Abstract A4: Do social class gradients predict breast cancer burden differently for black and white women?

Abstract

Abstract This research utilizes 50,000 cases of breast cancer diagnosed during 1992-2003 reported to the Maryland Cancer Registry, and linked by residential address to U.S. Census block group data. The goal of the research is to model social class patterns in breast cancer burden for white and black women, and examine the ways in which social class influences differ across racial groups. After geocoding cases to their Census blockgroup, we used factor analysis to create two indices capturing distinct aspects of area-level social resources. A composite index for social class captured area-level educational attainment, overall employment rates, portion of white collar employment, and income. The material resources index combined the portion of households with cars, telephones, living above poverty, and owning their own home. These indices were used to predict four aspects of case disease characteristics related to adverse outcome: tumor size, histologic grade, histologic type, and stage of disease at diagnosis. Multilevel random and fixed effects models estimated social resource effects across the State of Maryland's 3600 blockgroups. Race-specific models were used to examine whether social gradients in disease burden differed between black and white women. Results indicate that the two indices effectively capture social gradients in breast cancer burden for both ethnic groups, but that there are important differences. For white women, social class is a stronger predictor for tumor grade, size and type but both social class and material resources are predictive of stage at diagnosis. For black women, material resources are a stronger predictor for grade and tumor size, and social class is a stronger predictor for stage at diagnosis and tumor type. We found significant or borderline significant effects of race on the effect of both indices on all tumor characteristic outcomes. In almost all models, the direction of the interaction suggests that the protective effect of social resources on the outcome of interest is stronger for white women, than for black women. Additional analyses further examine the impact of neighborhood racial composition on social class influences. We conclude that investigating race-specific patterns of social class influences is one important strategy for understanding how social class patterns influence breast cancer disparities, as well as identifying possible avenues for intervention. Citation Information: Cancer Epidemiol Biomarkers Prev 2010;19(10 Suppl):A4.