Background: Pancreatic cancer is often diagnosed at advanced stages and is highly fatal. Many tumors are not surgically resectable. We aimed to identify features associated with survival in patients with surgically non-resected pancreatic cancer in the Military Health System.

Methods: We used the MilCanEpi database to identify Department of Defense beneficiaries aged 18 and older diagnosed with a primary pancreatic adenocarcinoma between January 1998 and December 2014 who did not receive oncologic surgery as treatment. We used Cox Proportional Hazard regression with step-wise procedures to select the socio-demographic and clinical characteristics related to 2-year overall survival, expressed as adjusted hazard ratios (aHR) and 95 % confidence intervals (CIs).

Results: Among 1,148 patients with surgically non-resected pancreatic cancer, sex, race-ethnicity, marital status, and socio-economic indicators were not selected in association with survival. A higher comorbidity count (aHR 1.30, 95% CI 1.06-1.59 for 5 v. 0), jaundice at diagnosis (aHR 1.57, 95% CI 1.33-1.85 v. no), tumor grade G3 or G4 (aHR 1.32, 95% CI 1.05-1.67 v. G1/G2), tumors location of pancreas tail (aHR 1.49, 95% CI 1.22-1.83 v. head) or body (aHR 1.30, 95% CI 1.04-1.62 v. head), and metastases were associated with survival. Patients receiving chemotherapy (aHR 0.66, 95% CI 0.57-0.76) had better survival compared to no treatment.

Conclusion: In a comprehensive health system, socio-demographic characteristics were not related to survival in surgically non-resected pancreatic cancer. This implicates access to care in reducing survival disparities in advanced pancreatic cancer and emphasizes the importance of treating patients based on clinical features.