Abstract
Purpose :
We performed a prospective study to facilitate the training of future vitreoretinal (VR) surgeons by testing the construction validity of a virtual reality VR-training program.
Methods :
A virtual VR-training program was composed on the Eyesi surgical simulator (VRmagic, Manheim, Germany) and tested on the following three groups:
Group 1 (n=3) included trained VR-surgeons, Group 2 (n=8) consisted of ophthalmologist residents, and Group 3 (n=20) included medical students without clinical ophthalmology training. Each group completed the program twice up to two weeks apart. The results from the second training session were recorded.
The program included relevant test modules with different levels of complexity and included: Navigation level 2 (Nav2), Forceps Training level 5 (ForT5), Bimanual Training level 3 (BimT3), Laser Coagulation level 3 (LasC3), Posterior Hyaloid level 3 (PostH3) and Internal Limiting Membrane Peeling level 3 (ILMP3). The level for each module was selected to ensure that the tasks would be difficult enough to enable discrimination based on surgical skills.
Results from all the modules were summarized. The scores in each module were assessed from 2-5 different factors (i.e. Target Achievement and Tissue Treatment). A total maximum of 600 points could be achieved (100 points in each level). Construction validity for a module was shown if the median score for Group 1 was higher than for Group 2, which is in turn was higher than for Group 3.
Kruskal Wallis test by ranks was used for statistical analyses.
Results :
For the following modules, statistical significant differences were found between the three groups: As compared to Groups 2 and 3, participants in Group 1 achieved higher scores for Nav2 (median 78 vs 63 vs 48.5 for Groups 1, 2 and 3, respectively, p=0.01), BimT3 (92 vs 81 vs 63.5, p<0.01), PostH3 (53 vs 0 vs 0, p=0.02), ILMP3 (83 vs 50 vs 0, p=0.03), and overall score (466 vs 399.5 vs 272.5, p=0.01).
Conversely there was no difference according to ForT5 (median 89 vs 86 vs 41.5 for Groups 1, 2 and 3, respectively, p=0,37), or LasC3 (median 94 vs 90 vs 90, p=0.52).
Conclusions :
Overall, it was possible to develop a training program in virtual VR-surgery with construction validity for four out of six modules. Two modules did not show statistical differences between the groups and were excluded from the training program. These findings should be confirmed in a larger study.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.