Racial, Socioeconomic, and Geographic Disparities of Female Breast Cancer in Texas
MetadataShow full metadata
Breast cancer, as the most common cancer among women, has been reported to display remarkable health disparities in the continuum of late-stage diagnosis, utilization of mammography, treatment options, as well as survival and mortality. These inequalities are experienced by different subpopulations and related to a complex set of factors, such as education, socioeconomic, race, geography, unequal access to health resources, and health-related policies. Existing knowledge about these factors and mechanisms causing these disparities in breast cancer outcomes is limited. Few studies have examined how racial disparities in breast cancer vary across geographic regions, which is key information for any attempt to allocate limited health resources more effectively and efficiently. To date, research on racial and socioeconomic disparities of breast cancer mainly proceeds by combining epidemiological data and investigating risk factors for the population under study. By doing so, one overlooks the geographical variations of breast cancer outcomes, a piece of information that is critical to target regions in intervention programs. Investigating racial disparities across regions can provide useful insights and reveal unknown risk factors for breast cancer, thereby helping health-policy makers to improve the overall health outcome of breast cancer among women.
Within the framework of Geographic Information Systems (GIS), this research conducted spatial and statistical analysis to identify the census tracts that displayed significant racial disparities in late-stage diagnosis and breast cancer mortality for both African-American and Hispanic women compared with their non-Hispanic white counterparts in Texas from 1995-2005. These disparities were measured in terms of rate difference (RD) and rate ratio (RR) and accounted for the population size of each census tract. The significance of these disparities was evaluated statistically and the results were corrected for multiple testing using the false discovery rate approach. For African-American women with the RD measurement, 278 and 188 census tracts displayed significant racial disparities for breast cancer mortality and late-stage diagnosis respectively in the 4,388 census tracts in Texas. These figures were larger for Hispanic women, with 328 and 266 census tracts respectively. Fewer Census tracts tested significant for the RR measurement. Most of the census tracts with significant racial disparities were located in the metropolitan areas of Houston, Dallas, and Austin-San Antonio for African-American women. Hispanics were also found to have significant racial disparities in the Southwest border of Texas. Logistic regression between the significance of the RD statistic for the two types of health outcomes indicated that a census tract with significant racial disparities for late-stage diagnosis was 30 times more likely to test significant for racial disparities in breast cancer mortality.
Logistic regression was also utilized to investigate the spatial connection of significant racial disparities in late-stage diagnosis and breast cancer mortality. The socioeconomic status (SES) was categorized into groups of low, middle, and high based on the percentage of population living under the federal poverty line. For the two minority groups, low and middle SES census tracts were more likely to report significant racial disparities for both health outcomes. About 40% of the census tracts with significant racial disparities for breast cancer mortality also displayed significant disparities for late-stage diagnosis. Linear regression was then used to quantify the relationships between the magnitude of racial disparities in mortality and late-stage diagnosis for breast cancer. The correlation coefficient was 0.23 for the RD measurement and 0.45 for the RR measurement for both minority groups. Moran‘s I, however, indicated the presence of spatial autocorrelation in the regression residuals, which might reflect the non-stationarity of the regression coefficients and/or the existence of unknown spatial factors. Therefore, the regression models were weighted geographically to account for the spatial variations of observations.
Furthermore, potential risk factors such as demographic characteristics, SES, and spatial accessibility were added to racial disparities in late-stage diagnosis as covariates of a logistic regression to investigate their contributions to the significance of racial disparities in breast cancer mortality. Principal component analysis (PCA) was utilized to reduce the multicollinearity among covariates and summarized the correlation structure displayed by the fourteen variables that were used to measure socio-demographic conditions and spatial accessibility to mammography facilities. The logistic regression analysis revealed that a census tract with significant racial disparities in late-stage diagnosis was 4 times more likely to have significant racial disparities in breast cancer mortality. Lower SES played an important role in determining whether a census tract displayed significant racial disparities in breast cancer mortality. However, proximity to mammography facilities had no impacts on the presence of significant racial disparities in breast cancer mortality for Hispanics, while centroids of census tracts that were closer to mammography facilities were more likely to have significant racial disparities for African-American women. For these women, most census tracts with significant racial disparities were located within the metropolitan areas which had higher concentration of health care facilities. In addition to the metropolitan areas, significant racial disparities for Hispanics were also found along the Southwest border of Texas, which lacked health care and had longer driving distance and time to mammography facilities.
This research analyzed the spatial patterns of racial disparities in late-stage diagnosis and breast cancer mortality, which shed new insights on the location of problematic areas and could help prioritizing the areas for effective intervention programs by accounting for population distributions. The identified risk factors in racialdisparities could help develop community-based intervention models and lead to a more efficient allocation of limited health resources with the ultimate goal of saving women‘s life. Subsidized health insurances and free mammograms for disadvantaged African-Americans and Hispanics could be applied at local communities to reduce racial disparities in breast cancer. The long term goal to improve the African-American and Hispanic women‘s health is to boost their income and enhance their social status through educational attainment.