Socioeconomic Status is a Cryptic Reference to Race
Clinical and Health Affairs
Race, Socioeconomic Status, and Premature Mortality
By Rhonda Jones-Webb, Dr.P.H., Xinhua Yu, Ph.D., M.D., Melanie Wall, Ph.D., Yue Cui, Ph.D., Wendy Hellerstedt, Ph.D., and John Oswald, Ph.D.
Health disparities between blacks and whites have been well-documented in the United States. For example, blacks in New York’s Harlem, central Detroit, the south side of Chicago, and Los Angeles’ Watts have about the same probability of dying by age 45 as whites throughout the entire country do by age 65.1
Socioeconomic status is an important variable in understanding differences in premature mortality rates between blacks and whites in this country. Socioeconomic status is defined by a range of factors that influence an individual’s social position (eg, education and income).2 Indeed, studies have shown that socioeconomic status is strongly correlated with race and can be an even stronger predictor of premature mortality than race.2-5 However, both race and socioeconomic status are important factors in understanding differences in premature mortality.
To shed light on the causes of premature mortality among blacks and whites, we designed a study using data from the Minnesota Department of Health and the U.S. Census. Our hypothesis was that the association between race and premature death would be affected by socioeconomic status.
Census tracts (n=578) served as our geographic units of analysis in multivariate models. A census tract is a geographic area that consists of about 4,000 persons on average.
All research was approved by the ethics committee at the Minnesota Department of Health and conforms to the principles of the Declaration of Helsinki.
Independent variables included race, sex, age, and socioeconomic status.
Data on race were obtained from the standard categories used by the U.S. Census Bureau in 1990 and limited to blacks and whites (non-Hispanic and Hispanic combined). Non-Hispanic whites and blacks comprised the overwhelming majority of each group.
Data on socioeconomic status were determined using information from the U.S. Census about the percentage of people living below poverty in a neighborhood and the average educational attainment of that neighborhood’s residents. These measures served as proxies for measures of individual socioeconomic status.
Data on each neighborhood’s economic and educational status were matched to each subject’s permanent address at the census tract level. Addresses were obtained from death certificates. Economic status was determined by the percentage of persons within a census tract living below the poverty level (an annual income of $12,700 for a family of 4 in 1990).6 Less affluent areas were defined as census tracts in which 20% or more of the population lived below the poverty line. Educational status was measured by the percentage of persons in a census tract with a high school education. Areas with poorly educated residents were those in which 15% or more of the population had less than a high school education. The correlation coefficient between neighborhood poverty and households with less than a high school education was .76 at the census tract level. Measures of neighborhood poverty and educational status were dichotomized given the small number of blacks in our sample.
Because we were interested in whether socioeconomic status moderated the association between race and premature mortality, we focused on testing the interactive effects of race with either neighborhood poverty or educational status on premature mortality. However, we first analyzed the separate effects of race, neighborhood poverty, and neighborhood educational status. We then considered all 2-way interactions. We also tracked differences between men and women.
Differences in mortality were greatest among blacks and whites who lived in census tracts where the population was more affluent and better educated (Table 1). For example, the ratio of black-to-white deaths in areas where less than 20% of the population lived below poverty was 6:1 compared with 1.5:1 in census tracts where more than 20% of the population lived below poverty.
We next compared data from Poisson regression models for men and women of both races. Significant interactions were observed between race and poverty on all causes of premature mortality (Table 2). Black men (OR=4.99 vs. 1.5) and black women (OR=6.18 vs. 1.57) who lived in less affluent areas had greater odds of dying than white men and women who lived in similar tracts. Racial differences in mortality were most striking, however, for men and women who lived in more affluent census tracts. Black women who lived in more affluent neighborhoods were 6 times more likely than their white counterparts to die prematurely.
Neighborhood educational status was independently related to mortality among men and women of both races. Men and women who lived in census tracts with lower educational attainment were at greater risk of dying than those who lived in tracts with higher levels of education, regardless of their race (data not shown).
♦ Strengths and Limitations
Second, the socioeconomic characteristics of some neighborhoods may have changed between the time of the census and the time of the individual’s death. We do not expect this to introduce significant bias because it generally takes a number of years for neighborhoods to change significantly.7
Third, statistical power to detect significant interaction effects was limited by the small number of deaths in certain subgroups (eg, the number of deaths among black men who lived in census tracts with higher educational status). Despite this limitation, we found a significant interaction effect between race and neighborhood poverty on premature mortality among both men and women. To increase power, we pooled data from 1992 through 1998 and dichotomized the neighborhood poverty and education variables. Future studies may wish to confirm the findings from this study using mortality data across multiple states that have similar demographic characteristics.
Our analyses also did not take into account the effects of living in a poor census tract that is adjacent to one like it versus one that is more affluent.8 Nor did our study include individual measures of socioeconomic status. We note, however, that other studies using area-based and individual-level socioeconomic data have yielded findings similar to ours.9
Finally, it may not be possible to generalize our findings to populations outside of Minnesota. The 5-county metro area we studied is highly segregated and has one of the highest percentages of minority children living in poverty in the United States.10 Future studies may wish to include both urban and rural samples of blacks and whites to confirm our findings.
♦ Race, Socioeconomic Status, and Mortality
Second, although blacks have made socioeconomic gains over the past few decades, they still experience racism and discrimination, which may induce psychological distress that can lead to higher rates of hypertension and other chronic health problems.11 This could account for the higher rates of premature mortality among blacks.12-14
Our findings also have implications for those trying to educate the public on how to live healthier lives. We found that racial disparities in terms of mortality risk were greatest among black women and white women who lived in more affluent census tracts. Media campaigns may be an important way to raise awareness among black women about improving their health. Messages that link health with the ability to support and care for a family are likely to resonate.16
Rhonda Jones-Webb is an associate professor in the University of Minnesota School of Public Health, Division of Epidemiology and Community Health. Xinhua Yu is a research associate in the University of Minnesota School of Public Health’s Division of Health Management and Policy. Melanie Wall is an associate professor of biostatistics. Yue Cui is with the University of Minnesota School of Public Health’s Division of Biostatistics. Wendy Hellerstedt is an associate professor with the University of Minnesota School of Public Health’s Division of Epidemiology and Community Health. John Oswald is senior director of health care analytics for OptumHealth at UnitedHealth Group.
This study was supported by a Center for Excellence in Health Statistics grant from the Centers for Disease Control and Prevention.