Predictors of Short-Term Readmission After Pancreaticoduodenectomy.

BACKGROUND: Readmissions are a common complication after pancreaticoduodenectomy and are increasingly being used as a performance metric affecting quality assessment, public reporting, and reimbursement. This study aims to identify general and pancreatectomy-specific factors contributing to 30-day readmission after pancreaticoduodenectomy, and determine the additive value of incorporating pancreatectomy-specific factors into a large national dataset.

METHODS: Prospective American College of Surgeons-National Surgical Quality Improvement Project (ACS-NSQIP) data were retrospectively analyzed for patients who underwent pancreaticoduodenectomy (PD) between 2011 and 2015. Additionally, a subset of patients with pancreatectomy-targeted data between 2014 and 2015 were analyzed.

RESULTS: Outcomes of 18,440 pancreaticoduodenectomies were analyzed, and found to have an 18.7% overall readmission rate. Multivariable modeling with pancreatectomy-specific variables increased the predictive value of the model (area under receiver operator characteristic 0.66 to 0.73). Statistically significant independent contributors to readmission included renal insufficiency, sepsis, septic shock, organ space infection, dehiscence, venous thromboembolism, pancreatic fistula, delayed gastric emptying, need for percutaneous drainage, and reoperation.

CONCLUSIONS: Large registry analyses of pancreatectomy outcomes are markedly improved by the incorporation of granular procedure-specific data. These data emphasize the need for prevention and careful management of perioperative infectious complications, fluid management, thromboprophylaxis, and pancreatic fistulae.