Skip Navigation


American Journal of Epidemiology Advance Access originally published online on March 25, 2008
American Journal of Epidemiology 2008 167(11):1295-1304; doi:10.1093/aje/kwn043
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
167/11/1295    most recent
kwn043v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (3)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Osypuk, T. L.
Right arrow Articles by Acevedo-Garcia, D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Osypuk, T. L.
Right arrow Articles by Acevedo-Garcia, D.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology © The Author 2008. Published by the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.

ORIGINAL CONTRIBUTIONS

Are Racial Disparities in Preterm Birth Larger in Hypersegregated Areas?

Theresa L. Osypuk1 and Dolores Acevedo-Garcia2

1 Department of Health Sciences, Bouvé College of Health Sciences, Northeastern University, Boston, MA
2 Department of Society, Human Development, and Health, Harvard School of Public Health, Boston, MA

Correspondence to Dr. Dolores Acevedo-Garcia, Department of Society, Human Development, and Health, Harvard School of Public Health, 677 Huntington Avenue, Kresge 7th Floor, Boston, MA 02115 (e-mail: dacevedo{at}hsph.harvard.edu).

Received for publication October 8, 2007. Accepted for publication February 7, 2008.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
The causes of the racial/ethnic disparity in preterm birth (PTB) remain largely unknown; traditional risk factors such as smoking and prenatal care fail to account for it. The authors examined whether living in metropolitan areas (MAs) with high levels of residential racial segregation along multiple dimensions (hypersegregation) was associated with higher rates of PTB or larger racial disparities in PTB and whether segregation modified the established race-age association in PTB. The authors merged 2000 natality data (n = 1,944,703) with US Census measures of Black-White hypersegregation. They executed two-level hierarchical logistic regression analyses among White and Black mothers in 237 MAs to estimate the odds of PTB by hypersegregation, race, and age, after controlling for covariates. In unadjusted and adjusted models, Black infants in hypersegregated MAs were more likely to be preterm than Black infants in nonhypersegregated MAs (p < 0.001). Black-White PTB disparities were larger in hypersegregated areas than in nonhypersegregated areas (p < 0.001), and the age-race association with PTB was modified by hypersegregation (p < 0.001). Living in a hypersegregated MA had a more pronounced association with PTB among older Black women, and racial disparities in PTB were larger in hypersegregated areas among older mothers (p < 0.001). Since over 40% of Black childbearing women live in hypersegregated areas, residential segregation may be an important social determinant of racial birth disparities.

ethnic groups; health status disparities; infant, low birth weight; minority health; prejudice; premature birth; residence characteristics

Abbreviations: CI, confidence interval


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
The persistent racial/ethnic disparity in birth outcomes is one of the most startling health trends in the United States. US infants born to non-Hispanic Black women are over 50 percent more likely to be born preterm than infants born to non-Hispanic White women (1). However, the causes of preterm birth and of the racial disparity in birth outcomes are poorly understood (2–4). Although individual-level maternal risk factors, including age, marital status, and cigarette smoking, account for only half of the racial risk in low birth weight (5, 6), explanations for social inequalities in maternal health have generally focused on individual risk factors, directing attention away from structural explanations (7, 8).

To identify factors accounting for the large and persistent Black-White disparity in health, it is necessary to focus on factors that either differentially influence Blacks and Whites or uniquely affect Blacks or Whites. Residential segregation is such a phenomenon (9). Residential segregation is the degree to which two or more social groups live apart from one another in different areas of the urban environment (10). Although it has decreased modestly over the past 30 years, a high level of residential segregation for Blacks persists in the United States (11). The magnitude of residential segregation for any given dimension is higher for Blacks than for other racial groups. Moreover, Blacks experience simultaneous high segregation across multiple dimensions of segregation (hypersegregation), while other groups do not (12).

The pathways from racial segregation to birth outcomes are not clear, yet existing evidence points to the importance of individual and neighborhood deprivation acting through health behavior and stress pathways (2). For instance, persons living in Black neighborhoods are exposed to higher rates of crime and neighborhood poverty (13), and exposure to such chronic or acute stressors has been linked to worse birth outcomes (7, 8, 14–18) through immune and neuroendocrine pathways (2, 19). Individual-level stressors may be more prevalent in deprived neighborhoods (20), and four distinct pathways link individual-level stressors with preterm delivery: adverse health behaviors (smoking, poor nutrition), psychosocial factors (depression, lack of social support), stress hormones that may initiate labor, and depressed immune functioning that causes increased susceptibility to infection (21, 22).

