Table 1.
Summary of reviewed studies that used multilevel modeling to investigate the relationship between racial segregation and health
| Authors (Year) | Geographic Unit (Neighborhood Unit) | Segregation Measure | Comparison Groups | Health Outcome | Key Findings |
| Subramanian et al. (2005) | Metropolitan area (Census tract) | Dissimilarity index (evenness) Isolation index (exposure) | Dissimilarity: Blacks vs. Whites. Isolation: Blacks vs. Others. | Self-rated health | 1. High black isolation was associated with increased odds of poor self-rated health among blacks (using cross-level interaction between black isolation index and black). |
| Bell et al. (2006) | Metropolitan area (Census tract) | Isolation index (exposure) Spatial proximity (clustering) | Blacks vs. Whites. | Birth weight; Preterm birth; Intrauterine growth restriction | 1. Higher isolation was related to lower birth weight, higher odds of prematurity and higher rates of fetal growth restriction. 2. Higher clustering was associated with higher birth weight and lower odds of premature birth. |
| Chang (2006) | Metropolitan area (Census tract) | Isolation index (exposure) | Blacks vs. Whites. | Body mass index (BMI) | 1. Among non-Hispanic blacks, racial isolation increased BMI and odds of being overweight. No association was found for non-Hispanic whites. |
| Grady (2006) | New York City (Census tract) | Local spatial segregation index (exposure with consideration of spatial structure) | Blacks vs. Others. | Low birth weight | 1. Neighborhood segregation was positively associated with the odds of being low birth weight (< 2,500 grams). |
| Lopez (2006) | Metropolitan area (Census tract) | Dissimilarity index (evenness) | Blacks vs. Whites. | Physical activity (<10 minutes/week of moderate or vigorous physical activity) | 1. Segregation was positively related to the risk of being physically inactive and this relationship was found for Whites. |
| Mobley et al. (2006) | ZIP code (Neighborhood definition not specified) | Isolation index (exposure) *(a) | Not Specified. | Body mass index; Coronary heart disease risk (CHD) | 1. Segregation was negatively associated with CHD risk among Black, Hispanic, and Asian women and positively associated with CHD risk among American Indian women. This finding was found using interaction between individual race/ethnicity and isolation index. |
| Robert & Ruel (2006) | County (Census tract) | Dissimilarity index (evenness) | Blacks vs. Whites. | Self-rated health | 1. In the NSFH (National Survey of Families and Households), there was a positive association between county racial segregation and poorer self-rated health among Whites but not Blacks. 2. In the ACL (Americans’ Changing Lives), there was no statistically significant association between racial segregation and self-rated health. |
| Grady & McLafferty (2007) | New York City (Census tract) | Local spatial segregation index (exposure with a consideration of spatial structure) | Blacks vs. Others. | Low birth weight | 1. For U.S.-born Black women, segregation was associated with a higher odds of being low birth weight. |
| Lee & Ferraro (2007) | Neighborhood Cluster *(b) (Census block) | Isolation index and Interaction index (both exposure) | Isolation: Puerto Rican American vs. Others. Mexican American vs. Others. Interaction: Puerto Rican American vs. African American. Mexican American vs. African American. |
Acute physical symptoms and Disability (instrumental activities of daily living) | 1. Segregation was positively related to health problems for Puerto Rican Americans (not for Mexican Americans). 2. Cross-level interaction between generation and origin showed that segregation protects second or later generation of Mexican Americans from disability and physical symptoms. |
| Grady & Ramirez (2008) | New York City (Census tract) | Local spatial segregation index (exposure with a consideration of spatial structure) | Blacks vs. Whites. | Maternal and infant risk factors (Low birth weight) | 1. Segregation was positively associated with the odds of having a low birth weight infant for African-American and White women. |
| Osypuk & Acevedo-Garcia (2008) | Metropolitan area (Census tract) | Hypersegregation *(c) | Blacks vs. Whites. | Preterm birth (PTB) | 1. Black infants in hypersegregated metropolitan areas were more likely to be preterm than Black infants in non-hypersegregated metropolitan areas. 2. Hypersegregaion was positively associated with Black-White PTB disparities. 3. The age-PTB association for blacks was worse in hypersegregated areas (3-way interaction). |
