Mentorship and formal robotic proficiency skills curriculum improve subsequent generations' learning curve for the robotic distal pancreatectomy

Abstract

Background: Minimally invasive distal pancreatectomy is the recently accepted standard of care and national data show 50% of distal pancreatectomies are now performed in minimally invasive fashion. However, incorporating new surgical technologies necessitates overcoming a learning curve. Studies show the robotic distal learning curve to be 20-40 surgeries with operating time (ORT) as the most significant factor determining proficiency. This study evaluates whether formal mentorship and a robotic skills curriculum impact the learning curve and complications for second and third generation adopters. Methods: Consecutive robotic distal pancreatectomies ± splenectomy (RDP) from 2008 to 2017 were evaluated. Surgeries with concomitant procedures were excluded due to influence on ORT. First Generation was two surgeons who started program without training. Second Generation was the two subsequent surgeons who joined the program with direct mentorship. Third Generation was fellows who completed the entire procedure from the console after completing the proficiency based robotic skills curriculum. Multivariate models (MVA) were performed for ORT, clinically relevant pancreatic fistula (CR-POPF), and major complications (Clavien ≥ 3). Results: A total of 296 RDP were performed of which 187 did not include other procedures: First Generation (n = 71), Second Generation (n = 50), and Third Generation (n = 66). ORT decreased by generation (1st = 228.49 ± 80.27 min vs 2nd = 190.20 ± 45.31 min vs 3rd=171.83 ± 45.09 min; p < 0.001). There was no difference in other primary outcomes of CR-POPF (1st = 18.3% vs 2nd = 12% vs 3rd = 10.6%; p = 0.386) or Clavien≥3 (1st = 12.9% vs 2nd = 6.0% vs 3rd = 7.7%; p = 0.384) between generations. Estimated blood loss decreased by generation (1st = 165.99 ± 138.48 ml vs 2nd = 132.60 ± 109.66 ml vs 3rd = 106.67 ± 127.50 ml; p = 0.026). Lymph node harvest increased by generation (1st = 16.25 ± 10.33 vs 2nd = 22.78 ± 12.61 vs 3rd = 23.71 ± 14.44; p = 0.001). No other differences were found in pre-operative, intra-operative, or post-operative variables between generations. Patient factors predictive of ORT on MVA were pre-operative albumin (−36.17 min; p = 0.001) and pancreatic ductal adenocarcinoma (+29.96 min; p = 0.019). Surgeon factors predictive of decreased ORT were later generation (−28.76 min; p = 0.019) (Figure). Surgical generation was not predictive of CR-POPF nor Clavien ≥ 3 on MVA. Patient factors predictive of CR-POPF were increased BMI (Odds = 1.07 [1.00, 1.15]; p = 0.049) and neoadjuvant therapy (Odds = 7.30 [1.04, 51.30]; p = 0.046). No patient factors were identified to be predictive of increased Clavien ≥ 3. Fellow participation at the console increased over time with subsequent attendings. MVA showed that in patients with previous surgery fellows were less likely to complete entire case (Odds = 0.45 [0.22, 0.93]; p = 0.03) and they were more likely to complete entire case if operating with a 2nd Generation Surgeon (Odds = 3.66 [1.2, 11.13]; p = 0.022). Conclusion: In a high-volume center of excellence formal mentorship and a skills curriculum decreased the starting point and steepness of learning curve for operating room time in robotic distal pancreatectomy/splenectomy. Complications were largely dependent on patient factors and not affected by introduction of next generation surgeons. Health care delivery systems should seek to formally incorporate these strategies to maximize highest value and quality patient outcomes.