Additionally, the effects of disadvantaged contexts may be stronger among older Black women. According to the weathering hypothesis, Black women may experience worse birth outcomes with age, reflecting the cumulative effects of psychosocial and environmental hazards associated with population-level patterns of racial and social inequality (4). For example, prior research has documented that risk of adverse birth outcomes may rise with increasing age among disadvantaged women or in areas with greater poverty (4, 8, 18, 23). Although the pattern of differential age-related birth outcomes by race is established (4, 8, 18, 23, 24), this age hypothesis has not been tested with regard to metropolitan patterns of racial birth inequality.

In this study, we examined how residential racial segregation was associated with preterm birth. We asked: Do Black mothers who reside in metropolitan areas characterized by hypersegregation (i.e., simultaneous high segregation across multiple dimensions) have worse preterm birth outcomes than those who reside in less segregated metropolitan areas? Are racial preterm birth disparities higher in hypersegregated areas than in nonhypersegregated areas? We examined the weathering hypothesis by asking: Does hypersegregation modify the race-age association with preterm birth? We built upon prior literature on the relation between segregation and health to incorporate multilevel analytic techniques, as well as a measure of segregation that may better represent the severity of Black segregation than single segregation measures but has not been employed in health studies despite its conceptual strength (25).


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
Individual health data
We used the National Center for Health Statistics' 2000 natality data set (26), the annual census of livebirths based on the U.S. Standard Certificate of Live Birth (27). We excluded multiple births, rural births (since segregation is a metropolitan-area-level phenomenon), births occurring in metropolitan areas with a population of less than 100,000 (masked), births occurring in metropolitan areas with fewer than 5,000 Blacks (28, 29), and births to foreign-born mothers, since they display different birth patterns (30, 31).

The study outcome was preterm birth, operationalized as birth before the 37th week of gestation, where gestation was determined by the last menstrual period or by clinical estimate. We excluded births with unknown length of gestation or unknown birth weight (0.7 percent). Our first individual-level predictor of interest was the mother's self-reported race/ethnicity. We restricted our analysis to non-Hispanic Blacks ("Blacks") and non-Hispanic Whites ("Whites") because of differential patterns of Latino segregation (12) and Latina birth weight differences (30, 31) in comparison with Blacks. Our second individual predictor of interest was maternal age (centered and modeled as a quadratic variable). We excluded births to women under age 15 years and women over age 45 years. Based on the work of Geronimus (4), we hypothesized that hypersegregation would modify the associations of age and race with preterm birth. We modeled a three-way (cross-level) interaction among race, age, and segregation with nine terms to examine how segregation might be more detrimental for preterm birth at higher ages among Blacks. Other individual-level demographic covariates (see table 1 for details) included parity, infant sex, maternal education, marital status, smoking or alcohol drinking during pregnancy, and prenatal care (measured by the Adequacy of Prenatal Care Utilization Index (32)).


View this table:
[in this window]
[in a new window]

  		TABLE 1. Natality data for non-Hispanic White and non-Hispanic Black infants in 237 metropolitan areas, United States, 2000

 
Segregation data and measures
We operationalized residential segregation at the metropolitan level, since metropolitan areas approximate housing markets (10) and since we hypothesized that risks and resources which affect preterm birth are partially related to racial inequality in housing markets. Metropolitan areas may be a more appropriate level of aggregation than neighborhoods for modeling racial health disparities, for several reasons. Metropolitan areas may capture effects of housing market inequality, including not only effects of neighborhoods but also effects of social class advancement, housing discrimination, and housing stock (33–35). Moreover, distributions of neighborhood quality are nonoverlapping for Whites and Blacks in highly segregated metropolitan areas, making it methodologically challenging to model racial health disparities with neighborhood-level data while maintaining high internal and external validity (33).

Residential segregation is a global construct with five distinct dimensions of spatial separation within a metropolitan area, including exposure (the probability that Blacks have contact with Black neighbors), unevenness (the degree to which each neighborhood incorporates the same proportions of Blacks and Whites as the metropolitan area overall), clustering (the tendency of Black neighborhoods to cluster together), centralization (the degree to which Black neighborhoods are at the center of the metropolitan area), and concentration (population density) (10). The health literature has overlooked the complexity of residential segregation by testing singular dimensions of segregation (25, 29, 36–38). A high level of segregation with regard to any one dimension may be harmful, but since Blacks experience high segregation across multiple dimensions simultaneously (hypersegregation), the detrimental effects of segregation multiply (12). We chose to test this aspect of segregation in the present study, since examining just one dimension at a time may understate the severity of Black segregation (12).

