Abstract A5: Spatially informed investigations of race-specific social gradients in breast cancer disparities

Background: The goal of this DOD-funded Synergistic Idea Award is to examine the extent to which social class, health behaviors, and access to health care, measured at the community level, explain geographic variation in breast cancer burden among African American and white women in the State of Maryland. The goal of this exploratory scientific activity is to elucidate possible proximal mechanisms that drive enduring but poorly understood associations between African American ethnicity and aggressive breast cancer biology, late-stage diagnosis, and excess disease burden. Although ethnic disparities are often speculated to be caused by social and economic resource differentials, few studies have examined the race-specific role of social resources on breast cancer outcomes or explicitly tested how specific elements of social class actually drive biological outcomes. Thus, it is difficult to identify which aspects of social class disparities should be prioritized for more refined studies or for cancer control planning and interventions.

Methods: Data: Data sources include 50,000 cases of breast cancer occurring in Maryland in 1992–2003, reported to the Maryland Cancer Registry. We have geocoded the residential address of each case to a point location within the State and linked to area-level descriptive information on social class, consumer behaviors, and health services availability. Social class measures are created using 1990 and 2000 Census data. Consumer purchasing behaviors are created from commercially available coverages, based on the Bureau of Labor Statistics Consumer Expenditure Surveys. Health services data are derived from geographically referenced Medicare identifier codes.

Methods: Statistical Analyses: Variation in incidence, histologic grade and other tumor characteristics, stage of disease at diagnosis, and to the extent data permit, estrogen receptor status, lymph node involvement, and survival, are being examined, using multilevel modeling. Outcomes are modeled as a function of individual case characteristics, including age and year of diagnosis, as well as Census area behavioral profiles, socioeconomic resources, and availability of health services.

Analyses identify single behaviors as well as composite characteristics of communities having excess breast cancer burden for African American cases and compare these significant characteristics to those associated with excess burden for white cases. Finally, spatial analysis techniques, including unadjusted and adjusted cluster analyses, will examine the extent to which these significant behavioral and social risk factors explain geographic patterns of breast cancer disparities in the state.

Findings to Date: Consistent with previous research, African-American women present at earlier ages than white women, with more advanced, and more aggressive disease. However, preliminary analyses suggest some of this excess burden is due to social class differences. Furthermore, analyses suggest that behavioral factors may explain some of the social class gradient. Next steps include factor analytical techniques to reduce and refine both social class and behavioral measures.