| Haas et al. (2008) | County (Census tract) | Isolation index (exposure) | Blacks vs. Whites. Hispanics vs. Whites. | Breast Cancer Care and Mortality | 1. Higher black segregation were associated with worse breast cancer care for both blacks and whites, but not associated with mortality. 2. Higher Hispanics segregation were associated with worse breast cancer care for both Hispanics and whites, but not associated with mortality. |
| Haas et al. (2008) | County (Census tract) | Isolation index (exposure) | Blacks vs. Whites. Hispanics vs. Whites. | Early versus. Late stage diagnosis of cancer (breast, colorectal, prostate, and lung) | 1. Black/white disparities in early-stage cancer diagnosis (except lung) are the largest in low segregation areas. 2. Hispanic/white disparity in early-stage diagnosis was less obvious, but the largest disparity in breast cancer was observed in low segregation areas. |
| Lee (2009) | Neighborhood Cluster *(b) (Census block) | Isolation index (exposure) | Puertorican American vs. Others. Mexican American vs. Others. | Depressive symptoms and anxiety | 1. Neighborhood segregation was positively associated with depressive symptoms and anxiety among Mexican Americans, rather than Puerto Rican Americans. |
| Walton (2009) | Metropolitan area (Census tract) | Isolation index (exposure) Spatial proximity (clustering) | Isolation: Asian vs. Whites. Hispanics vs. Whites. Blacks vs. Whites. Spatial Proximity (clustering): Asian vs. Whites. Hispanics vs. Whites. Blacks vs. Whites. |
Birth weight | 1. Segregation was negatively related to the odds of having a low birth weight baby among Asian Americans. 2. Segregation marginally increased the odds of low birth weight among African Americans after taking poverty into account. Segregation does not affect birth weight among Latino Americans. |
| Grady (2010) | Detroit Metropolitan Area (Automated zone) *(d) | Spatial isolation index (exposure with a consideration of spatial structure) G* (clustering) | Blacks vs. Others. | Low birth weight | 1. High segregation was associated with a higher risk of having low birth weight infants and the disparities in having low birth weight infants between black mothers and others was less in racially isolated areas. 2. Segregation was also related to a higher odds of having preterm births and the disparities in having preterm births between black mothers and others was reduced in racially clustered areas. |
| Kramer et al. (2010) | Metropolitan area (Census tract) | Isolation index (exposure) Dissimilarity index (evenness) both consider spatial structure | Blacks vs. Others. Whites vs. Others. | Preterm birth | 1. For Black mothers, isolation was associated with very preterm birth and moderately preterm birth. |
| Mellerson et al. (2010) | Metropolitan area (Census tract) | Dissimilarity index (evenness) | Hispanics vs. Whites. | Exercise | 1. For Hispanics, the odds of exercise was lower when living in highly segregated areas than in non-segregated areas. |
| Osypuk et al. (2010) | Metropolitan area (Census tract) | Interaction index (exposure) | Mexican Immigrant vs. other Mexican Immigrant. Mexican Immigrant vs. US-born Mexican-origin. US-born Mexican-origin vs. US-born Mexican-origin. US-born Mexican-origin vs. Mexican Immigrant. |
Birth weight | 1. Living in metropolitan areas with high segregation of US-born Mexican-origin residents was associated with lower infant birth weights for US-born Mexican-origin mothers. |
| Bellatorre et al. (2011) | Metropolitan area (Census tract) | Dissimilarity index (evenness) Interaction index (exposure) | Multi-group measures but not specified | Total allostatic load (AL) and inflammatory response AL | 1. Inflammatory response AL was positively related to both the evenness and exposure domains of segregation and this relationship did not differ by race/ethnicity. 2. The positive relationship between total AL and segregation was not found till the cross-level interactions between race/ethnicity and segregation measures were considered. |
| Debbink & Bader (2011) | Michigan Metropolitan area (Census tract) | Local isolation index (exposure with a consideration of spatial structure) | Blacks vs. Whites. | Low birth weight | 1. Odds of low birth weight was higher in segregated neighborhoods in Michigan metropolitan area. |
