One size doesn't fit all: different antibiotics are associated with decreases in surgical site infections in pancreaticoduodenectomy versus distal pancreatectomy

Abstract

Background: The utility of prophylactic antibiotics before skin incision to decrease surgical site infection is well-documented. However, there is limited data comparing the use of different classes of antibiotics to mitigate this risk in pancreatic surgery. Multiple different antibiotic regimens are utilized nationwide, and there remains a gap in knowledge as to the risks and benefits of various antibiotic choices. It is hypothesized that certain antibiotics may be associated with lower rates of surgical site infection in pancreatic procedures. Methods: Demographic, comorbidity, perioperative, intraoperative, and outcomes data were captured from the 2016 ACS National Surgical Quality Improvement Program (NSQIP) targeted pancreatectomy database. Pre-operative antibiotics were classified into first generation cephalosporin (1stCE), second or third generation cephalosporin (2-3CE), and broad-spectrum antibiotics (br-ABX). Br-ABX include anti-pseudomonals, ticarcillin-clavulanate, piperacilin/tazobactam, fourth generation cephalosporins, aminoglycosides, and fluoroquinolones. Univariate and multivariate analysis was performed to compare various outcomes between the different groups. Subset analysis was performed by surgery type (pancreaticoduodenectomy vs. distal pancreatectomy) and approach (open vs. minimally invasive (MIS)). Distal pancreatectomy with enteric reconstruction were excluded. The primary end-point was surgical site infection. Results: Of the 5,803 pancreatectomies performed throughout the U.S. in 2016, 38% received a first generation cephalosporin (1CE), 26% a second or third generation cephalosporin (2-3CE), and 36% a broad spectrum antibiotic (br-ABX). After controlling for patient demographics, comorbidities, perioperative variables including blood administration, gland characteristics, receipt of neoadjuvant therapy, stent and drain use; br-ABX, and not 2-3CE, was associated with a decrease in superficial surgical site infection (SSI) (OR = 0.67, p = 0.014) and organ space SSI (OR = 0.81, p = 0.05) compared to 1CE. The decreases in SSI occurred after open and not MIS pancreatectomy (p < 0.006). Subset analysis identified that the br-ABX-associated decrease in superficial and organ space SSIs were only associated with open pancreaticoduodenectomy (superficial: p = 0.003; organ space: p = 0.025), and not open distal pancreatectomy. Interestingly, an increase in organ space SSI was observed with br-ABX for MIS distal pancreatectomy (OR = 3.15, p = 0.002). For all pancreatectomies, there was a two-fold increase in clostridium difficile infection with 2-3CE (OR = 2.08, p=0.014), and a two-fold decrease in urinary tract infections (UTI) with Br-abx (OR = 0.52, p = 0.012), compared to 1CE. There were no differences in deep incisional SSI, wound dehiscence, sepsis, and length of stay or readmission rates across all groups of antibiotics. Conclusion: Broad-spectrum antibiotics are associated with decreased SSIs after open pancreaticoduodenectomy, but increased SSIs after MIS distal pancreatectomy. There was no difference in SSI for MIS pancreaticoduodenectomy or open distal pancreatectomy regardless of antibiotic regimen. Thus, since 2-3CE were associated increased clostridium difficile infection, Br-ABX may be appropriate for open pancreaticoduodenectomy, and 1CE for MIS pancreatectomy and all distal pancreatectomy in order to decrease SSIs. Prospective trials are warranted.