Learning Curve of a Laparoscopic Pancreaticoduodenectomy Program at a Second Institution

Abstract

Background Historically, pancreaticoduodenectomy (PD) has been performed via a laparotomy, but increasingly, laparoscopic and robotic platforms are being employed for PD. Laparoscopic PD has a steep surgeon specific learning curve and programmatic elements that must be optimized. These factors may limit a surgeon who is proficient at laparoscopic PD to develop a program at another institution. We hypothesize that the learning curve for a surgeon transferring a program to a second institution is shorter than the initial laparoscopic PD learning curve for the same surgeon. Methods A retrospective review of patients who underwent laparoscopic PD for any indication at the first institution (FI) from 2012 to 2017 and the second institution (SI) from 2018 to 2021 was conducted. Standard statistical analysis was performed. The learning curve was identified using one-sided CUSUM analysis of operative times. Result We identified 110 participants, 90 from the FI and 20 from the SI. More patients at the FI were diagnosed with periampullary adenocarcinoma on final pathology compared to the SI (65.6% vs 40.0%, P = .0132). FI operative times stabilized after the 25th laparoscopic PD and SI operative times stabilized after the 5th operation. No statistically significant difference was identified in postoperative complications. Conclusions The learning curve and average operative time of an SI laparoscopic PD program was shorter than the initial learning curve for a single surgeon with comparable outcomes. This suggests that complex minimally invasive surgical programs can be safely transferred to another high-volume institution without significant loss of progress.