Background: Pancreaticoduodenectomy (PD) has long been associated with high rates of morbidity and mortality. The objective of this study was to identify preoperative risk factors for serious complications and mortality after PD and to construct a prediction tool to facilitate risk stratification prior to surgery.
Materials and methods: Patients who underwent elective PD from 2005 to 2009 were identified from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) database. Multivariate logistic regression identified predictors of 30-day serious complications and mortality. A prediction tool was created and validated in a sample of 1254 patients.
Results: Of 4945 patients who underwent PD, 1342 (27.1%) suffered a serious complication and 127 (2.6%) died within 30 days. The most frequent complications were sepsis (15.3%), surgical site infection (13.1%), and respiratory complications (9.5%). After adjusting for potential confounders, the significant predictors of morbidity included older age, male gender, overweight and obesity, dependent functional status, chronic obstructive pulmonary disease (COPD), steroid use, bleeding disorder, leukocytosis, elevated serum creatinine, and hypoalbuminemia. Significant predictors of 30-day mortality included COPD, hypertension, neoadjuvant radiation therapy, elevated serum creatinine, and hypoalbuminemia. Multivariable models were used to construct a preoperative risk stratification tool.
Conclusions: Preoperative factors are associated with perioperative outcomes after PD. The prediction tool estimates the probability of early morbidity and mortality for patients undergoing PD. The tool may be used to provide information for patient counseling during the informed consent process and to identify high-risk patients for the purpose of tailoring perioperative care.