Socioekonomska stratifikacija i urbano stanovanje u postsocijalističkom Beogradu

Teaching and Learning Guide for: Racial Residential Segregation in Urban America

The extent to which residential segregation results from differences in the socioeconomic status of racial/ethnic groups remains a topic of debate. Although recent studies have argued that improvements in minority groups' socioeconomic status will lead to reduced residential segregation, such analyses have been cross-sectional. This study uses longitudinal data from 27 central cities in Texas to examine the effects of 1970,1980, and 1970–80 changes in socioeconomic status on Black-White, Anglo-Spanish, and Black-Spanish segregation. The findings indicate that socioeconomic factors are not significant determinants of racial/ethnic segregation in these periods. Rather, age of city, population growth, and percent of the population of minority status appear to explain patterns of residential segregation.

In the post-socialist period, there were many changes in the post-Yugoslav region. The dissolution of the state and the civil war were just some of the most important events. The successor states went through the post-socialist transition at a different pace. The political and economic transition had a significant impact on the social status of the population. According to numerous international urban studies, there is a causal relationship between the social status of the population and the housing situation. Looking at the social status of the population in Belgrade, it is obvious that the social status decreases from the centre to the periphery. The housing situation in Belgrade is very heterogeneous. First, there is a historical centre filled with buildings from different eras: from the pre-World War II period, from the socialist period and from the post-socialist period. Secondly, there are socialist neighbourhoods that surround the historic core of the city. Finally, there are the neighbourhoods on the periphery, which are mostly informally built. The withdrawal of the state and the reintroduction of market mechanisms in the housing sector, drastic proportions of owner-occupied housing, the expansion of illegal construction activity, international isolation during the 1990s and the subsequent reconnection to the global economy, albeit with extremely weakened institutions, characterised post-socialist housing and urban development in Belgrade. Added to this was another wave of demographic pressure from the former Yugoslav republics and Kosovo. Through the lens of social status and housing situation, we observe the patterns of residential segregation in the post-socialist period. Due to the incomparability of statistical data from different periods, we use a combination of sources and research approaches. The first two decades of the post-socialist period were analysed at the level of census units, using data on the education of the population as a proxy for their social status.

MariaKrysan and KyleCrowderCycle of Segregation: Social Processes and Residential StratificationNew York: Russell Sage Foundation, 2017. 288 p. $35.00

Using individual-level data from the neighborhood files of the 1970 Public Use Sample, the components of individual socioeconomic status (SES) education, occupation, and income-are treated as resources; and racial differences in the return on these resources, differences in neighborhood residential quality, are estimated, decomposed, and subjected to component analysis. It is documented that blacks receive a lesser return than whites; they reside in inferior neighborhoods despite similar resources. The notion is tested and supported that one mechanism explaining the lesser return is that blacks are channeled into predominantly black neighborhoods and are thus less able than whites to achieve class segregation. Channeling has a greater impact on economic factors than on social factors; on the latter, intraracial segregation seems more possible. As a final note, given the theoretical importance of community and neighborhood contexts, and thegeneral assumption in stratification theory that individual status (or parental status) is an adequate proxy for such context, a reexamination of studies of racial differences previously presumed to be net of class is suggested. The persistence of residential segregation by race has been demonstrated on many occasions (Erbe; Kantrowitz), and it has been shown to persist independent of occupational segregation and other forms of economic segregation (Duncan and Duncan; Kantrowitz; Simkus; Taeuber; Taeuber and Taeuber). Most recently, Simkus has demonstrated that up to one-fourth of occupational residential segregation can be attributed to racial segregation in both the occupational and residential distributions, and he redemonstrated the point that within-occupation residential racial segregation is extremely high. Using detailed tract level data from one SMSA, Erbe found that black professionals and managers, black college graduates, and blacks with incomes in excess of $25,000 lived in tracts comparable, respectively, to those of whites who were unskilled workers, school dropouts, and earned less than $3,000. Others have examined the effects of residential socioeconomic segregation, both structural effects (e.g., Morgan; Smith)

