Small Area Estimation of Black-White Racial Disparity in Female Breast Cancer Survival Through a Bayesian Spatial Shared Component Model

Keywords: breast cancer, racial disparity, Bayesian spatial shared component model
Abstract Type: Paper Abstract

Abstract

Female breast cancer is the most diagnosed cancer in Georgia. The literature has consistently found significant Black-White disparities in breast cancer incidence and mortality by geography and sociodemographic characteristics. However, the spatial-temporal patterns of racial disparity in female breast cancer survival have not been well understood. Women diagnosed with breast cancer were identified from the Surveillance, Epidemiology, and End Results (SEER) program database. A Bayesian spatial shared component modeling approach was used to jointly model the county-level relative excess risk (RERs) of female breast cancer for non-Hispanic Black and non-Hispanic White women in Georgia from 2000 to 2017. The results identified counties with high and low RERs unique to non-Hispanic Black and non-Hispanic White women as well as counties with shared RERs. Counties of high shared relative excess risk exist in the southeast of Georgia. The middle of North Georgia has areas of low shared risk. However, after accounting for income, the shared risk is attenuated such that only parts of them remain at an elevated shared risk. After adjusting for the county-level socioeconomic position (measured by Darden–Kamel Composite Index), the between-county variance of the shared RERs slightly changed. The analysis and modeling approach in this study can be a generally applicable method to measure and visualize disparity, evaluate the effectiveness of interventions, and generate new hypotheses about health disparities across geography and sociodemographic characteristics.