Risk Model for Distal Gastrectomy When Treating Gastric Cancer on the Basis of Data From 33,917 Japanese Patients Collected Using a Nationwide Web-based Data Entry System

Abstract

OBJECTIVE: To establish a risk model for distal gastrectomy in Japanese patients with gastric cancer. BACKGROUND: Risk stratification for distal gastrectomy in Japanese patients with gastric cancer improves surgical outcomes. METHODS: The National Clinical Database was constructed for risk determination in gastric cancer-related gastrectomy among Japanese individuals. Data from 33,917 gastric cancer cases (1737 hospitals) were used. The primary outcomes were 30-day and operative mortalities. Data were randomly assigned to risk model development (27,220 cases) and test validation (6697 cases) subsets. Stepwise selection was used for constructing 30-day and operative mortality logistic models. RESULTS: The 30-day, in-hospital, and operative mortality rates were 0.52%, 1.16%, and 1.2%, respectively. The morbidity was 18.3%. The 30-day and operative mortality models included 17 and 21 risk factors, respectively. Thirteen variables overlapped: age, need for total assistance in activities of daily living preoperatively or within 30 days after surgery, cerebrovascular disease history, more than 10% weight loss, uncontrolled ascites, American Society of Anesthesiologists score (≥ class 3), white blood cell count more than 12,000/μL or 11,000/μL, anemia (hemoglobin: males, <13.5 g/dL; females, <12.5 g/dL; or hematocrit: males, <37%; females <32%), serum albumin less than 3.5 or 3.8 g/dL, alkaline phosphatase more than 340 IU/L, serum creatinine more than 1.2 mg/dL, serum Na less than 135 mEq/L, and prothrombin time-international normalized ratio more than 1.25 or 1.1. The C-indices for the 30-day and operative mortalities were 0.785 (95% confidence interval, 0.705-0.865; P < 0.001) and 0.798 (95% confidence interval, 0.746-0.851; P < 0.001), respectively. CONCLUSIONS: The risk model developed using nationwide Japanese data on distal gastrectomy in gastric cancer can predict surgical outcomes.