BackgroundBlack adults experience the highest hypertension burden of any racial group. It is unclear whether different dimensions of residential segregation have different relationships with hypertension risk for Black and White people, or whether socioeconomic status explains any of the relationships.MethodsWe studied 6787 Black and White participants from the REGARDS (Reasons for Geographic And Racial Differences in Stroke) study, without prevalent hypertension and with follow‐up for hypertension 9.4 years later. County‐level segregation was defined using dissimilarity, isolation, and interaction indices. Associations between residential segregation indices and incident hypertension were estimated. Racial differences in associations and mediating effects of socioeconomic factors were assessed.ResultsRisk ratios (RRs) of incident hypertension were 1.06 per SD higher isolation index (95% CI, 1.03–1.09) and 0.96 per SD higher interaction index (95% CI, 0.93–0.99). The dissimilarity index was not associated with hypertension risk. Neighborhood socioeconomic status explained 19% of the association between the isolation index and hypertension incidence. Although there were no statistically significant racial differences in associations, the isolation index was marginally associated with higher risk of hypertension in Black people only (RR, 1.05 [95% CI, 0.98–1.12]), and neighborhood socioeconomic status was a marginally significant mediator of that association (RR indirect effects, 0.993 [95% CI, 0.98–1.0004]).ConclusionsFindings indicated that living in racially segregated counties was associated with higher risk of hypertension, particularly for Black people. Higher neighborhood socioeconomic status may mitigate the negative effects of that aspect of residential segregation on hypertension development.

Southeastern Geographer Vol. XXXVIII, No. 2, November 1998, pp. 125-141 RESIDENTIAL SEGREGATION OF ASIAN AMERICANS IN THE ATLANTA METROPOLITAN AREA, 1990 Qian Zhang The residential segregation patterns of Asian Americans in U.S. metropolitan areas have changed over time and have also varied in different urban settings. Atlanta, unlike major immi­ gration centers, has virtually no long-established ethnic enclaves and no traditional immigra­ tion networks. It thus provides a different urban setting for the study of Asian-American residential segregation patterns. The validity of previous conclusions about Asian-American segregation patterns, which are based on studies of traditional immigration centers in earlier time periods, needs to be tested in different urban settings such as Atlanta. In the Atlanta MSA in 1990, the most recently arrived Asian-American group, the Vietnamese, display the highest levels of segregation from Whites, Blacks, Hispanics, and other Asian-American groups. The well-established Asian group, the Chinese, are the most integrated with Whites. All AsianAmerican groups are highly segregated from Blacks in the Atlanta MSA; they are also, to a con­ siderable extent, residentially segregated from each other. Their segregation levels from His­ panics are relatively moderate. INTRODUCTION. The 1980 census provided a full set of cross-tabulations at the census-tract level for persons who identified themselves as Asians or Pacific Is­ landers. These data formed the basis for a number of studies of Asian-American residential segregation (Denton and Massey, 1988; Lanberg and Farley, 1985; Massey and Denton, 1987). It was in the 1970s that Asian Americans, as a group, emerged as an important object of the study of urban residential patterns, and some common attributes of Asian Americans, in terms of their residential pat­ terns, were found. Asian Americans display low to moderate segregation from Whites, and the levels of Asian-White segregation have fallen steadily over time (Lanberg and Farley, 1985; Massey and Denton, 1987). However, Asians, like Whites, displayed a high degree of residential segregation from Blacks, which cannot be simply explained by the two groups’ difference in socioeconomic status (Lanberg and Farley, 1985; Massey and Denton, 1987; Denton and Massey, 1988; Massey and Fong, 1990; Fong, 1996). The better established Asian ethnic groups, such as the Chinese and the Japanese, had lower segregation levels com­ pared to more recent immigrants such as the Vietnamese (Denton and Massey, 1988; Massey and Denton, 1992). Asian Americans have also been successful in obtaining suburban residences and are becoming more spatially integrated into the suburbs (Alba and Logan, 1991). In their path from residential segregation to spatial assimilation, and then to economic and cultural assimilation, Asian Mr. Zhang is a doctoral student in the Department o f Sociology at Yale University, New Haven, CT 06520. 126 S o u t h e a s t e r n G e o g r a p h e r Americans have largely conformed to the basic tenets of spatial assimilation the­ ory, 1they have confronted fewer barriers, and they have displayed lower levels of segregation compared to many other ethnic and racial groups (Massey and Denton, 1992). The intergroup difference within the broad category “Asian American,” nev­ ertheless, is notable in two aspects. First, their segregation patterns from AngloWhites , Blacks, and Hispanics are different (Lanberg and Farley, 1985). Each group’s specific history of immigration to the United States has had its impact on residential patterns. Usually, the well-established groups show lower levels of segregation from Whites, Blacks, and Hispanics, while recently arrived groups are more segregated. Second, the residential segregation among different AsianAmerican groups is remarkable (White et al., 1991; Zhou and Logan, 1991). For instance, the three East Asian (Chinese, Korean, and Japanese) groups’ segrega­ tion levels from each other are lower than their segregation levels from other Asian groups (Massey and Denton, 1992). Moreover, there is also notable geo­ graphic variation of Asian-American segregation patterns within the United States, as those in metropolitan areas of the Northeast and Midwest differ from those in the West and the South (Alba and Logan, 1991). PURPOSE. The purpose of this study is to identify the Asian-American residen­ tial segregation patterns for 1990 in the Atlanta metropolitan area. This study...

