Racial residential segregation and geographic heterogeneity in black/white disparity in poor self-rated health in the US: a multilevel statistical analysis
Introduction
There is large variation in the degree of racial residential segregation experienced by African Americans across US metropolitan areas. According to the 2000 Census, among the 50 metropolitan areas with the largest black populations, residential segregation (measured by the index of dissimilarity) ranged from “very high” in Detroit, Milwaukee, New York, Chicago and Newark, to “moderate” in Greenville, Riverside, Norfolk, Raleigh and Augusta (Lewis Mumford Center For Comparative Urban and Regional Research, 2001a). Residential segregation has been postulated to be associated with racial disparities in individual well being and neighborhood environment (Massey, 2001; Massey & Fischer, 2000; Massey & Hajnal, 1995). Studies have documented substantial variation in racial disparities in individual level outcomes (e.g. employment and education) (Cutler & Glaeser, 1997), and in neighborhood environment (e.g. exposure to poverty) (Galster & Killen, 1995; Galster & Mikelsons, 1995; Logan, 2002; Massey, Condran, & Denton, 1987) across metropolitan areas. Residential segregation has also been hypothesized to negatively affect the health of African Americans through its detrimental effects on individual level (e.g. employment, education) and neighborhood level (e.g. concentrated poverty) factors (Acevedo-Garcia & Lochner, 2003; Acevedo-Garcia, Lochner, Osypuk, & Subramanian, 2003a; Collins & Williams, 1999; Ellen, 2000; Williams, 1997; Williams, 2001).
Despite the variation in the degree of residential segregation across metropolitan areas, social epidemiologic studies have not explicitly examined the degree of geographic variation in black/white health disparities, nor whether such variation is systematically related to the geographic variation in the level of black/white residential segregation. The only previous study of African American health that examined the effect of residential segregation across the US after controlling for individual level factors used Huber standard errors to account for clustering at the metropolitan area level (Ellen, 2000), but did not model geographic variation explicitly.
A systematic examination of contextual heterogeneity (Subramanian, Jones, & Duncan, 2003), defined as the metropolitan area level variation in racial/ethnic disparities in health after adjusting for individual level factors, is central in understanding the potential role of racial residential segregation. Patterns of population health are shaped by a complex constellation of individual as well as contextual factors that may conceivably vary for whites and blacks. An investigation of racial disparities in health at the metropolitan area level can provide valuable insights into the manner in which contextual factors may relate to individual health. For instance, if we find empirical support for the hypothesis that the metropolitan area variation in health is greater for blacks (as compared to whites), this would suggest that the effect of contextual factors at the metropolitan area level on individual health may be different for whites than for blacks.
The idea of metropolitan area heterogeneity is especially pertinent to evaluating the multilevel association between racial/ethnic residential segregation and individual health. First, in the US, socioeconomic disparities between blacks and whites have a clear spatial expression at the metropolitan area level both along the central-city/suburban divide, and across neighborhoods (Altshuler, Morrill, Wolman, Mitchell, & Committee on Improving the Future of US Cities Through Improved Metropolitan Area Governance, 1999). Second, perhaps more importantly, some features of the organization of US metropolitan areas have shaped racial/ethnic disparities. For example, fragmentation in metropolitan governance contributes to residential segregation and unequal access to opportunities. Local control over land use, and taxes/public services creates segmentation of municipalities by property values (i.e. housing prices), which in turn results in segregation by income and race/ethnicity, and unequal spatial access to public goods such as education (Altshuler et al., 1999; Pendall, 2000; Schill & Wachter, 1995). Third, socioeconomic and neighborhood quality indicators across metropolitan areas are not only significantly worse for blacks than for whites, but they exhibit considerably more variation for blacks than for whites across metropolitan areas (Acevedo-Garcia, Osypuk, & Krimgold, 2003b). It has been hypothesized that high levels of racial/ethnic residential segregation are a plausible cause of the sharp disparities in neighborhood environment (Altshuler et al., 1999; Massey, 2001; Massey et al., 1987; Massey & Denton, 1993; Williams & Collins, 2001). Therefore, it is reasonable to postulate that black health may vary more across metropolitan areas than white health, and that residential segregation may be more detrimental to black health than to white health.
Even though the hypothesized relationships between residential segregation and health are multilevel in nature, previous empirical research on residential segregation and health has largely been based upon single-level aggregate analyses (Acevedo-Garcia & Lochner, 2003; Acevedo-Garcia et al., 2003a; Ellen, 2000). For instance, several studies showed that, at the metropolitan area level, segregation (usually measured by the dissimilarity index) is positively associated with infant and adult mortality rates among African Americans (LaVeist (1989), LaVeist (1993); Polednak (1993), Polednak (1996a), Polednak (1996b), Polednak (1997)). Only one previous national study showed that after controlling for several individual level factors, segregation (measured by the centralization index) was positively associated with higher odds of having a low birth weight baby among black women (Ellen, 2000). However, as pointed out earlier, even this study is not multilevel in terms of its analytical strategy, although the sample data itself was multilevel.
While aggregate studies have provided a compelling indication that segregation may contribute to racial disparities in health, only multilevel analysis may allow us to determine the independent effects of area residential segregation on individual health, i.e., whether segregation operates through individual and/or neighborhood characteristics, as well as whether metropolitan-area segregation has an effect on health after controlling for both individual and neighborhood level factors. Aggregate studies cannot separate out the individual factors that contribute to racial health disparities from the contextual (i.e. neighborhood and metropolitan) ones. With multilevel analysis we may also examine whose health is negatively impacted by segregation. Segregation may have health consequences for members of racial minority groups but not for members of racial majority groups.
