Validation of a colonoscopy simulation model for skills assessment

OBJECTIVE: The purpose is to provide initial validation of a novel simulation model's fidelity and ability to assess competence in colonoscopy skills. METHODS: In a prospective, cross-sectional design, each of 39 endoscopists (13 staff, 13 second year fellows, and 13 novices) performed a colonoscopy on a novel bovine simulation model. Staff endoscopists also completed a survey examining different aspects of the model's realism as compared to human colonoscopy. The groups' simulation performances were compared. Additionally, individual performances were correlated to patient-based performance data. RESULTS: Median model realism evaluation scores were favorable for nearly all parameters evaluated with mucosa appearance, endoscopic view, and paradoxical motion parameters receiving the highest scores. During simulation procedures, each group outperformed the less experienced groups in all parameters evaluated. Specifically, median cecal intubation times were: staff 226 s (IQR [interquartile range] 179-273), fellows 340 s (282-568), and novices 1,027 s (970-1,122) (P < 0.05). Median total procedure times on the model were: staff 468 s (416-501), fellows 527 s (459-824), and novices 1,350 s (1,318-1,428) (P < 0.05). Finally, individual cecal intubation times on the simulation model had a very high correlation to their respective patient-based times (r = 0.764). CONCLUSIONS: Overall, this model possesses a favorable degree of realism and is able to easily differentiate users based on their level of colonoscopy experience. More impressive, however, is the strong correlation between individual's simulated intubation times and actual patient-based colonoscopy data. In light of these findings, we speculate that this model has potential to be an effective tool for assessment of colonoscopic competence.