Objectives: Our group investigates objective performance indicators (OPIs) to analyze robotic colorectal surgery. OPIs include kinematics, energy use, instrument clutching, camera movement and robotic arm usage. Analyses of OPI data are more difficult in dual-console procedures (DCPs) as there is no distinct surgeon identifier. Currently, there is no reliable, efficient technique to assign console-specific OPIs during a DCP. We developed a metric for assigning tasks to the appropriate surgeon during review of robotic proctectomy videos.
Methods: An experienced robotic colorectal surgeon and colorectal surgery fellow reviewed 21 unedited robotic proctectomy videos. The procedures were all DCPs (attending and trainee working from separate consoles) with no audio or visual information to identify the two surgeons. The two consoles have unique identifiers to discriminate console 1 vs console 2. The videos were annotated to delineate specific surgical tasks based on a previously developed annotation card. The two reviewers watched a small number of tasks for 30 seconds and assigned “attending” or “trainee” to each. Based on this sampling, the remainder of task assignments for the procedure was extrapolated.
For comparison, we defined a novel metric called the ratio of economies of motion (rEOM) as an OPI-based calculation to assign surgical tasks to console 1 or console 2. Conventional economy of motion (EOM) describes total linear distance traveled and is calculated from the console hand controllers and patient cart robotic arms. rEOM is calculated by dividing hand controller EOM by robotic arm EOM. We compared assignments provided by video review with assignments provided by OPI evaluation (Figure 1).
Results: During 21 videos, an average of 86.2 individual surgical tasks (total 1811) were performed. A median of 6.5 tasks were reviewed per video (total 137), resulting in an audit rate of 7.6%. Task assignment agreement was 91.2% (125/137) for video review vs rEOM. There were 13 tasks incorrectly assigned to trainees and all tasks were correctly assigned to attendings. By excluding 27 tasks with incomplete rEOM data, task assignment agreement improved to 99.5% for the remaining 1784 tasks. This contributed to an additional 12 operations having task assignment agreement (137/137).
Conclusion: rEOM is an accurate way to assign tasks to the appropriate surgeon during dual-console robotic proctectomy. This will be useful for ensuring tasks are attributed to the appropriate surgeon in future studies of OPIs. Our next research plans will focus on identifying OPIs that correlate with surgeon experience and patient outcomes.
