|Home | About | Journals | Submit | Contact Us | Français|
In the United States, Black persons are disproportionately affected by sexually transmitted infections (STIs), including gonorrhea. Individual behaviors do not fully explain these racial disparities. We explored the association of racial residential segregation with gonorrhea rates among Black persons and hypothesized that specific dimensions of segregation would be associated with gonorrhea rates.
We used 2003 to 2007 national STI surveillance data and 2000 US Census Bureau data to examine associations of 5 dimensions of racial residential segregation and a composite measure of hypersegregation with gonorrhea rates among Black persons in 257 metropolitan statistical areas, overall and by sex and age. We calculated adjusted rate ratios with generalized estimating equations.
Isolation and unevenness were significantly associated with gonor-rhea rates. Centralization was marginally associated with gonorrhea. Isolation was more strongly associated with gonorrhea among the younger age groups. Concentration, clustering, and hypersegregation were not associated with gonorrhea.
Certain dimensions of segregation are important in understanding STI risk among US Black persons. Interventions to reduce sexual risk may need to account for racial residential segregation to maximize effectiveness and reduce existent racial disparities.
Sexually transmitted infections (STIs) remain an important public health problem in the United States, with approximately 19 million new infections per year.1 Black persons, especially adolescents, bear a disproportionate burden of most STIs, including gonorrhea.2–6 In 2008, rates of gonorrhea were highest among Black individuals, aged 15 to 19 and 20 to 24 years, compared with any other racial/ethnic and age groups.7,8 Among 15- to 19-year-old adolescents, rates of gonorrhea were nearly 21 times higher for Black (2201.9 per 100 000) than for White adolescents (107.0 per 100 000).7 Untreated gonorrhea can have serious and long-term sequelae, including the facilitation of HIV transmission, infertility, and adverse outcomes for infants born to infected mothers.8
Exposure to and infection with STIs are conditioned by many factors, including individual behaviors, relationship patterns, and characteristics of the social environment. Substantial attention has been paid to differences in individual risk behaviors, such as condom use and number of sexual partners, but they do not fully explain racial disparities in STI risk.4 Therefore, focusing solely on these proximate factors to reduce risk and disparities may have only limited effect.4,9 A growing body of research has examined the contribution of contextual factors, such as neighborhood attributes, to sexual risk. Specifically, numerous studies have examined whether living in a neighborhood with lower socioeconomic status is associated with sexual risk behaviors, such as younger age at first sexual intercourse and unprotected sexual intercourse.10–20 The findings have been equivocal, with some showing an association and others not. Therefore, a better understanding of the possible effects of other contextual factors on sexual risk is necessary.
Racial residential segregation—the extent to which 2 or more racial groups live separate from one another in a metropolitan area—is a characteristic of the social environment that many Black individuals continue to experience.21 Nearly two thirds of Black persons live in highly segregated areas.9 The available evidence suggests that Black individuals living in more segregated areas, compared with less segregated areas, are at higher risk for certain poor health outcomes, such as low birth weight, mental health conditions, and mortality.9,22–28 No published studies to date have examined the association of racial residential segregation with sexual risk, but recent commentary has identified racial residential segregation as a possible cause of disparities in sexual risk.5,9,22,29–31
Racial residential segregation, which describes the racial composition of neighborhoods and the spatial distribution of these neighborhoods in larger metropolitan areas, may be more conceptually relevant to understanding racial disparities than are individual and neighborhood characteristics because it captures the unequal structure for Black and White people across the entire housing market. It has been conceptualized in 5 distinct dimensions—exposure, concentration, centralization, clustering, and unevenness. Metropolitan areas are defined by
Racial residential segregation is hypothesized to lead to differential exposure to STIs through a variety of mechanisms. First, segregation might lead to increased rates of STIs among Black persons by affecting the sexual network (e.g., partner availability and density of individuals).5,31 Second, segregation may create or foster environments (e.g., restricted economic and employment opportunities, disordered neighborhoods) that are conducive to sexual risk behaviors and increased STI risk.9,24,29,31 Each dimension of segregation may have varying degrees of salience in describing distinct mechanisms that affect sexual risk and STI transmission.22,24,31According to a conceptual model proposed by Acevedo-Garcia,22 exposure, concentration, and, to a lesser extent, centralization are relevant to understanding infectious disease risk because of their effect on transmission patterns and social networks.