We received special calculations for segregation indices from the US Census Bureau for non-Hispanic Blacks and non-Hispanic Whites (E. Steinmetz, Bureau of the Census, personal communication, 2004), calculated from 2000 census-tract-level data, for each metropolitan area. Metropolitan areas encompassed both Metropolitan Statistical Areas and Primary Metropolitan Statistical Areas of the US Census. We defined hypersegregated metropolitan areas a priori using criteria defined by segregation demographers (39, 40). First, we selected five measures characterizing the five dimensions of segregation (10, 12): absolute centralization, spatial proximity, relative concentration, dissimilarity, and isolation. Higher segregation scores indicate higher Black segregation, and metropolitan areas falling above 0.60 on any measure are considered highly segregated (39). (See the paper by Massey and Denton (10) for formulas and descriptions of segregation measures.) To create our hypersegregation predictor variable, we scored metropolitan areas with a 1 if they were highly segregated (>0.60) with regard to each of the five segregation measures and 0 otherwise, and then summed across the five measures to create a score ranging from 0 to 5 for each metropolitan area (see table 2). Hypersegregated metropolitan areas were those with scores of 4 or 5—areas found to be highly segregated on at least four dimensions.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Distribution of metropolitan areas and infants according to level of residential segregation,* United States, 2000

 
Using data from the 2000 US Census, we adjusted for the following metropolitan-area-level covariates, centered at their grand mean value: population size (log), poverty rate, median annual household income (log), and proportion Black. Lastly, we included fixed effects of the four census regions (Northeast, Midwest, South, and West).

Analytic methods
We merged the segregation measures with the birth observations based on the mother's metropolitan area of residence. We conducted bivariate analyses to check for collinearity and to determine the form of the associations. We then created two-level hierarchical logistic regression models for individuals within metropolitan areas, using HLM 6.01 software (41). First we modeled race, unadjusted for covariates (model 1). Although we initially tested for a main effect of segregation, we found that segregation modified the race-age-preterm birth association (results not shown), so we present results from the models demonstrating effect modification. We built two models to test three-way age-race-hypersegregation interactions. Model 2 included just these interactions and variables. Model 3 built on model 2 and adjusted for covariates. The significance of the segregation associations, interaction terms, and model fit were determined by means of log-likelihood estimates from full maximum likelihood estimation models with robust standard errors, at {alpha} = 0.05 (two-tailed). To test the hypersegregation interactions, we examined the improvement of each model in comparison with simpler models.

We specified a random-effects model as more flexible for modeling racial disparities, allowing the association between race and preterm birth to vary across metropolitan areas (42). We then stratified models of preterm birth by hypersegregation and race (models 4a and 4b) and by hypersegregation and age (models 4c and 4d). Models 4a and 4c were unadjusted, and models 4b and 4d were covariate-adjusted. Lastly, we recognize that this data set conveys considerable power to investigate associations. Since we hypothesized segregation as a distal contextual cause, we anticipated that its effect would be small, thereby requiring a large data set. We therefore focused on substantively meaningful associations in addition to statistically significant ones.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
Our final sample of births included 1,944,703 infants born in 237 metropolitan areas. (See table 1 for descriptive statistics.) The prevalence of preterm birth was 8.5 percent for Whites and 16.0 percent for Blacks. Figure 1 displays the differential age distribution of births by race. Table 2 displays the distribution of hypersegregation for metropolitan areas and infants. Twenty-two US metropolitan areas were hypersegregated for Blacks in the year 2000, constituting 9 percent of metropolitan areas and 28 percent of births. The 22 hypersegregated metropolitan areas were: Albany, Georgia; Atlanta, Georgia; Baltimore, Maryland; Baton Rouge, Louisiana; Beaumont-Port Arthur, Texas; Birmingham, Alabama; Buffalo, New York; Chicago, Illinois; Cleveland-Lorain-Elyria, Ohio; Detroit, Michigan; Flint, Michigan; Gary, Indiana; Los Angeles-Long Beach, California; Milwaukee-Waukesha, Wisconsin; Mobile, Alabama; Monroe, Louisiana; New York, New York; Newark, New Jersey; Philadelphia, Pennsylvania–New Jersey; Saginaw, Michigan; St. Louis, Missouri–Illinois; and Washington, DC–Maryland–Virginia–West Virginia.


Figure 1
View larger version (8K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 1. Distribution of births by maternal age and race, United States, 2000. The density of births was calculated using R software (version 2.3.1; R Project for Statistical Computing (http://www.r-project.org/)), applying a Gaussian smoothing kernel estimated over 512 points of each age distribution with a smoothing bandwidth equal to the standard deviation of the smoothing kernel.

