E183: Social Vulnerability as a Neighborhood-level Determinant of Pancreatic Cancer Survivorship: A National Analysis of Veterans Affairs Patients

Introduction: Veterans Affairs (VA) patients experience lower barriers to access within the largest integrated U.S. health system, but still experience disparate mortality by race and community in several malignancies. The Social Vulnerability Index (SVI) includes multiple social determinants of health, summarizing economic deprivation, minoritized populations, poor access to housing and transportation, in census areas under 4000 individuals. SVI is implicated in differential morbidity and cancer center utilization in non-VA settings, but relationships between neighborhood and pancreatic cancer prognosis in veterans are not well-known. We therefore measured associations between SVI and pancreatic disease-specific survival (DSS) in a national VA cohort.

Methods: Patients with stage I-IV pancreatic adenocarcinoma diagnosed January 2005 to December 2018 were included. Demographics, comorbidities, tumor, and treatment characteristics were abstracted from the VA Corporate Data Warehouse. Addresses were geocoded to obtain SVI. Mann-Whitney U and Chi-square tests compared baseline characteristics. DSS was compared using Kaplan-Meier and multivariable Cox proportional hazards methods.

Results: 4,004 patients were identified, of which 500 (12%) lived in census tracts in the top SVI quartile. These patients were more likely to report black race (48% vs. 17%, p< 0.001), Hispanic ethnicity (4% vs. 2%, p< 0.001), be unemployed (55% vs. 38%, p< 0.001), unmarried (66% vs. 53%, p< 0.001), and have higher comorbidity (CCI≥3: 73% vs. 67%, p=0.02). In those with available histology (30%), high SVI patients had more poorly differentiated tumors (19% vs. 13%, p=0.02). There were no differences in treatment delay, receipt of non-VA cancer care, tumor location, clinical stage, neoadjuvant treatment, or therapeutic modality. Overall, median DSS was comparable [6.0 months (95% CI 5.6 - 6.8 months) vs. 7.0 months (95% CI 6.4 -7.6 months); log-rank p< 0.001]. In a sub-group of 633 (16%) who underwent resection, high SVI patients had a median DSS of 27 months (95% CI 19.3 - 29.6 months) versus 36 months (95% CI 27.7 - 46.6 months) for patients with low SVI (Figure 1). High SVI remained associated with mortality in multivariate analysis (HR 1.36, 95% CI 1.02 - 1.82, p=0.04).

Conclusions: High SVI was independently associated with poorer prognosis in patients undergoing pancreatic cancer resection. Acknowledging residual confounding, our findings highlight a potential need for neighborhood-level measures to improve postoperative survivorship interventions in veterans.