Variability in postoperative resource utilization after pancreaticoduodenectomy: Who is responsible

Background. We aimed to quantify and predict variability that exists in resource utilization after pancreaticoduodenectomy and determine how such variability impacts postoperative outcomes. Methods. The University HealthSystems Consortium database was queried for all pancreaticoduodenectomies performed between 2011-2013 (n = 9,737). A composite resource utilization score was created using z-scores of 8 clinically significant postoperative care delivery variables including number of laboratory tests, imaging tests, computed tomographic scans, days on antibiotics, anticoagulation, antiemetics, promotility agents, and total number of blood products transfused per patient. Logistic, Poisson, and gamma regression models were used to determine predictors of increased variability in care between patients. Results. Having a high (versus low) resource utilization score after pancreaticoduodenectomy correlated with increased duration of stay; (odds ratio 2.28), cost (odds ratio 1.89), readmission rate (odds ratio 1.46), and mortality (odds ratio 7.54). Patient-specific factors were the strongest predictors and included extreme severity of illness (odds ratio 114), major comorbidities/complications (odds ratio 5.99), and admission prior to procedure (odds ratio 2.72; all P < .01). Surgeon and center volume were not associated with resource utilization. Conclusion. Public reporting of patient outcomes and resource utilization, invariably tied to reimbursement in the near future, should consider that much of the postoperative variability after complex pancreatic operation is related to patient-specific risk factors.