 
In model 1, Black infants had almost twice the odds of preterm birth as White infants (unadjusted odds ratio = 1.94, 95 percent confidence interval (CI): 1.90, 1.98) (table 3). In model 2, Blacks in hypersegregated areas exhibited significantly worse preterm birth rates than Blacks in nonhypersegregated areas (Black race-hypersegregation interaction: p < 0.001). Race-age associations with preterm birth were modified by hypersegregation. Not only did Blacks exhibit an excess risk of preterm birth with older age (age-Black race interaction: p < 0.001), the age association for Blacks was also worse in hypersegregated areas (the three-way interaction coefficient was positive; p < 0.001). Adjustment for covariates did not appreciably reduce this association. Holding constant the higher age preterm birth risk for Blacks versus Whites, there was a steeper preterm birth incline with higher age for Blacks in hypersegregated areas than for Blacks in nonhypersegregated areas; for a Black woman 10 years older living in a hypersegregated area, the odds ratio was 1.10, as calculated from exponentiating the sum of the four hypersegregation logit coefficients.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Odds of preterm birth according to race, age, and hypersegregation in combined (unstratified) multiple hierarchical logistic regression models, United States, 2000

 
Figure 2 illustrates the three-way interaction and depicts the predicted probability of preterm birth by race, age, and hypersegregation, using coefficients from model 3. Black infants were at higher risk of preterm birth than White infants, on average. The solid line indicates that Black infants in hypersegregated metropolitan areas had a higher risk of preterm birth than did their Black counterparts in nonhypersegregated metropolitan areas (long-dashed line), and the association was more pronounced at higher maternal ages. These steeper age gradients with preterm birth are also illustrated in table 4, in models stratified by race and hypersegregation (models 4a and 4b). The adjusted odds ratio for preterm birth for a 10-year increase in age was 1.26 (95 percent CI: 1.22, 1.29) for Blacks in hypersegregated areas as compared with 1.18 (95 percent CI: 1.15, 1.21) for Blacks in nonhypersegregated areas, and both of these age coefficients were higher than those for Whites regardless of segregation (models 4b).


Figure 2
View larger version (10K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 2. Adjusted predicted probability of preterm birth by metropolitan area hypersegregation, maternal age, and race, United States, 2000.

 

View this table:
[in this window]
[in a new window]

  		TABLE 4. Odds of preterm birth by race, age, and hypersegregation in stratified multiple hierarchical logistic regression models, United States, 2000*{dagger}

 
Figure 3 depicts the racial difference in preterm birth (e.g., the adjusted predicted probability of preterm birth for Black mothers minus that for White mothers) by hypersegregation and age. The racial disparity in preterm birth was higher among older mothers, regardless of segregation. Moreover, the preterm birth racial disparity was higher in hypersegregated areas than in nonhypersegregated areas at nearly all ages, and the racial disparity was larger with each additional 10 years of age within each segregation level. These patterns are also illustrated in stratified models (table 4). For example, the unadjusted odds ratio for preterm birth for Black women aged 40–45 years versus White women aged 40–45 years was 3.04 (95 percent CI: 2.73, 3.37) in hypersegregated areas as opposed to 2.17 (95 percent CI: 1.96, 2.40) in nonhypersegregated areas (models 4c). For women aged 35–39 years, the preterm birth odds ratio for Black race was 2.75 (95 percent CI: 2.52, 3.02) in hypersegregated areas as compared with 2.34 (95 percent CI: 2.23, 2.45) in nonhypersegregated areas. These odds ratios declined in the adjusted model (models 4d), but the trends persisted.


Figure 3
View larger version (9K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 3. Adjusted racial disparity in predicted probability of preterm birth by metropolitan area hypersegregation and maternal age, United States, 2000. The racial difference was calculated as the Black predicted probability of preterm birth minus the White predicted probability.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
In this analysis, we found that the probability of preterm birth for Blacks was higher, and that the racial disparity in preterm birth was larger, for infants born in metropolitan areas characterized by hypersegregation—that is, simultaneous high residential racial segregation across four or more dimensions of segregation. Moreover, the well-known association of race and age with preterm birth was modified by hypersegregation, and associations were maintained in the presence of maternal demographic, health behavior, and socioeconomic characteristics. The slope of the association between preterm birth and age was not only steeper among Black women than among White women but it was also steeper for Black women in hypersegregated areas than for Black women in nonhypersegregated areas. This finding suggests that age-related preterm birth patterns may be related to stressors associated with residential hypersegregation facing Black women. To our knowledge, this is the first study to have documented an association between hypersegregation and racial health disparities.

