Impact of surgical experience on management and outcome of pancreatic surgery performed in high- and low-volume centers.

Pancreaticoduodenectomy (PD) is one of the procedures in general surgery with the highest rate of life-threatening complications. The positive impact of the volume-outcome ratio on outcomes and mortality in pancreatic surgery (PS) has led to policy-level efforts toward centralization of care for PS that is currently under evaluation by some Regional Health Services. The role of the surgeon's experience and training is still under debate. The aim of this paper is to compare the outcomes of PS by the same surgeon in a high volume (HV) and in a low volume (LV) hospital to assess whether a specific training in PS could outdo the benefits of hospital volume. 124 pancreatic resections (98 PD) were conducted by a single surgeon from 2004 to 2014 in two different Italian hospitals with different PS volumes as well as in general surgical activities. The results were retrospectively analyzed. All data regarding demographics, oncological characteristics, surgical parameters and post-operative outcomes were compared between patients operated on in the HV (group A) and LV hospital (group B). The surgical experience in the LV hospital has been then divided into a first period (group B1) and in a second period (group B2). χ 2 test or Fisher's exact test (when variables were dichotomous) was used. The unpaired t test was used to compare continuous data between the two groups. Values are expressed as n. of cases (percent) for categorical data or as mean (standard deviation) for continuous data. A p value less than 0.05 was considered as significant. From 2004 to 2014, 124 patients underwent pancreatic resection by the same surgeon: 69 in an HV hospital (group A) and 55 in an LV hospital (group B). We focused our attention on PD outcomes, 54 in group A and 44 in group B (22 in group B1 and 22 in group B2, accordingly to the aforementioned criteria). A higher incidence of ASA 3 patients, although not statistically significant, was found in group B than in group A (34 vs. 18%; p = 0.064). With regard to post-operative outcome between group A and B, no statistical differences were found in mortality rate (4 vs 7% p = n.s.), morbidity rate (overall, medical and surgical), Clavien-Dindo complications grade, reoperation rate, pancreatic fistula rate and grade, and post-operative length of stay. Oncologically, there were no differences in lymph nodes retrieval between the two groups. With regard to comparison between the two LV hospital groups, mortality rate was nearly significantly higher in group B1 than in group B2 (14 vs. 0%; p = 0.073), whereas no differences were found in the comparison between group A (4%) and group B2 (0%) (p = n.s.). A previous surgical experience in an HV hospital overcomes or reduces the differences in the outcome of pancreatic surgery reported in the literature between HV and LV hospitals. There was a time-related improvement trend in terms of post-operative mortality in the LV, probably related to the accustomedness and skills in managing severe complications related to PS. The surgeon's experience together with the selection of patients, the availability of resources and the development of team experience at LV hospital are probably important variables which can overcome hospital volume and should, therefore, be taken into account in PS accreditation programmes.