Heart failure (HF) is a cardiovascular disease (CVD) outcome that is associated with high morbidity and mortality as well as high healthcare costs.1 Given that HF is the end stage of most CVDs, both conditions share common risk factors such as type 2 diabetes (T2D), hypertension, smoking and obesity.2 Socioeconomic status (SES) has been recognized to have a measurable and significant effect on cardiovascular health. It has been reported that low SES may confer a cardiovascular risk that is equivalent to conventional risk factors.3 Low SES has been shown to be a powerful and independent predictor of HF development and adverse outcomes.4 Biological, behavioural and psychosocial risk factors prevalent in socioeconomically deprived individuals are known to accentuate the relationship between low SES and cardiovascular outcomes such as HF.3 These include lower levels of education, unhealthy lifestyles such as excessive alcohol consumption, limited access to health care and higher prevalence of comorbid conditions. The beneficial effects of regular physical activity (PA) and exercise in preventing vascular disease and promoting overall health are well established and documented. These benefits also extend to HF prevention.5 Though cardiorespiratory fitness (CRF) reflects habitual aerobic PA, it is a separate measure that captures the capacity of the cardiovascular and respiratory systems to supply oxygen to skeletal muscles during progressive PA or incremental exercise to volitional fatigue.6 The gold standard for CRF assessment is direct measurement of the highest attained oxygen consumption (VO2) during cardiopulmonary exercise testing. Similar to PA, high levels of CRF are strongly and independently associated with lower risk of vascular outcomes including HF.7, 8 The inverse associations between CRF and vascular outcomes have been reported to be stronger than that of traditional risk factors such as T2D and smoking; this has led to CRF being proposed as a vital sign.9 There is increasing evidence showing that higher levels of CRF can attenuate the adverse impact of other risk factors; for instance, we and others have previously shown that high CRF levels can attenuate the impact of risk factors associated with mortality,10 pneumonia11 and COVID-19 hospitalization.12 Given the evidence, we hypothesized that high CRF levels would attenuate the increased risk of HF due to low SES. To explore this, we aimed to evaluate the joint effects of SES and CRF on the risk of incident HF using a population-based prospective cohort of 1831 middle-aged Finnish men without a history of HF at baseline. We also evaluated the separate associations of SES and CRF with the risk of HF to confirm previous evidence of these associations. Reporting of the study conforms to broad EQUATOR guidelines13 and was conducted according to STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) guidelines for reporting observational studies in epidemiology (Appendix S1). The current analysis is based on the Kuopio Ischaemic Heart Disease (KIHD) risk factor study, a general population-based prospective cohort study comprising of a representative sample of men aged 42–61 years recruited in eastern Finland. A detailed description of the study design, recruitment methods, risk marker assessment and physical examinations have been described previously.8 Baseline measurements were performed between 01 March 1984 and 31 December 1989. The research protocol was approved by the Research Ethics Committee of the University of Eastern Finland and written informed consent was obtained from all the participants. A self-reported questionnaire was used to assess SES, which involved a summary index that combined factors such as income, education, occupational prestige, material standard of living and housing conditions. The composite SES index ranged from 0 to 25, with higher values indicating lower SES. Maximal oxygen uptake (VO2max) was used as a measure of CRF, which was assessed using a respiratory gas exchange analyser (Medical Graphics, MCG, St. Paul, Minnesota) during cycle ergometer exercise testing.14 We excluded men with a prevalent history of HF for the current analysis. We included all HF events that occurred from study entry through to 2018. The diagnostic classification of HF cases was coded according to the ICD-10 codes. Hazard ratios (HRs) with 95% confidence intervals (CIs) for HF were calculated using Cox proportional hazard models and these were adjusted for in three models: (Model 1) age; (Model 2) Model 1 plus systolic blood pressure (SBP), body mass index (BMI), heart rate, smoking status, history of T2D, history of coronary heart disease (CHD), total cholesterol, high-density lipoprotein cholesterol (HDL-C) and PA; and (Model 3) Model 2 plus mutual adjustment for each exposure. For consistency with previous reports,10, 15 the exposures (SES and CRF) were categorized into low and high levels based on their median cutoffs. The exposures were also modelled as continuous variables given evidence of linear relationships with HF risk using multivariable restricted cubic spline curves. Evaluation of the joint association of SES and CRF with HF risk was based on the following four combinations: high SES-low CRF; low SES- low CRF; high SES-high CRF and low SES-high CRF. Tests of interaction were used to formally assess if the risk of HF due to one exposure is modified by the other exposure and vice versa. To put our findings into clinical context, we also calculated the number needed to treat (NNT) associated with high SES-high CRF using the formula proposed by Altman and Anderson16: NNT (t) =1/[SB(t))HR – SB(t)], where SB(t) denotes the Kaplan–Meir survival probability in the reference group (High SES-Low CRF) at time t and HR refers to the Cox regression estimate comparing the exposure group with the reference group. Stata version MP 16 (Stata Corp, College Station) was employed for all analyses. The overall mean (standard deviation, SD) age, SES and CRF of study participants at baseline was 52 (5) years, 8.26 (4.24) and 30.8 (7.9) ml/kg/min, respectively (Table 1). There were significant differences in baseline characteristics between low and high CRF groups. Overall Mean (SD) or median (IQR) or n (%) High CRF Mean (SD) or median (IQR) or n (%) Low CRF Mean (SD) or median (IQR) or n (%) During a median (interquartile range) follow-up of 27.3 (18.6–31.2) years, 364 incident HF cases occurred. In an analysis adjusted for age, SBP, BMI, heart rate, smoking status, history of T2D, history of CHD, total cholesterol, HDL-C and PA, low compared with high SES was associated with an increased risk of HF 1.43 (95% CI: 1.15–1.79), which remained similar on further adjustment for CRF. On adjustment for the confounders as above, high CRF was associated with a decreased risk of HF compared with low CRF 0.70 (95% CI: 0.55–0.89), which remained similar on additional adjustment