diagnosis of DIE at the different locations ranged from 0.91 to 0.98 for bladder DIE, from 0.80 to 0.94 for rectosigmoid DIE, from 0.90 to 0.94 for forniceal DIE, from 0.79 to 0.82 for USL DIE and from 0.89 to 0.98 for RVS DIE. Conclusions: The suggested 2-week training program, based on a mixture of offline and live scanning sessions, is feasible and apparently provides effective training for the ultrasonographic diagnosis of DIE.