Why might hypersegregation be associated with higher rates of preterm birth for Black infants and with larger racial disparities in preterm birth? Segregation is a racialized phenomenon: the spatial manifestation of institutional racism and separation of racial groups across neighborhood lines within metropolitan areas. Racial discrimination in housing markets is a likely cause of residential segregation and differential neighborhood environments (43). Experimental evidence indicates that when Blacks seek to rent an apartment or buy a house, they experience worse treatment than comparable Whites approximately 20 percent of the time (44). Residential segregation relegates Blacks to living in the worst neighborhoods, including neighborhoods characterized by high levels of poverty and violent crime, and the racial disparity in neighborhood quality is largest in highly segregated areas (33, 45). The metropolitan-area-level segregation associations we observed here partially captured neighborhood inequalities. Neighborhood quality may affect preterm birth through several mechanisms—for instance, through stress pathways or constant triggering of the "fight or flight" response in high-violent-crime areas, which exerts wear and tear on the body (19, 46). Neighborhood quality may also affect preterm birth through area-related factors pertaining to risks and resources, including the availability of nutritious food, access to adverse health products (cigarettes, alcohol), and social pathways (6, 14). We were unable to test neighborhood as a mediator because lower-level geographic identifiers are masked in the National Center for Health Statistics natality data (26).

In this analysis, we found that Black women experienced worse birth outcomes with advancing age in hypersegregated areas, which may provide support for the weathering hypothesis (4). Our study could not test whether the same women experienced worse outcomes over time, since it was not a longitudinal study, but the preterm birth patterns we observed among Black women in hypersegregated areas may signal a greater presence of hazards or stressors associated with racial inequality there (e.g., higher exposure to crime, neighborhood poverty, or housing discrimination) (18). However, the racial disparity in the age-preterm birth association may alternately be due to childbearing patterns, including differential timing/delay of and selection into childbearing (4, 47, 48), short interpregnancy intervals (49), differential intentions to bear children (50), and obstetric factors arising from prior births. We could not sort out these alternate explanations with our data, although sensitivity models confirmed that the age-race-hypersegregation interaction was not explained by prior preterm births among multiparous women or by better control for parity (results not shown). Although distributions of fertility by age may not be entirely comparable between racial groups, it is reasonable to assume that within racial groups, fertility distributions among women in different metropolitan areas are comparable, aside from the consequences of residential segregation. Thus, under a conservative assumption, even if race-age-preterm birth associations were due to childbearing patterns, we still observed differential patterns in hypersegregated areas, which would suggest that segregation could influence childbearing.

Although we found a detrimental association between hypersegregation and preterm birth for Blacks, the racial disparity in preterm birth far outweighed the segregation association. However, a large number of women at a small risk (e.g., 41 percent of Black childbearing women live in hypersegregated areas) may give rise to more cases than the small number who are at high risk; thus, a small effect of an area variable can have large effects on population health (51, 52). Distal causes at the population level, which may include residential segregation, are expected to have smaller effects on disease than more proximal causes, since there is greater uncertainty associated with distal causes (53). Although we employed a common approach in the literature for modeling contextual associations, the association of a contextual cause with a health outcome may not be easily reduced to individual-level counterparts that enter the body in a simple causal chain. Rather, the pathways may be interactive and various (53). Therefore, we expected small effect sizes for segregation that operates through multiple mediators to ultimately influence birth outcomes. Furthermore, given that segregation is an indicator of institutional racism, some of the effects of segregation may have been captured by the race coefficient at the individual level. Models 3, 4b, and 4d may also have underestimated segregation, since they adjusted for variables that have been conceptualized as mediators of segregation and birth outcomes (29), including education and health behaviors. In sum, the public health significance of the small effect of segregation resides in pursuing social context as a cause of preterm birth, which suggests new avenues for preventive strategies, by addressing broader determinants than those specified within a narrow biomedical approach—an approach that has heretofore been unable to explain racial/ethnic disparities in preterm birth (2, 18).

Threats to validity exist in our study, given the cross-sectional design and the limits of the data. First, our association of segregation with preterm birth may have resulted from bias due to omitted variables or measurement error. Relevant omitted variables include social class, given its strong association with preterm birth (54). We controlled for maternal education but could not control for income, poverty, or wealth in these data. However, since segregation is hypothesized to affect health through social class pathways (35, 55), it may be improper to control for it as a mediator. Therefore, we presented estimates that were unadjusted for covariates. Our segregation association might also reflect some other aspect of metropolitan confounding, despite our control for metropolitan area factors; however, segregation associations for Blacks were robust to the addition of metropolitan area covariates. Alternately, the segregation associations may have been due to between-metropolitan-area migration-related selection. However, selection may be less of a threat in metropolitan segregation studies than in neighborhood studies, because migration is much more common between neighborhoods within metropolitan areas than between metropolitan areas (55, 56).

