Understanding rural mortality disparities in cancer: a multi-level approach

点击翻译按钮获取中文摘要

Abstract Cancer mortality rates in the US declined by 26% between 1991 and 2015, but the decline has not been equal across all populations. Approximately 19% of the US population resides in rural areas; for over three decades, this population has experienced increasingly inferior outcomes when compared with their urban counterparts. The underlying cause(s) of the widening rural/urban gap in cancer outcomes is hypothesized to be multifactorial, with socioeconomic status (SES) both at the individual-level and geographic area-level likely playing a significant role. Gaps in evidence include the use of multiple “rural” definitions, a lack of evidence on associated geographic factors, and limited evidence based on multi-level approaches to understand the complex nature of rural disparities. Therefore, the overall objective of this study is to conduct a comprehensive examination of the underlying causes of rural/urban disparities in mortality among individuals diagnosed with cancer. In the first aim, the Surveillance, Epidemiology, and End Results (SEER) population-level data will be used to examine mortality among individuals diagnosed with cancer across three definitions of rurality, providing researchers and policy makers with the magnitude of differences by each definition. In the second aim, SEER will be linked via county- indicators to 16 databases (US Census Bureau, Area Deprivation Index, Bureau of Labor Statistics, County Health Rankings and Roadmaps, AMA Healthcare Workforce Mapper, BRFSS, Social Vulnerability Index, Health Information National Trends Survey, etc.). These linkages will allow us to estimate the contribution of specific area- level factors (e.g., area-level SES, access to high-quality care) on rural/urban mortality differences using effect decomposition methodology. In the third aim, the Reasons for Geographic and Racial Differences in Stroke (REGARDS) cohort (U01 NS041588) will be linked to state level cancer registry data using the Virtual Pooled Registry Cancer Linkage System (VPR-CLS). REGARDS includes longitudinally collected data for 30,239 participants (44% blacks), oversampled from stroke belt/buckle states (56% of participants from NC, SC, GA, TN, AL, MS, AK, LA) with large rural populations experiencing the highest mortality. REGARDS collects information at the individual level on sociodemographics, health literacy, and distance to healthcare, transportation, and risky health behaviors. Multi-level modeling and mediation modeling approaches will allow for the examination of the contribution of individual-level characteristics and the area-level characteristics simultaneously. Results will provide estimates of how much of the mortality disparity is explained by differences in urban and rural geographic characteristics overall as well estimates that describe the potential impact of hypothetical interventions on specific mediating factors. Findings will provide the critical evidence needed to inform policy and intervention development aimed at addressing the systemic disparities in mortality experienced by rural patients with cancer.