Poster of Distinction–P014

ESTABLISHING THE RELIABILITY AND VALIDITY OF A VIRTUAL REALITY UPPER GASTROINTESTINAL SIMULATOR USING A NOVEL VIDEO-ENDOSCOPIC ASSESSMENT TECHNIQUE. Krishna Moorthy MBBS, Yaron Munz MD, Avril Chang MBBS, Mustafa Jiwanji, Ara Darzi MD, Department of Surgical Oncology and Technology, Imperial College of Science, Technology and Medicine, St. Mary’s Hospital, London, UK

Background
This study aims to establish the construct validity of an Upper GI simulator (Simbionix, Israel) and to cross validate its parameters using a novel video-endoscopic assessment technique developed by us.
Methods:
Study participants were requested to undertake 2 modules on the simulator, twice each. The simulator’s assessment of percentage of mucosa visualized fails to give details regarding the part of the upper gastro-intestinal tract which was missed or inadequately visualized. Using a video-endoscopic assessment method developed by us, two independent blinded observers rated the performance by scoring different components of the procedure on a 5 point Likart scale. Statistical analysis was done using the Kruskal Wallis test, Mann-Whitney U test and Spearman’s correlation.
Results:
There were 11 novices (Group 1), 11 residents (20-80 procedures, Group 2) and 10 experienced endoscopists (> 200, Group 3). There was a significant difference in the total time taken to perform the procedure (p<0.001), percentage of mucosa visualized (p<0.001), percentage of pathologies visualized (p=0.002) and the number of inappropriate retroflexions (p=0.016) across the groups. Inter-group analyses were significant between 1 and 3 for all parameters and in between 2 and 3 for three of the parameters.
There was also a significant difference in the video-endoscopic score across the groups (p<0.001). This was significant between 1 and 3 (p<0.001) and 1 and 2 (p=0.001) but not between 2 and 3 (p=0.50). There was a significant correlation between the video-endoscopic score and percentage of mucosa visualized (rho= .64, p< 0.001) and number of inappropriate retroflexions (rho= 0.67, p<0.001).
Conclusion:
This study has established the construct validity of the simulator which can strongly discriminate between groups with different levels of experience.
The video-endoscopic method developed by us validates the simulator’s assessment parameters.


Up