County-Level Factors Associated with High Age-Adjusted Incidence Rates of Pancreatic Cancer in the US

Introduction: This study examined demographic, socioeconomic, and geographic associations with county-level pancreatic cancer incidence in the US. Methods: We obtained county-level age-adjusted incidence rates of pancreatic cancer (2012-2016) from Surveillance, Epidemiology, and End-Results. We merged data from the American Community Survey (demographics), USDA Economic Research Service (Urban Influence Code, vehicle availability, food desert, income/access census tracts), County Health Rankings (Primary Care Provider ratio, smoking), US Diabetes Surveillance System (diabetes, inactivity, obesity), and the US Renal Data System (end-stage renal disease). We categorized variables based on quartiles and employed chi-square tests to evaluate associations with having high incidence of pancreatic cancer greater than the national average (10.3/100,000). A logistic regression model with backward selection estimated the odds (odds ratio, 95%CI) of having an incidence of pancreatic cancer greater than the national average. Results: In a univariable setting, higher percentages of black population, inactivity, diabetes, obesity, smokers, uninsured, food deserts, low vehicle availability, and lower median income are associated with high pancreatic cancer incidence. The only variable that remained after backward selection was current smokers. Compared with counties in the 1st quartile, counties in the 3rd (5.4, 2.0-14.1) and 4th (4.3, 1.7-11.2) quartiles of percent current smokers were more likely to have high pancreatic cancer incidence. Conclusion: Our results suggest that intervention focused on social determinants of health, particularly lowering tobacco use, could decrease the pancreatic cancer burden. Future investigation should leverage implementation science to alleviate pancreatic cancer disparities, particularly for medically underserved populations that lack access to specialized cancer services.