Virtual Reality Simulation Has Weak Correlation with Overall Trainee Robot-Assisted Laparoscopic Hysterectomy Performance

Abstract

Both simulator practice and intraoperative performance serve to inform surgical trainee training, but the skill transfer from simulation to the intraoperative setting remains unclear. This study evaluates the correlation between trainee performance on virtual reality simulation and (1) overall intraoperative performance during robotic-assisted laparoscopic hysterectomy (RALH) procedures and (2) suturing performance during vaginal cuff closure portion of the case. Retrospective subgroup analysis of randomized controlled trial. Academic hospital. Patients with RALH (N=29). Gynecological trainees (N=21) performed simulation tasks using the da Vinci skills simulator on the day of surgery before performing RALH. Attending surgeons assessed participants' intraoperative performance using Global Evaluative Assessment of Robotic Skills (GEARS). Performance of the vaginal cuff closure step was subsequently assessed using GEARS scoring of anonymized videos. Spearman's correlation was used to quantify the relationship between simulation and intraoperative performances. Trainees achieved a median intraoperative GEARS score of 18.5/30 (interquartile range: 17-22) and a median total simulator score of 84.4/100 (interquartile range: 78.1-87.5). More advanced residents exhibited worse overall simulator performance (median score 86.6/100 compared with 78.8/100, p=.03) and similar intraoperative GEARS scores during overall RALH and vaginal cuff closure compared with less experienced trainees. Total simulation performance score was negatively correlated with GEARS Bimanual Dexterity (ρ=-0.46, p=.02) and Force Sensitivity subscores (ρ=-0.39, p=.05). There was no correlation between total GEARS intraoperative vaginal cuff closure scores and overall simulation performances; however, total Tubes simulation score was correlated with higher GEARS Force Sensitivity subscore (ρ=0.73, p=.05). In this study, there was limited correlation between simulation score metrics and trainees' overall intraoperative performance. Furthermore, we identified that GEARS scores could not distinguish between similar trainee skill levels. These findings underscore the need to develop intraoperative assessment tools that can better discriminate different but similar skill levels.