We used 5 years of national sexually transmitted disease (STD) surveillance data to study the associations of racial residential segregation with gonorrhea rates among Black people in the United States at the metropolitan statistical area (MSA) level. According to Acevedo-Garcia's model, we hypothesized that certain dimensions of segregation, such as exposure and concentration, would be more positively associated with gonorrhea rates compared with other dimensions of segregation. Additionally, we hypothesized that the associations would be modified by sex and age because of differences across sex and age groups in patterns of social influence.
Racial residential segregation is measured across 2 different geographic scales, areas representing neighborhoods and larger metropolitan areas.33,34 The smaller scale should represent neighborhoods, and the larger metropolitan area should approximate labor and housing markets.21 We used census tracts, which are geographic subdivisions created by the US Census Bureau and intended to represent homogeneous neighborhoods,35 and MSAs, which are census-defined subdivisions that consist of at least 1 core urban area, the associated counties, and adjacent counties with a high degree of social and economic ties to the urban core(s).35
Data on number of gonorrhea cases for all MSAs in the United States for 5 consecutive years (2003–2007) were provided by special request by the Statistics and Data Management Branch, Division of STD Prevention, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention at the Centers for Disease Control and Prevention. Gonorrhea is a national notifiable disease, and data are reported by all US states and territories. During 2003 to 2007, 1 535 464 cases of gonorrhea were reported in 366 MSAs. Number of cases and total population were provided by race/ethnicity, 5-year age groups, and sex. MSA population estimates also were obtained from the CDC's Centers for Disease Control and Prevention's surveillance system, which incorporates population estimates from the US Census Bureau. The STD data and population estimates are described in more detail in the annual STD surveillance report.7 Only participants classified as Black, non-Hispanic were included for analysis (n = 830 130). Additionally, participants for whom complete information on age and sex could not be determined were excluded (2241; 0.3%).
We assessed racial residential segregation with indexes identified by Massey and Denton32 as best representing each dimension of segregation (Table 1).36 Values for each index were obtained from the US Census Bureau Housing and Household Economic Statistics Division, which used US Census Bureau 2000 data to calculate these indexes.
The isolation index, a measure of exposure, sums the proportion of Black residents in each neighborhood, weighted by the proportion of Black individuals in the metropolitan area who live in that neighborhood. We refer to the exposure dimension by its converse label, isolation, because higher levels of isolation represent higher levels of segregation.
The relative concentration index enumerates the amount of physical space occupied by Black people in the metropolitan area relative to the space occupied by White people and then compares this value with the ratio that would be obtained if Black persons were completely concentrated and White persons were completely scattered.
The absolute centralization index represents the distance that Black residents live from the center of the MSA rather than outlying areas by comparing Black individuals’ distribution around the MSA center with their distribution in outlying areas.
Clustering was measured with the index of spatial proximity. This index first enumerates the average proximity of Black people to one another and the average proximity of Black to White individuals. It then takes the average of these values, weighted by the proportion of Black and White persons in the metropolitan area, to indicate the extent to which Black people live nearer to one another than they live to White people.
Unevenness was measured with the dissimilarity index, which sums the weighted difference of the proportion of Black residents in each neighborhood from the proportion of Black residents in the MSA and then compares this value with the theoretical value under complete segregation. It reflects the proportion of Black individuals who would have to change neighborhoods to achieve an even distribution in the metropolitan area. (For further details on these indexes, including formulas, see http://www.census.gov/hhes/www/housing/housing_patterns/app_b.html.)