Meanwhile, from a methodological perspective, to date, most applications of multilevel methods in public health research principally has estimated the “average” effect of a predictor measured at an area level on individual-level health outcomes (Subramanian et al., 2003). This approach can potentially obscure contextual heterogeneity among metropolitan areas, which as discussed above, is of substantive interest when examining racial/ethnic disparities in health and their association with residential segregation. Specifically, we raise and answer three main research questions:
- (1)
Are there significant differences across metropolitan areas in self-rated health among blacks and whites after accounting for individual level factors?
- (2)
Are there significant differences across metropolitan areas in the white/black disparity in self-rated health?
- (3)
Does residential segregation account for the geographic variation in racial disparities in self-rated health across metropolitan areas?
Section snippets
Data
We used the 2000 March Supplement (Annual Demographic Survey) of the Current Population Survey (CPS) conducted by the US Census Bureau for the US Bureau of Labor Statistics to obtain data pertaining to the individual level response (i.e. self-rated health) and predictor variables (Unicon Research Corporation, 2003). The final sample from the CPS data consisted of 51,316 non-Hispanic white and non-Hispanic black adults aged 18+residing within those US metropolitan areas with a total population
Results
Table 1 presents the results from Models 1 and 2. The reference group is a 45-year old, married/partnered, white male with graduate-level education and earnings above $125,000. Model 1 reports the independent effect of individual race on poor self-rated health, after accounting for age, sex, marital status, education and income. Compared to whites, blacks are 1.5 times more likely (95% CI: 1.40–1.62) to report poor health. Before adjusting for sex, marital status, education and income, the OR
Discussion
Using a multilevel methodological approach, we examined the geographic variation in racial disparities in self-rated health across metropolitan areas, and its association with the geographic variation in racial residential segregation, after accounting for individual factors. We turn to discussing the key findings of our analysis.
On average, blacks are 1.5 times more likely to report poor health than whites, even after we take into account several key individual demographic and socioeconomic
Conclusion
Blacks living in metropolitan areas characterized by high black isolation have a higher likelihood of reporting poor self-rated health than blacks living in low black isolation areas, after controlling for individual level factors. Although not strictly comparable (due to differences in the outcomes studied, and the segregation dimensions tested), this finding offers some support for those previous aggregate studies that found that segregation was positively associated with mortality rates and
References (65)
- et al.
Elderly people's ratings of the importance of health-related factors to their self-assessments of health
Social Science & Medicine
(2003) - et al.
Metropolitan area income inequality and self-rated health—a multi-level study
Social Science & Medicine
(2002) - et al.
The relationship of income inequality to mortalitydoes the choice of indicator matter?
Social Science & Medicine
(1997) - et al.
Self-rated health status as a health measurethe predictive value of self-reported health status on the use of physician services and on mortality in the working-age population
Journal of Clinical Epidemiology
(1997) - et al.
Geographies of health perception in Quebeca multilevel perspective
Social Science & Medicine
(1999) - et al.
Does the state you live in make a difference? Multilevel analysis of self-rated health in the US
Social Science & Medicine
(2001) Race and healthbasic questions, emerging directions
Annals of Epidemiology
(1997)Zip code level risk factors for tuberculosisneighborhood environment and residential segregation, New Jersey, 1985–1992
American Journal of Public Health
(2001)- et al.
Residential segregation and health
- et al.
Future directions in residential segregation and health researcha multilevel approach
American Journal of Public Health
(2003)
Modelling binary data
Segregation and mortalitythe deadly effects of racism
Sociological Forum
Being black, living in the redrace, wealth and social policy in America
The starting gatebirth weight and life chances
Relationship between premature mortality and socioeconomic factors in black and white populations of US metropolitan areas
Public Health Reports
Are ghettos good or bad?
Quarterly Journal of Economics
Measuring health perceptions in the health insurance experiment
Are African Americans still hypersegregated?
A methodological analysis of segregation indices
American Sociological Review
Is segregation bad for your health? the case of low-birth weight
Understanding AIDShistorical interpretations and the limits of biomedical individualism
American Journal of Public Health
Validity of self-rated health among Latino(a)s
American Journal of Epidemiology
The geography of metropolitan opportunitya reconnaissance and conceptual framework
Housing Policy Debate
The geography of metropolitan opportunitya case study of neighborhood conditions confronting youth in Washington, DC
Housing Policy Debate
Multilevel statistical models
Improved approximations for multilevel models with binary responses
Journal of the Royal Statistical Society A
Self-rated health and mortalitya review of twenty-seven community studies
Journal of Health and Social Behavior
Health status measurementa brief but critical introduction
Neighborhoods and health
Income distribution, socioeconomic status, and self rated health in the United Statesmultilevel analysis. (see comment)
BMJ
Cited by (209)
Empirical evidence supporting the inclusion of multi-axes segregation in assessing US county health
2023, Social Science and MedicineNeonatal motor functional connectivity and motor outcomes at age two years in very preterm children with and without high-grade brain injury
2022, NeuroImage: ClinicalCitation Excerpt :Factors included age ≤18 years at time of delivery, Black race, no high school degree, public health insurance, and single-parent household (Hack et al., 1994; Mangin et al., 2017; Manley et al., 2015). Black race was included in the demographic stressor index to account for social and health inequities that result from experiencing structural and individual racism in America, particularly as the region of recruitment is heavily stratified by race (Bureau, 2010; Subramanian et al., 2005). UPT and BI neonates underwent MRI scanning at/near term-equivalent age (35–45 weeks postmenstrual age, mean 38.4 weeks).
Cross-sectional comparison of health care delivery and reimbursement between segregated and nonsegregated communities in Hungary
2024, Frontiers in Public HealthInequalities in noise will affect urban wildlife
2024, Nature Ecology and Evolution