for SES. There was evidence of significant associations when both exposures were modelled as continuous variables (Table 2). Restricted cubic spline curves with adjustment for age, SBP, BMI, heart rate, smoking status, history of T2D, history of CHD, total cholesterol, HDL-C and PA showed that HF risk increased continuously with decreasing SES across the range 7–19 (p-value for nonlinearity =.83) (Figure 1A), whereas HF risk decreased continuously with increasing CRF across the range 18–58 ml/kg/min (p-value for nonlinearity =.79) (Figure 1B). The spline curves were qualitatively similar in subgroups of CRF and SES (Figure 2). In multivariable analysis, low SES-low CRF was associated with an increased HF risk 1.32 (95% CI: 1.01–1.74), high SES-high CRF with a decreased HF risk 0.62 (95% CI: 0.43–0.89), with no evidence of an association for low SES-high CRF and HF risk 1.01 (95% CI: 0.73–1.39) when compared with men with high SES-low CRF (Table 2). The association of SES with HF risk was not modified by CRF (p-value for interactions >.10) and neither was the association between CRF and HF risk modified by SES (p-value for interactions >.10), when both exposures were modelled as continuous or categorical variables (Figure 3). The absolute risk reduction of HF associated with high SES-high CRF was 0.21 during the entire duration of follow-up, which translated into a NNT of 10 (95% CI: 6–35) to prevent one HF. Our results based on a general population-based prospective cohort study of middle-aged to older Finnish men confirms the previously reported independent associations of low SES with increased HF risk and high CRF levels with lowered risk of HF. The associations were also potentially consistent with graded dose-response relationships. Evaluation of the joint associations of SES and CRF with HF risk showed that increased CRF levels appeared to attenuate the increased risk of HF associated with low SES. However, formal tests showed no significant evidence of interactive effects of SES and CRF on the long-term risk of HF, suggesting the effect of each exposure on HF risk may be independent of the other. Given the low sample size and event rates in the exposure categories, studies with larger samples are needed to confirm or refute potential interactive effects of SES and CRF on HF risk. Finally, our findings suggest that the NNT for high aerobic fitness levels and high SES to prevent a HF event over long-term follow-up ranged from 6 to 35 in approximately healthy middle-aged to older men. The interaction between SES and HF has been reported to be complex and the precise mechanisms accounting for the association between low SES and increased HF risk remain elusive.4 Socioeconomic differences in potential aetiological risk factors such as alcohol consumption, hypertension and systemic inflammation, have been reported to contribute to the risk. Social deprivation is also associated with lower rates of treatment, dose and adherence to therapy for, and delayed presentation of hypertension, diabetes and CHD,4 which consequently lead to HF. Psychosocial factors such as stress and depression, which are strongly associated with cardiovascular outcomes, also disproportionately affect individuals of low SES.3 Though CRF is determined by many non-modifiable factors such as age, sex and heritability, it remains a modifiable risk factor. The most established methods of increasing CRF are via exercise training and increased PA.9 Greater PA and exercise reduce HF risk through various mechanisms including (i) reducing the prevalence of standard and novel cardiovascular risk factors such as hypertension, obesity, blood glucose and coronary artery disease; (ii) preventing adverse changes in cardiac structure and function; (iii) promoting physiologic remodelling and (iv) improving cardiac, neurohormonal, skeletal muscle, pulmonary, renal and vascular performance.5 These findings may have important clinical implications. They add to the overwhelming evidence on the benefits of high CRF levels (via regular aerobic PA) on chronic diseases and their potential ability to attenuate the adverse effects of traditional risk factors. Despite guideline recommendations and population-wide strategies to promote PA levels, most populations do not achieve general PA recommendations. Populations at high cardiovascular risk including the socioeconomically deprived need more education on the substantial benefits of PA. Furthermore, there should be widened access to PA resources that are both feasible and attractive for these populations. This is the first evaluation of the separate and joint associations of SES and CRF with HF risk. We also assessed the nature of the dose-response relationships of the exposures with HF risk. Other strengths of this analysis included the use of a prospective cohort design with exclusion of men with pre-existing HF, the long-term follow-up duration of the cohort and the use of a gold standard measure of CRF. Limitations deserving consideration included the relatively low sample size due to the categorization of exposures, use of self-administered questionnaires in assessing SES, findings may only be generalizable to middle-aged and older northern European men and potential for biases such as residual confounding and regression dilution bias. In a general male Finnish population, both SES and CRF were each independently associated with HF risk, potentially consistent with graded dose-response relationships. High levels of CRF may attenuate the increased risk of HF due to low SES, but further study is needed to confirm if there are true interactive effects of SES and CRF on the long-term risk of HF. The authors thank the staff of the Kuopio Research Institute of Exercise Medicine and the Research Institute of Public Health and University of Eastern Finland, Kuopio, Finland for the data collection in the study. J.A.L. acknowledges support from The Finnish Foundation for Cardiovascular Research, Helsinki, Finland. These sources had no role in design and conduct of the study; collection, management, analysis and interpretation of the data; and preparation, review or approval of the manuscript. No potential conflict of interest was reported by the authors. S.K.K.: Study design, data analysis and interpretation, drafting manuscript, and revising manuscript content and approving final version of manuscript; S.Y.J.: Study design and revising manuscript content and approving final version of manuscript; T.H.M: Study design and revising manuscript content and approving final version of manuscript; J.A.L.: Study design and conduct, responsibility for the patients and data collection, and revising manuscript content and approving final version of manuscript. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Discussing school socioeconomic segregation in territorial terms: the differentiated influence of urban fragmentation and daily mobility