To account for potentially nonlinear associations with gonorrhea rates, we used established cutpoints for the indexes of segregation: very low, low, moderate, and high. These cutpoints have been established previously as meaningfully distinguishing between levels of segregation and being relevant across indexes.25,32,37 Because of small cell sizes for the very low category for some indexes, we combined very low and low. See Table 1 for ranges and cutpoints for each index. Metropolitan areas that were highly segregated on at least 4 dimensions were considered hyper-segregated and were compared with areas that were not highly segregated on 4 or 5 dimensions (i.e., not hypersegregated).27,36
We obtained MSA-level sociodemographic data that were hypothesized to be potential confounders from the US Census Bureau 2000 Census Summary 1 and Summary 3 Files.38,39 Specifically, census region (Northeast, Midwest, West, and South), population size (log), and population density (people per square mile of land area) were considered.
Additionally, we controlled for racial composition (percentage Black) because we wanted to determine whether the spatial organization of Black people compared with White people was associated with gonorrhea rather than simply the racial composition of the areas. Because of the concern that the effect of racial residential segregation on health may simply represent socioeconomic differences, we also included MSA-level socioeconomic status, a marker for occupational markets, housing markets, and economic status of the MSA, according to the socioeconomic position index developed by Krieger et al.40 In brief, data on percentage working class, percentage unemployed, percentage of houses worth more than $300 000 (reverse coded), median household income (reverse coded), percentage of persons with income below the federal poverty level, and percentage with less than high-school education were obtained for all individuals in the MSAs. We calculated and summed z scores for each measure to create a socioeconomic position index, with higher scores indicating higher disadvantage.
The independent variables were linked to gonorrhea cases by MSA of residence of participants. We used 2008 MSA definitions to categorize cases into MSAs and 2003 MSA definitions to calculate segregation measures. Data for 8 newly defined MSAs in 2008 compared with 2003 were excluded (n = 3688). Additionally, to remove unreliable estimates, we excluded MSAs that had a population less than 100 000 and a Black population of less than 5000 (n = 6670),21,25,27,28,41 resulting in a final sample of 817 531 gonorrhea cases in 257 MSAs.
We used generalized estimating equations (GEE) with a Poisson distribution to examine associations of segregation measures and covariates with 5-year average gonorrhea rates. Analyses were standardized by the size of the Black population at risk, accounting for the greater precision of rates in larger populations and directly addressing the potential problem of heterogeneity of error variance.42 We used GEE to account for variance clustering that may occur when one level of data (individuals) is nested within another (MSAs).43 We specified an exchangeable correlation structure and used robust variance estimators to calculate 95% confidence intervals (CIs) around each rate ratio (RR).44
We first developed a series of unadjusted models to separately examine the association of each segregation measure with gonorrhea rates among Black persons. Second, we assessed adjusted associations by including MSA-level covariates. Finally, we tested sex and age modification by including interaction terms and created sex- and age-specific models. To assess overall associations and interaction effects, likelihood ratio tests were executed, and P values were reported. Although the MSAs represent a census, the P values were interpreted as the probability that the findings were the result of chance. The data points that were organized by MSA do not represent a census, and we made inferences across unsampled time points; therefore, uncertainty in point estimates remained.
Statistical analyses were conducted with SAS version 9.1.3 (SAS Institute, Cary, NC).
The final sample included 817 531 cases of gonorrhea among Black, non-Hispanic individuals from 2003 to 2007 in 257 MSAs. The 5-year average gonorrhea rate was 502.1 per 100 000. Among Black, non-Hispanic case participants, across the 257 MSAs, 46% were female, 28% were 19 years old or younger, and 59% were 20 to 34 years old.
In absolute terms, approximately half of the MSAs were located in the South, followed by the Midwest, West, and Northeast. On average, the total population of each MSA was 842 000, with 322.3 people per square mile. Additionally, on average, 13.0% of the population in each MSA were Black, non-Hispanic, and the socioeconomic position index was –0.08 (Table 2).
