The Learning Curve in Pancreatic Surgery
Jennifer F Tseng1, Jeffrey E Lee2, Peter WT Pisters2, Charlotte C Sun2, Henry F Gomez2, Douglas B Evans2
1University of Massachusetts Medical School, Worcester, MA;2UT MD Anderson Cancer Center, Houston, TX
Objectives: to investigate the learning curve for pancreaticoduodenectomy.
Design: retrospective record review.
Setting: single institution.
Patients: 650 consecutive patients who underwent pancreaticoduodenectomy 1990-2004 by three surgeons who began their attending careers at one center.
Interventions: Operative time, estimated blood loss (EBL), length of stay (LOS), and margins (for pancreatic adenocarcinoma) were analyzed. The chi-square and Mann-Whitney U tests were used to evaluate categorical and continuous variables, respectively.
Main Outcome Measures: Using serial groups of 20 cases, median OR time, EBL, and LOS were calculated and the trend over time modeled using 3rd-order polynomial equations, and retroperitoneal margins (R0/R1) were assessed.
Results: Surgeons had initial variability in EBL, OR time, and LOS over their 1st 60 pancreaticoduodenectomies, followed by improvement over their 2nd 60 cases. EBL dropped (1100 vs. 725 mL,p<.001), OR time decreased (589 vs. 513 min,p<.001), and LOS diminished (15 vs. 13 days,p=.004)(FigA-C). The proportion of positive/suspicious pancreatic adenocarcinoma margins (R1) decreased between surgeons' 1st and 2nd 60 cases (30% vs. 8.2%,p<.001). Extended analysis of a single surgeon's cases suggested that although the greatest progress occurred between the 1st and 2nd 60 cases, additional cases provided further incremental improvement (p<.001)(FigD-F).
Conclusions: Pancreaticoduodenectomy has an inherent learning curve. After 60 cases, surgeons achieved significantly decreased blood loss, operative time, and hospital stay, and performed more margin-negative resections. Improvement in measured outcomes continues during the operative career. 
Back to Final Program
