The influence of procedural volume on short-term outcomes for robotic pancreatoduodenectomy-a cohort study and a learning curve analysis.

BACKGROUND: An increasing number of robotic pancreatoduodenectomies (RPD) are reported, however, questions remain on the number of procedures needed for gaining technical proficiency in RPD. Therefore, we aimed to assess the influence of procedure volume on short-term RPD outcomes and assess the learning curve effect. METHODS: A retrospective review of consecutive RPD cases was undertaken. Non-adjusted cumulative sum (CUSUM) analysis was performed to identify the procedure volume threshold, following which before-threshold and after-threshold outcomes were compared. RESULTS: Since May 2017, 60 patients had undergone an RPD at our institution. The median operative time was 360 min (IQR 302.25-442 min). CUSUM analysis of operative time identified 21 cases as proficiency threshold, indicated by curve inflexion. Median operative time was significantly shorter after the threshold of 21 cases (470 vs 320 min, p < 0.001). No significant difference was found between before- and after-threshold groups in major Clavien-Dindo complications (23.8 vs 25.6%, p = 0.876). CONCLUSIONS: A decrease in operative time after 21 RPD cases suggests a threshold of technical proficiency potentially associated with an initial adjustment to new instrumentation, port placement and standardisation of operative step sequence. RPD can be safely performed by surgeons with prior laparoscopic surgery experience