Introduction : Pancreaticoduodenectomy (PD) is a complex and high skill demanding procedure often associated with significant patient morbidity, with improved postoperative outcomes over the past two decades. However, there is a paucity of literature concerning the learning curve for PD. Therefore, the present study aims to analyze the factors that could influence it.

Methods : A prospective review of outcomes after PD during 2013-2019 was conducted. From 121 cases, 110 were PD (5 laparoscopic, 105 open) and 11 were total PD (1 laparoscopic, 10 open). Subsequent statistics were performed on the first 100 cases, with attention related with the learning curve.

Results : The most frequent histopathological tumor type was adenocarcinoma (41,32%). Thirty two percent of patients had preoperative biliary drainage and 37,19% presented with bactibilia. The preferred reconstruction technique included the pancreaticogastrostomy (99%), in situ hepaticojejunostomy (70%) and precolic gastrojejunal anastomosis (88%). Postoperative complications included biliary fistula (1%), pancreatic fistula (8%), pancreatic stump bleeding (4%), delayed gastric emptying (13%). The mean operative time decreased after the first 50 cases (P<0.001) and the blood loss after 60 cases (P=0.046). R1 resections decreased after 25 cases (P=0.025). Vascular resections(16%) do not influence the complications rate (p=0.8).

Conclusions : Outcomes improve as surgeon's experience increases, the proper training being the most important factor for minimizing the impact of the learning curve over the postoperative complications. Analysing the learning curve of a single surgeon is mandatory for accurate statistical results and interpretation.