Learning curve of totally robotic pancreaticoduodenectomy in early-career surgeons: a dual analysis using CUSUM and moving average
Related Experiment Video
Updated: May 31, 2026
Technical Detail for Robot Assisted Pancreaticoduodenectomy
Published on: September 28, 2019
Boram Lee1, Ho-Seong Han2,3, Yoo-Seok Yoon1
1Department of Surgery, Seoul National University Bundang Hospital, Seoul National University College of Medicine, 82, Gumi-ro 173 Beon-gil, Bundang-gu, Gyeonggi-do, Seongnam, 13620, Republic of Korea.
Background:
Robotic pancreaticoduodenectomy (RPD) is increasingly performed, but most learning curve studies involved senior surgeons with laparoscopic pancreaticoduodenectomy (LPD) experience. Evidence is lacking for early-career surgeons transitioning directly from open surgery to robotics. This study aimed to evaluate the learning curve of RPD using dual analytic methods and to explore efficiency gains through sub-phase analysis of operative time.
Methods:
We retrospectively analyzed 78 consecutive RPD performed by a single early-career surgeon (May 2023-August 2025). Learning curves were assessed using cumulative sum (CUSUM) and five-case moving average analyses of operative time. Perioperative and postoperative outcomes were compared across phases. In addition, operative time was subdivided into docking, resection, and anastomosis to evaluate phase-specific efficiency gains.
Results:
CUSUM identified proficiency at case 23, and moving average showed stabilization at case 41, delineating three phases: initial learning (1-22), proficiency (23-41), and stabilization (42-78). Mean operative time decreased significantly across phases (523.6 → 309.3 min; p < 0.001). Clinically relevant pancreatic fistula declined from 31.8 to 5.4% (p = 0.012). Conversion to open surgery was 3.8%. Sub-phase analysis demonstrated that most reductions were attributable to shorter resection and docking times, while anastomosis time remained stable. Oncologic adequacy was preserved (mean lymph nodes 18.6, R0 resection 97.4%). No 90-day mortality occurred.
Conclusion:
In this single-surgeon series, operative proficiency appeared after approximately 23 cases and stabilized after 41 cases. Dual analytic methods revealed complementary insights into the learning curve, and sub-phase analysis highlighted that efficiency gains were primarily achieved in resection and docking. RPD may represent a feasible training pathway for early-career surgeons in structured high-volume environments with appropriate mentorship.