| Hao et al. (2011) | County (Census block) | Isolation index (exposure) | Blacks vs. Whites. | Health-related quality of life (HRQOL) *(e) | 1. HRQOL was poorer among cancer survivors living in highly segregated counties, whereas individual race/ethnicity was no longer significant. |
| Kershaw et al. (2011) | Metropolitan area (Census tract) | Isolation index (exposure) | Blacks vs. Whites. | Black-White disparities in hypertension | 1. Racial disparities in the odds of hypertension were significantly smaller in low-than in high-segregation areas. 2. Race-segregation interaction showed that the race-hypertension association was modified by racial segregation, increasing hypertension for blacks but decreasing for whites. |
| Anderson (2012) | Metropolitan area (Census tract) | Isolation index (exposure) | Blacks vs. Whites. | Health-care coverage | 1. For blacks, higher segregation was associated with lower odds of having health insurance. |
| Corral et al. (2012) | Metropolitan area (Census tract) | Isolation index (exposure) | Blacks vs. Whites. | Overweight/Obesity, exercise and fruit/vegetable consumption | 1. African Americans living in moderate segregation areas were more likely to consume 5+fruit or vegetable per day than their counterparts in low segregation areas. 2. African Americans in high segregation areas were more likely to be obese/overweight than those in low segregation areas. |
| Russell et al. (2012) | Metropolitan area/Micropolitan area[MiA] (Census tract) | Theil’s information theory index (evenness) | Not specified. | Breast cancer mortality, all-cause mortality | 1. For black women, as MA/MiA segregation increased, the risk for breast cancer mortality increased. Interaction between race and segregation was used to find the racial difference. |
| Biello et al. (2013) | Metropolitan area (Census tract) | Hypersegregation *(c) | Blacks vs. Whites. | Sexually transmitted infections (STIs) | 1. Hypersegregation was not associated with the sex risk index. |
| Biello et al. (2013) | Metropolitan area (Census tract) | Dissimilarity index (evenness) Isolation index (exposure) Relative concentration index (concentration) Absolute centralization index (centralization) Spatial proximity index (clustering) Hypersegregation *(d) | Blacks vs. Whites (implicit). | Age at first sexual intercourse | 1. High segregation was related to early age at first sexual intercourse, but isolation seems to show a U-shape effect on the dependent variable. 2. The interaction between individual race and segregation, to some extent, suggested that segregation may modify the relationship between race and age at first intercourse. 3. The adverse effect of segregation was multiplied in hypersegregated areas. |
| Britton & Shin (2013) | Metropolitan area (Census tract) | P-star index (isolation/exposure) Dissimilarity index (evenness) | P-star index: Mexican Immigrants vs. US-born Mexican-origin residents. Blacks vs. others. Dissimilarity: Blacks vs. Whites. Mexican immigrant vs. Whites. US-born Mexican vs. Whites. |
Very preterm birth | 1. Segregation was positively associated with very preterm birth for both African American and Mexican-origin women, but this association was robust across different measures of segregation only for African Americans. |
| Jones (2013) | Metropolitan area (Census tract) | Hypersegregation *(c) | Blacks vs. Whites. | Hypertension | 1. Segregation was positively related to hypertension and contributed to hypotension more in extremely hypersegregated areas than do individual characteristics. |
| Kershaw et al. (2013) | Metropolitan area (Census tract) | Isolation index (exposure) | Blacks vs. Others. Hispanics vs. Others. | Obesity (BMI) | 1. There was no relationship between segregation and obesity among men. For black women, high segregation was associated with high obesity. 2. Mexican-American women living in high segregation areas had lower obesity. |
| Mendez et al. (2013) | Metropolitan area (Census tract) | Dissimilarity (evenness) Isolation and interaction index (both exposure) | Blacks vs. Whites. | Preterm birth | 1. Residential segregation was not associated with preterm birth among this sample of women. |
| Nelson (2013) | Metropolitan area (Census tract) | Dissimilarity (evenness) | Hispanic vs. Whites. | Self-rated health (SRH) | 1. Residential Segregation was associated with poorer SRH for Hispanics and Mexicans. 2. For Cubans, residential segregation was related to better SRH. |