Residential segregation could be regarded as a process whereby two or more distinct communities who formerly lived together separate from one another due to many factors. Residential segregation is not only applicable to small communities but rather to a larger region. As a result of the ongoing civil unrest that engulfed the city of Jos, there has been a process of residential mobility and relocation among people of different faith. The paper is aimed at examining the implication of intangible location attributes on residential mobility, segregation and relocation in Jos town. Stratified random sampling technique was employed in order to come up with the sample needed to conduct the research. The data needed for the research were retrieved through structured, semi-structured and unstructured interview method of data collection. A qualitative method and approach of data analysis through the use of a thematic network analysis was incorporated in order to analyse the data gotten from the interview survey. The results uncovered that residential segregation in the study area leads to change in the residential pattern of Jos town. The variations and trends in the sales and rental value of residential properties were greatly affected as a result of the persisting residential segregation. The research concludes that residential segregation has a great implication on land and landed property value as variations in the values of residential properties is noticeable. There is a need for those in authority to take a decisive action in order to overcome and halt the persisting mobility and relocation in the study area.

The COVID-19 pandemic has been particularly devastating for those with limited economic resources. Extensive research demonstrates the negative relationship between wealth and mortality at both the individual and area levels. In addition, residential segregation has been linked to poor health and greater mortality. Home equity is the largest asset that many Americans own, but residential segregation devalues homes located in Black neighborhoods. Despite the interlocking relationships between wealth, residential segregation, and mortality, it remains unclear how wealth and residential segregation work to predict COVID-19 deaths. Using U.S. Census data and county-level COVID-19 data from Johns Hopkins University (n = 1164), I deploy median home value as a wealth proxy and negative binomial regression models to interrogate two questions. (1) What is the relationship between home value and COVID-19 deaths? (2) How does the relationship vary by level of residential segregation? Results indicate that COVID-19 mortality is 64 percent greater in the lowest wealth counties than in the wealthiest counties. At average median home value, the most segregated counties with the largest Black populations suffer 28 percent more COVID-19 deaths than similarly situated counties with low levels of residential segregation and small Black populations. This study underscores the importance of accounting for residential segregation in examinations of the well-established relationship between socioeconomic status and health and mortality.