Second, preterm birth may have been mismeasured in the birth certificate data. Missing or misclassified data on gestational age may produce misclassification of up to 5 percent of observations (57). Since our conceptual model hypothesized preterm birth as the most relevant outcome of interest, we modeled it here. We found comparable results when we used birth weight as the outcome instead of preterm birth (results not shown). It is also possible that segregation affects the gestational age estimate itself, by resulting in poor access to prenatal care and late registration and therefore erroneous gestational age. In sensitivity analysis (results not shown), we found some geographic differences in determination of gestational age and in prenatal care. However, excluding areas with high proportions using clinical estimates or areas with missing prenatal care data did not alter our results. Alternately, missing covariate data may have biased our associations; however, in sensitivity analyses that excluded missing data, we found comparable results (results not shown). We know of no better data for testing metropolitan segregation-birth hypotheses with such strong generalizability.

In conclusion, our analysis found that Black women living in hypersegregated metropolitan areas had a higher probability of preterm birth than did their counterparts in nonhypersegregated areas, and the associations were worse among older Black women. The racial preterm birth disparity was also larger for older mothers and women living in hypersegregated areas. Residential segregation is an important social phenomenon that may be a fundamental cause of racial health disparities (35). Since racial disparities in birth outcomes remain poorly understood and since the size of racial birth disparities appears to be large regardless of segregation, understanding the causes may require situating individuals and neighborhoods within a metropolitan context of racial inequality.


   ACKNOWLEDGMENTS
 
Financial support was provided by the Robert Wood Johnson Foundation through the Health and Society Scholars Program.

The authors thank Drs. Dawn Misra, Sandro Galea, and Ichiro Kawachi for providing helpful comments on the manuscript.

Partial results from this paper were presented at the 101st annual meeting of the American Sociological Association in Montreal, Quebec, Canada (August 11–14, 2006).