| Anderson & Fullerton (2014) | Metropolitan area (Census tract) | Isolation index (exposure) | Latino vs. Others. | Self-rated health (SRH) Personal Physician Health-care Coverage | 1. For Latinos, Latino segregation was associated with poorer SRH and lower odds of having a personal physician and health insurance. 2. For Whites, Latino segregation was not related to the outcomes. |
| Corral et al. (2014) | Metropolitan area (Census tract) | Isolation index (exposure) | Hispanic vs. Whites. | Obesity (BMI) | 1. Living in highly segregated metropolitan areas increased the odds of being obese for Hispanics. |
| Greer et al. (2014) | Metropolitan area (Census tract) | Isolation index (exposure) | Blacks vs. Others. | Cardiovascular mortality | 1. Among blacks, segregation was positively related to heart disease mortality rates in young and old age groups. But it was only associated with stroke mortality rates in old age group. 2. Among whites, segregation was associated with heart disease mortality rates in young and old age groups. |
| Kershaw et al. (2014) | Metropolitan or Micropolitan area (Census tract) | Isolation index (exposure) | Hispanics vs. Others. | Obesity (BMI) | 1. Among Hispanic men, segregation was unassociated with BMI. 2. Segregation was positively related to BMI among Hispanic whites but negatively associated with BMI among Hispanic Blacks. |
| Plascak et al. (2014) | City or Town (place- level) (Neighborhood definition not specified) |
Gini coefficient (evenness) | Blacks vs. Whites. | Diagnostic time to resolution | 1. Living in segregated area was not associated with diagnostic time to resolution following an abnormal breast or cervical cancer screening test. |
| Yang et al. (2014) | County (Census tract) | Interaction index (exposure) | Blacks vs. Whites. Asians vs. Whites. Hispanics vs. Whites. | Maternal Smoking during pregnancy | 1. Living in a less racially segregated area was related to a lower probability of smoking during pregnancy for black women. 2. Living in a racially segregated area increased the probability of maternal smoking for Asian women and Hispanic women. |
| Yi et al. (2014) | United Hospital Fund (UHF) Neighborhood in New York City (Zip code) | Isolation index (exposure) | Blacks vs. Others. Hispanics vs. Others. Asians vs. Others. | Biomarkers of sodium and potassium consumption | 1. Higher Hispanic segregation was related to lower potassium intake. 2. Higher Black and Hispanic segregation was associated with higher sodium-potassium ratio. 3. Higher Asian segregation was related to lower sodium-potassium ratio. |
| Bower et al. (2015) | Metropolitan area (Census tract) | Isolation index (exposure) | Blacks vs. Others. | Obesity | 1. Among black women, residential segregation was positively associated with odds of obesity. |
| Britton & Velez (2015) | Metropolitan area (Census tract) | P-Star index (isolation/exposure) Dissimilarity index (evenness) | Puerto Rican. Vs. Others. Puerto Rican. Vs. Anglos. | Preterm birth | 1. Among Puerto Rican women, Puerto Rican residential segregation was positively related to the risk of very preterm birth. 2. Among US-born women, it was not related to preterm birth outcomes. |
| Grigsby-Toussaint et al. (2015) | Metropolitan area (Neighborhood definition not specified) | Hypersegregation *(c) | Latinos vs. Whites. | Diabetes risk | 1. Among Latinos, level of segregation was not associated with diabetes risk. |
| Kovalchik et al. (2015) | County (Census tract) | Isolation index (exposure) | Blacks vs. Whites. Hispanics vs. Whites. | Cognitive ageing | 1. Higher Hispanic segregation was associated with greater cognitive function at baseline and higher cognitive decline over 10-year period. 2. Black segregation was not related to cognitive ageing. |
| Lutfi et al. (2015) | Core-based statistical areas (CBSAs) (Census tract) | Dissimilarity index (evenness) Isolation index (exposure) Relative concentration index (concentration) Absolute centralization index (centralization) Spatial proximity index (clustering) Hypersegregation *(d) | Blacks vs. Whites. | Risky sexual behavior 1. # of partners in the past year 2. condom use at last sex 3. Composite index of the previous two |
1. Among non-Hispanic blacks, higher centralization and concentration showed strong associations with risky sexual behavior. |
| Piontak et al. (2016) | County (Census tract) | Dissimilarity index (evenness) | Blacks vs. Whites. Hispanics vs. Whites. | Childhood obesity | 1. For students, living in minority-segregated area was higher associated with obesity than those in less segregated area. 2. The association was not statistically significant once school-level poverty was controlled. |
