G. T. Kennedy1, M. T. McMillan1, M. Sprys1, J. A. Drebin1, C. M. Vollmer1 1Hospital Of The University Of Pennsylvania,Philadelphia, PA, USA
Background: Surgeons with greater experience have demonstrated improved outcomes following pancreatoduodenectomy (PD), but little is known about what distinguishes their practice. Furthermore, the concept of experience has been variably interpreted in the surgical literature—some authors prioritize years in practice while others delineate cumulative career procedure volume or annual practice productivity. We sought to identify how these various forms of surgical experience influence management decisions for PD.
Methods: A survey assessing PD experience and practice patterns of pancreatic surgeons was distributed by six international gastrointestinal surgical societies. Questions regarding the practice patterns of individual surgeons were presented using a modified Likert scale. Multivariable, stepwise logistic regression analysis identified factors associated with three different forms of experience: years in practice, surpassing the PD learning curve (≥50 PDs), and high annual PD volume (≥25 PD/year).
Results: The median career PD volume of the 861 respondents was 80 (Interquartile range: 30-200). With regard to years in practice, forty-four percent of the surgeons were early-career (≤10 years of practice following training), 30% mid-career (11-20 years), and 26% senior-career (>20 years). Senior surgeons were more likely to use pancreatogastrostomy, dunking/invagination, external stents (all P<0.05). In terms of career total volume, the PD learning curve was surpassed by only 65% of respondents. Fewer early-career surgeons (39%) have attained this threshold compared with mid-career (80%) and senior-career (89%) surgeons (P<0.00001). Regression analysis identified factors independently associated with surpassing this learning curve of ≥50 cases (Table). Regarding annual productivity, surgeons in the upper quartile of annual PD volume (≥25/year) also demonstrated certain practice patterns; they were more likely to use the isolated roux limb technique (P=0.044) and the same type of pancreatico-enteric reconstruction on every case (P=0.016), but less likely to use autologous tissue patches (P=0.003) and multiple drains (P=0.0002).
Conclusion: The concept of experience in pancreatoduodenectomy encompasses several components: years of practice, attainment of the learning curve, and annual productivity. Each of these notions appear to influence decision-making during this complex operation in different ways.
