The Poliquin Resident Research Competition will take place on Monday June 18, 2018.
“Learning Curve and Predictive Validity of a Virtual Reality Sinus Surgery Simulator”
Mirko Manojlovic Kolarski, Christopher M.K.L. Yao, Stephen H.T. Chen, John M. Lee, Eric Monteiro, Allan D. Vescan
Objectives: To determine the learning curve (LC) and predictive validity of a virtual reality (VR) simulator in endoscopic sinus surgery (ESS).
Methods: Otolaryngology residents were allocated to VR or control arms. The VR arm completed 7-8 simulation sessions consisting of 3 tasks. Tasks were scored using proprietary performance metrics. Linear mixed effects models were used to model the LCs for time to completion and score. Following washout, both arms were recorded during cadaveric ESS. Videos were scored by 3 blinded staff rhinologists using task specific (TS) and global rating (GR) checklists. Intraclass coefficient between raters was calculated to assess inter-rater reliability.
Results: 18 residents (7 VR, 11 control) participated. Endoscopy task completion time improved by 84.9+/-40.8 (p=0.001) and 74.3+/-35.8 (p=0.001) seconds by the seventh and eighth attempts, respectively, without change in safety metrics. Sphenoethmoidectomy task score improved by 40.2+/-20.4 (p=0.09) and 45.3+/-13.9 (p=0.09) points by the seventh and eighth attempt. LC modeling showed that most residents plateaued after 5-6 attempts. During cadaveric surgery, average score on TS was 46.9+/-6.5 and 52.2+/-7.2 (p=0.17) and GR was 19.0+/-3.3 and 22.4+/-4.3 (p=0.12) in the control and VR arms, respectively. Intraclass coefficients were 0.199 (-0.07,0.524) and 0.155 (-0.105, 0.485) for TS and GR respectively.
Conclusion: Most residents plateaued after 5-6 attempts, but LCs vary by task and resident and may require more sessions. There was no significant difference between arms during cadaveric surgery. To evaluate the efficacy of simulation, it will be important to conduct further studies into standardized metrics of evaluation in ESS.
- By the end of this session the audience will be able to interpret the learning curves for simulation tasks when presented with regression models.
- By the end of the session residents and staff will be able to describe the challenges with sinus surgery evaluation metrics when presented with an predictive validity study.