Objective: To create a mortality risk model after pancreaticoduodenectomy (PD) using a Web-based national database system.
Background: PD is a major gastroenterological surgery with relatively high mortality. Many studies have reported factors to analyze short-term outcomes.
Subjects and methods: After initiation of National Clinical Database, approximately 1.2 million surgical cases from more than 3500 Japanese hospitals were collected through a Web-based data entry system. After data cleanup, 8575 PD patients (mean age, 68.2 years) recorded in 2011 from 1167 hospitals were analyzed using variables and definitions almost identical to those of American College of Surgeons-National Surgical Quality Improvement Program.
Results: The 30-day postoperative and in-hospital mortality rates were 1.2% and 2.8% (103 and 239 patients), respectively. Thirteen significant risk factors for in-hospital mortality were identified: age, respiratory distress, activities of daily living within 30 days before surgery, angina, weight loss of more than 10%, American Society of Anesthesiologists class of greater than 3, Brinkman index of more than 400, body mass index of more than 25 kg/m, white blood cell count of more than 11,000 cells per microliter, platelet count of less than 120,000 per microliter, prothrombin time/international normalized ratio of more than 1.1, activated partial thromboplastin time of more than 40 seconds, and serum creatinine levels of more than 3.0 mg/dL. Five variables, including male sex, emergency surgery, chronic obstructive pulmonary disease, bleeding disorders, and serum urea nitrogen levels of less than 8.0 mg/dL, were independent variables in the 30-day mortality group. The overall PD complication rate was 40.0%. Grade B and C pancreatic fistulas in the International Study Group on Pancreatic Fistula occurred in 13.2% cases. The 30-day and in-hospital mortality rates for pancreatic cancer were significantly lower than those for nonpancreatic cancer.
Conclusions: We conducted the reported risk stratification study for PD using a nationwide surgical database. PD outcomes in the national population were satisfactory, and the risk model could help improve surgical practice quality.