Conflict of interest: none declared.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. Martin JA, Hamilton BE, Sutton PD, et al. Births: final data for 2004. (National vital statistics reports, vol 55, no. 1). (2006) Hyattsville, MD: National Center for Health Statistics. 102.
  2. Behrman RE, Butler AS. Committee on Understanding Premature Birth and Assuring Healthy Outcomes. Preterm birth: causes, consequences, and prevention. (2006) Washington, DC: National Academies Press.
  3. Fiscella K. Racial disparities in preterm births: the role of urogenital infections. Public Health Rep (1996) 111:104–13.
  4. Geronimus AT. Black/white differences in the relationship of maternal age to birthweight: a population-based test of the weathering hypothesis. Soc Sci Med (1996) 42:589–97.[CrossRef][Web of Science][Medline]
  5. Kleinman JC, Kessel S. Racial differences in low birth weight: trends and risk factors. N Engl J Med (1987) 317:749–53.[Abstract]
  6. Buka SL, Brennan RT, Rich-Edwards JW, et al. Neighborhood support and the birth weight of urban infants. Am J Epidemiol (2003) 157:1–8.[Abstract/Free Full Text]
  7. Roberts EM. Neighborhood social environments and the distribution of low birthweight in Chicago. Am J Public Health (1997) 87:597–603.[Abstract/Free Full Text]
  8. Rauh VA, Andrews HF, Garfinkel RS. The contribution of maternal age to racial disparities in birthweight: a multilevel perspective. Am J Public Health (2001) 91:1815–24.[Abstract/Free Full Text]
  9. LaVeist TA. Racial segregation and longevity among African Americans: an individual-level analysis. Health Serv Res (2003) 38:1719–33.[CrossRef][Web of Science][Medline]
  10. Massey DS, Denton NA. The dimensions of residential segregation. Soc Forces (1988) 67:281–315.[CrossRef][Web of Science]
  11. Iceland J, Weinberg DH, Steinmetz E. Racial and ethnic residential segregation in the United States: 1980–2000. (Census 2000 special report series no. CENSR-3). (2002) Washington, DC: Bureau of the Census, US Department of Commerce.
  12. Massey DS, Denton NA. Hypersegregation in U.S. metropolitan areas: black and Hispanic segregation along five dimensions. Demography (1989) 26:373–91.[Web of Science][Medline]
  13. Massey D. Residential segregation and neighborhood conditions in U.S. metropolitan areas. In: America becoming: racial trends and their consequences—Smelser NJ, Wilson WJ, Mitchell F, eds. (2001) Vol 1. Washington, DC: National Academy Press. 391–434.
  14. Morenoff JD. Neighborhood mechanisms and the spatial dynamics of birth weight. AJS (2003) 108:976–1017.[CrossRef][Medline]
  15. O'Campo P, Xue X, Wang M-C, et al. Neighborhood risk factors for low birthweight in Baltimore: a multilevel analysis. Am J Public Health (1997) 87:1113–18.[Abstract/Free Full Text]
  16. Messer LC, Kaufman JS, Dole N, et al. Neighborhood crime, deprivation, and preterm birth. Ann Epidemiol (2006) 16:455–62.[CrossRef][Web of Science][Medline]
  17. Pearl M, Braveman P, Abrams B. The relationship of neighborhood socioeconomic characteristics to birthweight among 5 ethnic groups in California. Am J Public Health (2001) 91:1808–14.[Abstract/Free Full Text]
  18. Collins J, Simon D, Jackson T, et al. Advancing maternal age and infant birth weight among urban African Americans: the effect of neighborhood poverty. Ethn Dis (2006) 16:180–6.[Web of Science][Medline]
  19. Massey D. Segregation and stratification: a biosocial perspective. Du Bois Rev (2004) 1:1–19.
  20. Aneshensel CS, Sucoff CA. The neighborhood context of adolescent mental health. J Health Soc Behav (1996) 37:293–310.[CrossRef][Web of Science][Medline]
  21. O'Campo P, Schempf A. Racial inequalities in preterm delivery: issues in the measurement of psychosocial constructs. Am J Obstet Gynecol (2005) (192(suppl)):S56–63.
  22. Misra DP, O'Campo P, Strobino D. Testing a sociomedical model for preterm delivery. Paediatr Perinat Epidemiol (2001) 15:110–22.[CrossRef][Web of Science][Medline]
  23. Rich-Edwards JW, Buka SL, Brennan RT, et al. Diverging associations of maternal age with low birthweight for black and white mothers. Int J Epidemiol (2003) 32:83–90.[Abstract/Free Full Text]
  24. Reichman NE, Pagnini DL. Maternal age and birth outcomes: data from New Jersey. Fam Plann Perspect (1997) 29:268–72.[CrossRef][Web of Science][Medline]
  25. Acevedo-Garcia D, Lochner KA, Osypuk TL, et al. Future directions in residential segregation and health research: a multilevel approach. Am J Public Health (2003) 93:215–21.[Abstract/Free Full Text]
  26. National Center for Health Statistics. Centers for Disease Control and Prevention. 2000 natality data set. (CD-ROM). (2002) Hyattsville, MD: National Center for Health Statistics.
  27. National Center for Health Statistics. Centers for Disease Control and Prevention. Technical appendix from vital statistics of the United States 2000—natality. (2002) Hyattsville, MD: National Center for Health Statistics.
  28. Massey DS, Fischer MJ. Does rising income bring integration? New results for blacks, Hispanics, and Asians in 1990. Soc Sci Res (1999) 28:316–26.[CrossRef][Web of Science]
  29. Ellen IG. Is segregation bad for your health? The case of low birth weight. In: Brookings-Wharton papers on urban affairs 2000—Gale WG, Pack JR, eds. (2000) Washington, DC: Brookings Institution Press. 203–29.
  30. Acevedo-Garcia D, Soobader M-J, Berkman LF. The differential effect of foreign-born status on low birth weight by race/ethnicity and education. Pediatrics (2005) 115:e20–30.[Abstract/Free Full Text]
