Background Endoscopic sinus surgery is nuanced and technically challenging for novice trainees.
With modern resident duty limitations and an increased focus on quality metrics in
education and beyond, it is important to develop alternative tools for teaching this
skill. Neurotouch is a validated high-fidelity virtual-reality simulator that provides
haptic feedback and visual cues to simulate sinus surgery. Implementing the simulator
in the era of competency-based curriculum has not been thoroughly investigated. In
this study, we determined the learning curve for three tasks on the Neurotouch and
assessed the Neurotouch as a learning tool compared with standard practice during
cadaveric endoscopic sinus surgery.
Methods Residents were randomized to virtual reality (VR) or control arms. Residents (PGY
1−4) in the VR arm completed seven to eight sessions on the Neurotouch. Each session
consisted of two practice tasks (sphenoid endoscopy and polypectomy) and an evaluation
task (endoscopic sinus surgery). Residents in the control arm did not have access
to adjunctive tools. Participants were evaluated on performance metrics on quality,
efficiency, and safety. They received immediate feedback following the simulation,
displayed as a score out of 100 with points gained for successfully performing the
task and points lost for errors. These scores were aggregated to calculate the learning
curve for each of the tasks. After a washout period, residents in VR and control arms
were evaluated during a cadaveric endoscopic sinus course.
Results In the first task, the average time to completion of endoscopy for the first, third,
and eighth attempts were 123.2 ± 41.7, 67.0 ± 49.2, and 36.8 ± 13.8 seconds respectively,
with no significant change in overall score. There was significant improvement between
the first and third (p = 0.05) attempts, which was sustained during the eighth (p = 0.001) attempt. The variance between trainees also narrowed with successive practice
attempts. In the polypectomy task, there was also no significant difference between
the average scores for the first, seventh, and eighth attempts. Evaluation task scores
on first attempt, seventh, and eighth attempts were 28.6 ± 19.5, 68.8 ± 8.4, and 72.3 ± 8.9,
respectively. The change from first to seventh and first to eighth attempt was 40 ± 20.4
(p = 0.09) and 45 ± 12.9 (p = 0.09).
Conclusion In its current virtual reality iteration, there was a significant improvement in
time to completion after three sessions, which was maintained through further attempts.
The polypectomy task did not show a significant change in overall scoring. This may
be due to the simplicity of the task and high average scores at first attempt. For
the evaluation task, there was an increase in average score from first to last attempt,
which approached but did not reach significance. This learning curve data will assist
with implementing the Neurotouch as part of a simulation curriculum for novice trainees
prior to spending time in the operating room. Further evaluation of the efficacy of
the simulator in improving surgical skill and qualitative measures is pending.