The majority of MSAs were highly concentrated and centralized (56.4% and 71.2%, respectively) and moderately uneven (57.2%) (Table 2). Conversely, 93.4% of the MSAs had low levels of clustering, and 55.3% had low levels of isolation (Table 2). However, accounting for the size of the Black population shifted a larger proportion of MSAs to be more segregated across all measures. Moreover, after weighting by the size of the Black population, the mean isolation index indicated that the average Black adult or child lived in a census tract in which 57% of the tract population was Black. Seven percent of MSAs were hypersegregated (Table 2); 19.1% of MSAs were highly segregated on no dimensions, 24.5% on 1 dimension, 32.3% on 2 dimensions, 17.1% on 3 dimensions, and 7.0% on 4 or 5 dimensions.
In unadjusted analyses, isolation was associated with gonorrhea rates (P = .001). Compared with low levels of isolation, moderate levels of isolation were significantly associated with higher gonorrhea rates (RR = 1.31; 95% CI = 1.11, 1.56) among Black people. After adjusting for MSA sociodemographic characteristics, high (RR = 1.27; 95% CI = 0.99, 1.62) and moderate (RR = 1.55; 95% CI = 1.30, 1.85) levels of isolation were associated with higher gonorrhea rates (Table 3).
In unadjusted analyses, centralization was not associated with gonorrhea rates. However, in adjusted analyses, centralization was marginally associated with gonorrhea (P = .09). Specifically, high levels of centralization were significantly associated with higher gonorrhea rates (RR = 1.28; 95% CI = 1.05, 1.55) (Table 3).
In unadjusted analyses, clustering was associated with gonorrhea rates (P = .05). A high level of clustering was associated with higher gonorrhea rates (RR = 1.23; 95% CI = 1.04, 1.44). Conversely, a moderate level of clustering was inversely associated with gonorrhea rates (moderate vs low RR = 0.61; 95% CI = 0.44, 0.84). However, in adjusted analyses, these associations were not significant (Table 3).
In unadjusted analyses, unevenness was not associated with gonorrhea rates. However, in adjusted analyses, unevenness was associated with gonorrhea (P = .009). High (RR = 1.29; 95% CI = 1.01, 1.64) and moderate (RR = 1.32; 95% CI = 1.10, 1.58) levels of unevenness were associated with higher gonorrhea rates (Table 3).
Concentration and hypersegregation were not associated with gonorrhea rates (Table 3).
In sex-stratified analyses, high centralization appeared to be more strongly associated with gonorrhea rates among males (RR = 1.38; 95% CI = 1.13, 1.69) compared with females (RR = 1.17; 95% CI = 0.96, 1.44; Table 3). However, the sex–centralization interaction was not statistically significant (P = .12). The associations of isolation, concentration, clustering, unevenness, and hypersegregation with gonorrhea rates were not modified by sex (Table 3).
The association of isolation with gonorrhea rates differed by age (P < .001). Isolation was associated with gonorrhea rates among all age groups, but the magnitude of the effect was larger among those aged 19 years or younger and those aged 20 to 34 years (Table 3). The associations of concentration, centralization, clustering, unevenness, and hyper-segregation with gonorrhea rates were not modified by age (Table 3).