| Plascak et al. (2016) | Zip-Code (Census block) | Gini coefficient (evenness) | Latinos vs. Whites. | Self-rated health (SRH) | 1. The relationship between the Latino residential segregation and SRH was U shaped. 2. In lower segregated area, increases in Latino segregation was negatively associated with SRH, while in higher segregated area, increases in Latino segregation was positively associated with SRH. |
| Caldwell et al. (2017) | County (Census tract) | Isolation index (exposure) | Blacks vs. others Hispanics vs. others | Access to healthcare | 1. Living in rural areas with high segregation was associated with worse access to health care among both blacks and Hispanics. 2. Living in rural areas with high segregation was positively associated with the odds of reporting their health care needs were met among both blacks and Hispanics. |
| Harvey et al. (2017) | County (Census tract) | Dissimilarity index (evenness) | Hispanics vs. Others | Late stage melanoma diagnosis (LSMD) | 1. Segregation was not associated with the odds of LSMD. |
| Mobley et al. (2017) | County (Neighborhood definition not specified) | Isolation index (exposure) | Blacks vs. whites Hispanics vs. whites Asians vs. whites | Breast cancer mammography use | 1. Living in highly segregated Hispanic neighborhood was associated with increased odds of mammography use. 2. Living in highly segregated black or Asian neighborhood was associated with decreased odds of mammography use. |
| Mobley et al. (2017) | County (Census tract) | Isolation index (exposure) | Blacks vs. others Hispanics vs. others Asians vs. others | Colorectal cancer (CRC) diagnosis | 1. Living in highly segregated black neighborhood was associated with lower odds of late-stage CRC diagnosis. 2. Living in highly segregated Asian neighborhood was associated with higher odds of late-stage CRC diagnosis. 3. Living in highly segregated Hispanic neighborhood showed no significant association. |
| Mobley et al. (2017) | County (Census tract) | Isolation index (exposure) | Whites vs. others Blacks vs. others Hispanics vs. others Asians vs. others | Breast cancer diagnosis | 1. Living in a highly isolated Asian community was associated with lower odds of late-stage breast cancer diagnosis. 2. Living in a highly isolated White community was associated with higher odds of late-stage breast cancer diagnosis. |
| Nobles et al. (2017) | Metropolitan area (Census tract) | Dissimilarity index (evenness) Isolation index (exposure) | Latinos vs. Others | Mental distress | 1. Higher Latino isolation was associated with lower level of mental distress. 2. Higher dissimilarity index was associated with lower level of mental distress among men, but not among women. |
| Ojinnaka et al. (2017) | County (Census tract) | Isolation index (exposure) | Blacks vs. whites | Surgical treatment of early-stage breast cancer | 1. Living in highly segregated areas was associated reduced odds of surgical treatment of early-stage breast cancer. 2. Blacks living in highly segregated areas were less likely to receive surgical treatment than their counterparts living in areas with lower level of segregation. |
| Pennap et al. (2017) | County (Neighborhood definition not specified) | Isolation index (exposure) | Hispanics vs. Others | Attention-deficit/hyperactivity disorder (ADHD) diagnosis and stimulant use | 1. Hispanics living in segregated areas are less likely to receive an ADHD diagnosis and stimulant use than Hispanics living in less segregated areas. |
| Yang et al. (2017) | Census tract (Census block) | Entropy (evenness) Isolation (exposure) Absolute centralization index (centralization) Delta index (concentration) Spatial proximity index (clustering) | Whites vs. Blacks Whites vs. Hispanics | Self-rated health (SRH) | 1. High white/black concentration widened the SRH disparities between whites and blacks. 2. Higher levels of white/Hispanic centralization were associated with narrower gaps in SRH. 3. There was no single dimension of segregation that statistically outperforms others. |
| Goodman et al. (2018) | County (Census tract) | Dissimilarity & Entropy (evenness) Isolation, Correlation, & Local spatial segregation index (exposure) | Blacks vs. Whites | Body mass index | 1. Racial residential segregation (evenness and exposure) was positively associated with individual body mass index. |
