Danilo Miskovic, MD FRCS, Melody Ni, PhD, Susannah M Wyles, MSc MRCS, Amjad Parvaiz, FRCS, George B Hanna, PhD FRCS. Imperial College, London, United Kingdom
The objective of this study was to investigate if it is possible to establish construct and predictive validity for the observational clinical human reliability assessment (OCHRA) method in laparoscopic colorectal surgery (LCS) within a group of specialist surgeons (consultants).
Thirty-one full-length videos (15 right and 16 left colectomies) performed by 20 different established UK colorectal surgeons participating within the National Training Program in LCS were analysed. Of those, 10 were performed by experienced laparoscopic surgeons (“experts”, >200 laparoscopic colorectal resections) and 21 by less experienced (“apprentices”, <50 resections). A detailed task and error analysis was created and each video was assessed using scientific rating software. Procedures were broken down into 4 main task areas (exposure, vascular pedicle dissection, mobilization of colon, resection/anastomosis). Errors included instrument and tissue handling errors. Immediate consequences (e.g. bleeding, organ injury) were also noted. In addition, for the sub-task “lateral mobilization of colon”, the time spent on dissection of tissue and exposure were measured. The ratio between dissection to exposure time served as an indicator for efficiency (E:D ratio). For each video, two independent, blinded experienced surgeons (“faculty”) were also asked to globally assess the video with the aim to conclude if the performing surgeon demonstrated adequate skill to carry out such surgery independently (“pass/fail/inconclusive”). Between group differences were analyzed using the Mann Whitney U test. A Bayesian network was built to predict the global outcome (“pass/fail or inconclusive”) based on OCHRA data.
Based on the faculty ratings, 18 cases passed, 9 failed and 4 were inconclusive. All expert cases passed. A total of 385 errors were recorded by OCHRA. There was a significant difference between experts and apprentices for total number of errors (p=0.001), instrument errors (p=0.002), tissue errors (p=0.001) but not for consequential errors (0.092). Comparing surgeons who passed with those who failed showed similar results but there was also a significant difference found for consequential errors (p=0.002). Excluding the experts, the significant difference remained only for tissue errors (p=0.017), and for consequential errors (p=0.009). Overall, the between group differences were more distinct in the vascular dissection (p=0.146) and mobilization task (p=0.048) than other task areas. The predictive validity achieved through the Bayesian network was high (area under the ROC curve = 0.825). The model showed that tissue errors and consequential errors committed during vascular pedicle dissection, mobilization and resection of the colon were more diagnostic for the global outcome than other error types or task areas. Regarding efficiency, there was a significant difference between E:D ratios for experts and apprentices (0.98 vs. 0.53, p=0.002), but not between the pass and fail groups.
These data suggest that proficiency in advanced laparoscopic surgery can be assessed using OCHRA. The Bayesian network showed that the outcomes of two independent expert surgeons could be accurately predicted. The E:D ratio can be used as an indicator for expertise. This is promising data towards a very objective assessment tool that allows problem areas of individual trainees to be pinpointed, and thus providing detailed feedback reports.
Program Number: S089