Invited Perspective: Moving from Characterizing to Addressing Racial/Ethnic Disparities in Air Pollution Exposure.

According to ecological theory, the socioeconomic status of a minority group is inversely related to the group’s level of residential segregation from the majority group. This article determines whether the level of black socioeconomic status is related to the level of black residential segregation in the city of Detroit and Detroit’s suburbs. Data were obtained from the U.S. Bureau of Census, 1990 Summary Tape Files 4-A. The methods employed to measure residential segregation were the indexes of dissimilarity D and isolation P*. Indexes were computed by census tract to measure segregation and isolation between blacks and whites at the same level of occupation, income, or education. The results revealed that residential segregation between blacks and whites remained high (i.e., above 50%) in both the city and the suburbs despite comparable socioeconomic status. Blacks in the suburbs were more segregated and isolated than blacks in the city at each socioeconomic level.

Intro Racial separation of care contributes to racial disparities in health. Little is known about upstream drivers of receiving care in a minority-serving hospital (MSH) and differences by race/ethnicity. Methods Cross-sectional analysis of MCBS data (2010-2019) with binary outcome of hospitalization in MSH versus not. Exposures of interest were individual race/ethnicity and insurance; neighborhood disadvantage (SDI); and regional residential racial segregation. We included individual and regional covariates. In fully adjusted models, we tested whether race/ethnicity modified the association between SDI or residential segregation and hospitalization in MSH. Results Among 8,735 hospitalizations, race/ethnicity (aOR 3.19 for Black vs. White patients; 95% Confidence Interval (CI): 2.60-3.91; P < 0.001); dual-eligible status (aOR: 1.30, 95% CI: 1.05-1.60; P < 0.05), neighborhood disadvantage (aOR for most disadvantaged versus least: 2.03, 95% CI: 1.56-2.63, P < 0.001); and residential segregation (aOR for high segregation versus low: 2.99, 95% CI: 1.98-4.52; P < 0.001) were independently associated with hospitalization in MSH. In the least disadvantaged or segregated areas, all patients were unlikely to be hospitalized in MSH; however, in the most disadvantaged or segregated areas, White patients had only a small increase in odds of MSH hospitalization while Black patients had a very large increase (P for interaction terms = 0.001 and 0.04, respectively). Conclusion Our findings on differences by race/ethnicity suggest a need for further research on factors driving site of care, to address the upstream socio-structural determinants of health, and to invest in minority-serving hospitals given the practical challenges of eliminating neighborhood disadvantage and residential segregation.

Scholars have often discounted social class as a substantial contributor to residential segregation by race, in part as a result of using the dissimilarity index, which is likely to show high levels of uneven group distribution regardless of socioeconomic status (SES), and in part as a result of using limited categories of SES. This study expands on prior research by examining residential segregation between black-alone and white-alone households in 36 metropolitan statistical areas (MSAs) with 2000 decennial census data, using both spatial unevenness (dissimilarity) and two types of experiential indicators (exposure indices), measuring SES across income levels and accounting for the presence of other races. Findings show that black households with higher incomes live in neighborhoods with greater exposure and lower isolation than do black households with lower incomes. Additionally, while the dissimilarity of black households decreases with income, unevenness is not as strongly connected to income as are the experiential measures. While race remains a primary determinant of residential segregation, results indicate substantial class differences.