Residents were randomized to control (no training) and intervention (model training) groups. They were filmed while performing a series of tasks. Experts were also filmed. Each video was scored by 2 blinded raters.
Multicenter collaboration within the Western Society of Pelvic Medicine (Vancouver, Calgary, and Edmonton). Face and content validity were evaluated. A standard scoring tool was developed for performance evaluation. Interrater reliability was assessed using intraclass correlation coefficient. Cronbach α was calculated for internal consistency. Jonckheere-Terpstra test verified whether the scores increased with operator skill level.
A total of 14 junior gynecology residents, 2 urogynecology fellows, and 3 staff urogynecologists were rated by a total of 6 gynecologic surgeons who scored 42 videos each.
Experienced pelvic surgeons from 3 participating sites agreed the models captured essential elements of real surgical skills (face validity) and of the true procedures (content validity). Intraclass correlation coefficient was adequate (AR = 0.86, PR = 0.90, and VH = 0.87). Cronbach α for the total scores was adequate (AR = 0.85, PR = 0.8, and VH = 0.71). Performance score increased with operator skill level for all 3 procedures (AR, p = <0.001; PR, p = 0.008; and VH, p = 0.007).
Our low fidelity procedure-specific vaginal surgery models had adequate initial validity. Future research will investigate transferability of acquired skills to the operating room.