Robotic Whipple Biotissue Curriculum Improves Technical Performance for Fellows and Has Construct Validity

Melissa E Hogg, MD, MS1, Mazen Zenati1, Stephanie Novak1, Yong Chen2, Amer H Zureikat1, Herbert J Zeh III1. 1UPMC, 2Chinese Hospital

INTRODUCTION – Many view the pancreaticoduodenectomy as the most complex intra-abdominal operation. Additionally, the past five years have seen a rise in performing the procedure with minimally invasive technology. A major criticism of incorporating the robotic platform for the pancreaticoduodenectomy is that it is difficult to teach and disseminate the technique. Our group has experience with >300 robotic pancreaticoduodenectomies and has published a learning curve of 80 cases to optimize performance.  We hypothesize by instituting an advanced robotic training curriculum we can decrease that learning curve for robotic whipples.


METHODS – A three-step curriculum: 1) simulation, 2) biotissue and 3) operative coaching was implemented. The biotissue curriculum consisted of sewing artificial organs to simulate a hepaticojejunostomy (HJ), gastrojejunostomy (GJ) and pancreaticojejunostomy (PJ). These were evaluated for time, errors and Birkmeyer Score (30 maximum, 6 categories with a Likert scale of 5). Three attendings with experience of greater than 80 robotic pancreaticoduodenectomies performed each biotissue anastomosis once for validity.  Two blinded graders scored all videos.


RESULTS – Fourteen fellows performed 195 anastomoses during first year: 66 (HJ), 64 (GJ) and 66 (PJ). The attendings’ first attempt outperformed the fellows’ first attempt in all 9 categories. The fellows’ performances were analyzed as a group by attempt. For the HJ, time, errors and Birkmeyer all improved linearly over 7 analyzed attempts (p<0.007).  For the GJ, time, errors and Birkmeyer all improved linearly over 9 analyzed attempts (p<0.002). For the PJ, errors and Birkmeyer both improved linearly over 8 analyzed attempts (p<0.002); however, time trended down without plateauing but did not reach statistical significance (p=0.08).  The attendings’ first attempts were faster than fellows’ last attempt for all anastomoses (p<0.041).  For the GJ and PJ but not HJ, errors and Birkmeyer were on par between attendings’ first and fellows’ last attempt. The graders scoring correlated for errors and Birkmeyer (p<0.0001).  Since incorporating the biotissue curriculum, every fellow completing a three-month rotation has successfully performed a robotic pancreaticoduodenectomy (p<0.05; compared to before the curriculum).


CONCLUSION – A pancreaticoduodenectomy biotissue curriculum has face and construct validity.  It can improve fellow performance in both an inanimate and operative environment.  Time is the most difficult parameter in which to reach attending mastery. The PJ takes more attempts to show improvement, and the HJ is the most difficult to obtain attending mastery performance.  This curriculum is a valid tool for teaching robotic pancreaticoduodenectomies with established milestones for reaching optimum performance. 

« Return to SAGES 2016 abstract archive