Re: Race, socioeconomic status, and breast cancer treatment and survival.
N. O. Kwate
DOI: https://doi.org/10.1093/JNCI/94.16.1254
2002-08-21
Journal of the National Cancer Institute
Abstract:A recent report in the Journal (1) attempted to “disentangle the influence of race and socioeconomic status on breast cancer stage, treatment, and survival.” The complex constellation of racial and socioeconomic factors in chronic illness is not easily deconstructed, and I would argue that this study has not fully realized this goal. The problem is that, in the Bradley et al. study, race and socioeconomic status (SES) were inextricably linked, much as they are in real life. That is, only 13% of the white women lived in census tracts with 13% or more in poverty, whereas 84% of African-American women lived in census tracts with 13% or more in poverty. Thus, race and SES were essentially the same variable: most poor women were African-American, and most women who were not poor were white. Therefore, an analysis with race as an independent variable and SES as a covariate is tantamount to looking at the effect of race while controlling for race. What does it mean to attribute certain outcomes, as Bradley et al. have done, to residing in census tracts with 13% or higher poverty, when almost 90% of those residents are AfricanAmerican? A few critical issues should be considered when defining income and SES. First, using federal guidelines for poverty level may obscure how SES affects health, given that federal guidelines place individuals at the most marginal levels of existence. That is, the income levels that are used to connote poverty are too strict. Many families who subsist on minimal levels of income may still not be classified as living in poverty. For example, the poverty level for a family of 2 in 2001 is $11 610 (2). Thus, a woman caring for a child at an income level of $12 000 will not be classified as living in poverty, despite the harsh circumstances of her financial situation. Second, the authors recognize that census tracts are also not ideal units for measuring poverty. Indeed, in 1990, in Brooklyn, New York, two particular census tracts had essentially identical per capita incomes. However, in one, whites earned only $3023 more than African-Americans did; in the other, they earned $11 257 more—and these census tracts are separated by only two house numbers (3). Third, defining SES for African-Americans can be a tricky enterprise. African-Americans have lower total assets, are more likely to live in segregated neighborhoods and to be the first generation with wealth in the family (4), and earn less than whites with the same education. And, in many cities, wealthy and poor African-American families often live together in the same neighborhood, if not on the same street. These conditions are largely nonexistent in predominantly white neighborhoods. Thus, using typical factors of income and education may not be appropriate when comparing the two groups. In the final analysis, we must question whether separating race and SES may, in many ways, be an artificial distinction. Although we may be able to control for SES as a discrete statistical variable, can we do it as a real life experience? That is, in the United States, centuries of structured inequalities based on race have created conditions in which African-Americans face a very different social reality (including economic status) than European-Americans. Then, how are we to understand the ramifications on health of “being poor” when “being poor” disproportionately means “being black”?