Mitigation of the Robotic Pancreaticoduodenectomy Learning Curve through comprehensive training

Background: There is an associated lag in achieving competency for robotic pancreaticoduodenectomy (PD), resulting in a learning curve. We hypothesize that the reported learning curve can be mitigated through a comprehensive graduated training protocol. Methods: All patients (n = 237) who underwent an open (n = 197, 83.1%) or robotic (n = 40, 16.9%) PD between 2015-2019 were identified at The Ohio State University. The learning curve for operative time and surgical failure (defined as conversion to open, blood transfusion, or Clavien-Dindo complication grade >3) was analyzed using a risk adjusted cumulative summation technique. Results: After 10 cases, operative time plateaued to a mean of 468.3 +/- 96.3 minutes for robotic PD versus a mean of 332.5 +/- 103.9 minutes for open PD (P < 0.001). There was no further apparent learning curve over time relative to rates of operative time or surgical failure. After propensity score-matching, patients undergoing robotic PD had a similar incidence of major complications, grade B/C postoperative pancreatic fistula, and delayed gastric emptying versus patients undergoing open PD (all P > 0.05). Conclusion: Completion of a comprehensive procedure-specific robotic training protocol for PD mitigated the learning curve for this operative approach by shifting the curve into the training/simulation phase rather than the live operating phase. These data hold important implications for the future training and accreditation of surgeons embarking on robotic PD.