  31. Acevedo-Garcia D, Soobader M-J, Berkman LF. Low birthweight among U.S. Hispanic/Latino subgroups: the effect of maternal foreign-born status and education. Soc Sci Med (2007) 65:2503–16.[CrossRef][Web of Science][Medline]
  32. Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care Index and a proposed Adequacy of Prenatal Care Utilization Index. Am J Public Health (1994) 84:1414–20.[Abstract/Free Full Text]
  33. Osypuk TL, Galea S, McArdle N, et al. The distribution of neighborhood poverty and racial disparities in neighborhood context: the unequal American geography of opportunity. (2007) Presented at the annual meeting of the Population Association of America, New York, New York, March30.
  34. Acevedo-Garcia D, Osypuk TL. Impacts of housing and neighborhoods on health: pathways, racial/ethnic disparities, and policy directions. In: Segregation: the rising costs for America—Carr JH, Kutty NK, eds. (2008) New York, NY: Routledge. 197–235.
  35. Williams DR, Collins C. Racial residential segregation: a fundamental cause of racial disparities in health. Public Health Rep (2001) 116:404–16.[Web of Science][Medline]
  36. Subramanian SV, Acevedo-Garcia D, Osypuk TL. Racial residential segregation and geographic heterogeneity in black/white disparity in poor self-rated health in the U.S.: a multilevel statistical analysis. Soc Sci Med (2005) 60:1667–79.[CrossRef][Web of Science][Medline]
  37. Lopez R. Black-white residential segregation and physical activity. Ethn Dis (2006) 16:495–502.[Medline]
  38. Chang V. Racial residential segregation and weight status among US adults. Soc Sci Med (2006) 63:1289–303.[Web of Science][Medline]
  39. Massey DS, White MJ, Phua V-C. The dimensions of segregation revisited. Sociol Methods Res (1996) 25:172–206.[Abstract]
  40. Wilkes R, Iceland J. Hypersegregation in the twenty-first century. Demography (2004) 41:23–36.[CrossRef][Web of Science][Medline]
  41. Raudenbush SW, Bryk AS, Congdon R. HLM for Windows—hierarchical linear and nonlinear modeling. Version 6.01. (2005) Lincolnwood, IL: Scientific Software International.
  42. Hox J. Multilevel analysis: techniques and applications. (2002) Mahwah, NJ: Lawrence Erlbaum Associates.
  43. Charles CZ. The dynamics of racial residential segregation. Annu Rev Sociol (2003) 29:167–207.[CrossRef][Web of Science]
  44. Turner MA, Ross SL, Galster GC, et al. Discrimination in metropolitan housing markets: phase I. (2002) Washington, DC: Office of Policy Development and Research, US Department of Housing and Urban Development.
  45. Acevedo-Garcia D, Osypuk TL, McArdle N, et al. Toward a policy-relevant analysis of geographic and racial/ethnic disparities in child health. Health Aff (2008) 27:321–33.[Abstract/Free Full Text]
  46. Brunner EJ. Toward a new social biology. In: Social epidemiology—Berkman L, Kawachi I, eds. (2000) New York, NY: Oxford University Press. 306–31.
  47. Heck KE, Schoendorf KC, Ventura SJ, et al. Delayed childbearing by education level in the United States, 1969 –1994. Matern Child Health J (1997) 1:81–8.[CrossRef][Medline]
  48. Stevens-Simon C. The weathering hypothesis. (Letter). Am J Public Health (2002) 92:507–8.[Free Full Text]
  49. Rawlings JS, Rawlings VB, Read JA. Prevalence of low birth weight and preterm delivery in relation to the interval between pregnancies among white and black women. N Engl J Med (1995) 332:69–74.[Abstract/Free Full Text]
  50. Keeton K, Hayward RA. Pregnancy intention and birth outcomes: does the relationship differ by age or race? J Womens Health (Larchmt) (2007) 16:510–16.[CrossRef][Medline]
  51. Rose G. The strategy of preventive medicine. (1992) New York, NY: Oxford University Press.
  52. Rose G. Sick individuals and sick populations. Int J Epidemiol (1985) 14:32–8.[Abstract/Free Full Text]
  53. Schwartz S, Diez-Roux AV. Commentary: causes of incidence and causes of cases—a Durkheimian perspective on Rose. Int J Epidemiol (2001) 30:435–9.[Free Full Text]
  54. Behrman RE, Butler AS. Committee on Understanding Premature Birth and Assuring Healthy Outcomes. Sociodemographic and community factors contributing to preterm birth. In: Preterm birth: causes, consequences, and prevention—Behrman RE, Butler AS, eds. (2006) Washington, DC: National Academies Press. 111–28.
  55. Cutler DM, Glaeser EL. Are ghettos good or bad? Q J Econ (1997) 112:827–72.[CrossRef][Web of Science]
  56. Bureau of the Census. US Department of Commerce. Migration for the population 5 years and over for the United States, regions, states, counties, New England minor civil divisions, metropolitan areas, and Puerto Rico: 2000 (PHC-T-22). Table 1. (2003) Bethesda, MD: US Department of Commerce.
  57. David R. Commentary: birthweights and bell curves. Int J Epidemiol (2001) 30:1241–3.[Free Full Text]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Epidemiol RevHome page
M. R. Kramer and C. R. Hogue
Is Segregation Bad for Your Health?
Epidemiol. Rev., June 9, 2009; (2009) mxp001v2.
[Abstract] [Full Text] [PDF]

Home page
 Am J EpidemiolHome page
D. Acevedo-Garcia and T. L. Osypuk
Invited Commentary: Residential Segregation and Health--The Complexity of Modeling Separate Social Contexts
Am. J. Epidemiol., December 1, 2008; 168(11): 1255 - 1258.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
167/11/1295    most recent
kwn043v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (3)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Osypuk, T. L.
Right arrow Articles by Acevedo-Garcia, D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Osypuk, T. L.
Right arrow Articles by Acevedo-Garcia, D.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?