Rural-urban disparities in pancreatic cancer stage of diagnosis: understanding the interaction with medically underserved areas

Purpose: To estimate differences in pancreatic cancer diagnosis stage by rurality of patient residence and residence in a medically underserved area (MUA). Methods: Using 2010-2016 Pennsylvania Cancer Registry data, we restrict our analysis to adults diagnosed with pancreatic cancer. We categorize each patient's residence by Rural-Urban Continuum Codes (RUCC): (1) metro; (2) nonmetro adjacent with population ≥20,000; (3) nonmetro adjacent with population <20,000; (4) nonmetro nonadjacent, and (5) completely rural; also by whether the county was a full MUA, partial MUA, or non-MUA. We examine the percent of patients diagnosed with local and locoregional stage of disease for each residential rurality and MUA status grouping. We estimate multivariate linear probability models of local and locoregional stage of diagnosis while controlling for demographics, insurance type, year, rurality, and MUA status. Finally, we estimate models interacting rurality with MUA status to disentangle the relative impact of each on diagnostic stage. Findings: We find that no pancreatic patients living in completely rural areas were diagnosed at the local stage, the lowest of all categories of rurality. In adjusted models, we find that patients living in completely rural areas have significantly lower rates of local and local/regional stage of diagnosis. The estimates are attenuated slightly when controlling for MUA. Conclusions: We find significantly lower rates of early stage pancreatic cancer diagnosis for patients living in completely rural areas. While lack of primary medical care resources is an important factor, rural residence remains an important independent predictor in later stage of diagnosis.