| Lutfi et al. (2018) | Core-based statistical areas (CBSAs) (Census tract) | Dissimilarity index (evenness) Isolation index (exposure) Relative concentration index (concentration) Absolute centralization index (centralization) Spatial Proximity (clustering) | Blacks vs. Whites | STI diagnosis | 1. High segregation was associated with the odds of STI diagnosis. 2. The associations were the strongest when segregation was measured by dissimilarity and relative concentration indices. 3. The associations were stronger among males than females, but there was no age group pattern. |
| Mayne et al. (2018) | Chicago Metropolitan area (Census tract) | Local Getis-Ord Gi* statistic | Blacks vs. Others | Hypertensive disorder of pregnancy (HDP) | 1. Segregation was positively associated with odds of having HDP only among black women living in higher poverty neighborhoods. |
| Salow et al. (2018) | Chicago Metropolitan area (Census tract) | Local Getis-Ord Gi* statistic | Blacks vs. Others | Adverse pregnancy outcomes | 1. Overall and spontaneous preterm births are more prevalent in high segregation areas than low segregation areas. |
| Williams et al. (2018) | Hospital Reference Regions (Zip Code Tabulation Area: ZCTA) | Dissimilarity index (evenness) Isolation index (exposure) | Blacks vs. Whites | Stillbirth | 1. Low segregation was associated with decreased odds of stillbirth among blacks than among whites. 2. Decreasing segregation was associated with a reduction in odds of stillbirth among blacks, but not whites. |
| Yu et al. (2018) | County (Census tract) | Dissimilarity index (evenness) Isolation index (exposure) Concentration index (concentration) | Dissimilarity: Blacks vs. Whites Hispanics vs. Whites Asians vs. Whites Isolation: Whites vs. others Blacks vs. others Hispanics vs. others Asians vs. others | Obesity | 1. Blacks and Hispanics residing in segregated areas were more likely to be obese. These relationships were not found among Whites and Asians. |
| Bravo et al. (2019) | Census block (Neighborhood definition not specified) | Local, spatial index of racial isolation (exposure) | Blacks vs. Others | Type 2 diabetes mellitus | 1. Higher isolation of black residents was associated with increased risk of diabetes for whites and blacks. |
| Noah et al. (2019) | Census tract (blockgroup) | Isolation index (exposure) Delta index (concentration) | Blacks vs. Whites | Sexually transmitted diseases (STDs) during pregnancy | 1. Living in segregated neighborhood was associated with mother’s STDs during pregnancy. 2. White mothers living in segregated area were likely to have higher odds of acquiring STDs during pregnancy than black mothers. 3. Black mothers residing in segregated area were less likely to have acquired STDs during pregnancy than white mothers, if they live with co-ethnics. |
| Tempalski et al. (2019) | Metropolitan area (Census tract) | Isolation index (exposure) | Blacks vs. Others Latinos vs. Others | HIV testing | 1. For both blacks and Latinos, higher isolation was not associated with the odds of an individual getting tested for HIV. |
| Thomas-Hawkins et al. (2019) | County (Census tract) | Dissimilarity index (evenness) | Blacks vs. Whites | Emergency departments (ED) revisit | 1. Living in high segregation areas increased the odds of revisiting emergency departments (ED). 2. Blacks living in high segregation are more likely to revisit ED, compared to their white counterparts. |
*(a)
The author called it index of racial segregation; Kramer and Hogue (2009) categorized it as isolation index.
*(b)
Neighborhood Clusters (NC) are defined as the sampling unit for the Project on Human Development in Chicago Neighborhoods (PHDCN) community survey.
NCs consist of one or more geographically contiguous census tracts. Chicago’s 865 inhabited census tracts were aggregated to form 343 NCs.
*(c)
Hypersegregation is defined as highly segregation on ≥ 4 of 5 dimensions based on the work by Wilkes and Iceland (2004).
*(d)
The census tract boundary file was used in the development of Automated Zone Matching zones. AZM operates by recombining census tracts into a smaller set of output zones using an iterative process by which one tract is randomly selected and attribute constraint parameter(s) are evaluated.
*(e)
Health-related quality of life (HRQOL) was measured by the General Health subscale of the Short Form-36, which consists of five items that assess survivors’ perception of their health in general on 5-point Likert-type scales.