INSTITUTIONAL AND SURGEON-SPECIFIC LEARNING CURVE ANALYSIS OF 350 ROBOTIC-ASSISTED ESOPHAGECTOMIES AT A HIGH VOLUME CANCER CENTER.
David T. Pointer*, Sabrina Saeed, Jacques-Pierre Fontaine, Sean Dineen, David Boulware, Biwei Cao, Jose M. Pimiento
Surgical Oncology, H. Lee Moffitt Cancer Center, Tampa, FL
Introduction: Evolving technology has led to the rapid adoption of robotic techniques in esophageal surgery. Recent MIS experiences have demonstrated prolonged learning curves for stabilized outcomes after esophagectomy. With the potential advantage of instrument dexterity and 3-D visualization, robotic-assisted minimally invasive esophagectomy (RAMIE) has emerged as an alternative to traditional open or video-assisted techniques. Our objective was to analyze the learning curve in terms of pertinent perioperative outcomes to guide surgeons and institutions on expectations during implementation of this complex operation.
Methods: We retrospectively reviewed 350 patients who underwent RAMIE for malignancy. After stratification by chronological quartiles, patient characteristics and perioperative outcomes were evaluated for our institutional cohort then for three surgeon-specific cohorts. Cumulative-sum analysis was used for operative time, estimated blood loss (EBL), lymph node harvest, length of stay and anastomotic leak rate to determine the learning curve inflection points in measurable proficiency end-points.
Results: Institutional incidence of anastomotic leak was 16%; median length of stay 9 [4;65] days; operative time 418 [200;813] minutes; EBL 200 [20;1700] mL. Quartile analysis by institution demonstrated a significant reduction in LOS (10d v 9d; p=0.02) between the first and fourth quartiles. Significantly decreased anastomotic leak rate was observed in two of three surgeons between the first and third quartile (23% v 0%, p=0.01) and third and fourth (28% v 4%, p=0.05), respectively. Approximately 150 Cases were required to reach institutional learning curve plateau for lymph node harvest and LOS and 170 cases required for leak rate plateau. Two inflection points were observed for EBL (125 and 250 cases), operative time (75 and 275 cases) and leak rate (170 and 225 cases). Surgeon-specific learning curves were variable. Two inflection points were observed for each surgeon for operative time, lymph node harvest and EBL at mean case number 24 and 82, 23 and 51, and 30 and 97, respectively. Anastomotic leak rate curves demonstrated two inflection points for each of the three surgeons at 30 and 120 cases (surgeon 1), 50 and 95 cases (surgeon 2), and 35 and 55 cases (surgeon 3). The inflection point for LOS was observed at an estimated mean of 32 cases (20;50).
Conclusion: Case volume required for proficiency differs for each metric assessed. The learning curve for complex cases such as RAMIE varies for institutions and individual surgeons based on prior experience and institutional support. The effect of new technology, faculty and assisting personnel cannot be fully captured.
Back to 2020 Abstracts