This study was the first to our knowledge to empirically examine the association of racial residential segregation with gonorrhea rates, the second most common STI, among Black persons in the United States. Isolation and unevenness and, to a lesser extent, centralization predicted gonorrhea rates, whereas other dimensions (i.e., concentration and clustering) did not. These findings partially coincide with Acevedo-Garcia's conceptual model, describing how segregation may affect infectious disease risk.22
The isolation index, which measures the extent to which Black people are exposed in large part to only other Black people, was associated with gonorrhea rates, and this association was modified by age. Conceptually, high isolation of Black persons from White persons may lead to increased rates of STIs among Black persons because those who are likely to acquire infection are sexual contacts of those already infected (i.e., those exposed to the infectious agent). Studies examining sexual networks and gonorrhea infection have found that high sexual isolation (i.e., the tendency of individuals to select sexual partners among members of the same group) maintains the presence of groups within which transmission occurs.22,45 Because individuals tend to select sexual partners in the area in which they live (i.e., spatial assortativity),46 sexual isolation may be affected by geographic isolation. As a result, isolation of infected and susceptible persons in a geographic area will increase the likelihood that core network members, or individuals at high risk for infection, will have sexual intercourse and spread the infection to individuals with few sexual partners within that area.5,22,31,46,47
Isolation may increase the risk for gonorrhea through its effect on social factors as well. Social norms—which also can be transmitted—have been shown to be associated with sexual risk among adolescents and among Black women.48–50 In isolated communities, within-group norms for risky sexual behavior might be strengthened.25 Age moderated the isolation–gonorrhea rate association; the association was stronger among the younger age groups. Because they are experiencing major transitions, the social environment may confer more influence for younger individuals.51
We hypothesized that concentration of Black individuals in more densely settled areas may lead to higher rates of STI transmission among Black individuals because of denser sexual networks (i.e., more interconnectedness between sexual contacts)46 and by concentrating economic and social disadvantage,9,24,29,31,37,52,53 but it was not associated with gonorrhea rates among Black individuals in our analysis.
Centralization—the tendency of Black residents to live nearer to the center of the MSA relative to White residents—was marginally associated with gonorrhea rates overall. High centralization, which may be an indicator for crowding, poor neighborhood environment, and, consequently, a lack of social control, can affect individual behaviors by encouraging risky behaviors, including crime, drug use, and risky sexual behaviors.9,22,25,31,54 This risky behavior may lead to higher rates of transmission of STIs among those confined to living in these areas. Several studies have found an association of neighborhood physical environment with gonorrhea rates in US cities.31,55,56
We did not find any evidence of an association between clustering and gonorrhea rates, which was consistent with Acevedo-Garcia's model. However, in contrast to the model, we did find that unevenness was associated with gonorrhea rates. Unevenness is conceptually related to isolation in that if an area is highly uneven, Black people are not distributed evenly across neighborhoods in an MSA and live in separate neighborhoods. Therefore, in highly uneven areas, Black persons live on average in neighborhoods with a high percentage of Black residents, resulting in Black people being isolated from White people. Additionally, isolation and unevenness capture a summary of the racial composition of neighborhoods in a metropolitan area, whereas the other dimensions capture the spatial distribution of these neighborhoods. Therefore, isolation and unevenness may be functioning similarly in this context. Interestingly, these dimensions are the most commonly tested in the health literature. Although the research on the segregation–sexual risk associations is still too nascent to exclude other mechanisms, this may suggest that testing these dimensions may be sufficient in the future.
Hypersegregation, which assumes that all dimensions are equally important, was not associated with gonorrhea rates. However, as shown by the results, certain dimensions were not associated with gonorrhea, so this measure may not be a meaningful summary of elevated risk in this context.
These associations of isolation, centralization, and unevenness with gonorrhea rates remained after adjusting for MSA-level covariates, including region, population size, density, socioeconomic position, and racial composition. This suggests that segregation is associated with gonorrhea rates above and beyond its effect on MSA socioeconomic position. Socioeconomic position was positively associated with gonor-rhea (RR = 1.05; 95% CI = 1.02, 1.08), and even though it may be a partial mediator, it did not completely explain the association between segregation and gonorrhea rates. This finding is supported by previous studies.24 Additionally, although neighborhood racial composition has been shown to be associated with rates of gonorrhea,57 in our analyses, MSA racial composition was not associated with gonorrhea among Black individuals, and models without MSA racial composition did not differ substantially from models with racial composition, suggesting that the observed associations were not artifacts of racial composition. This finding is supported by previous research among injection drug users58 and may indicate that the racial composition of neighborhoods and their spatial organization explain disparities in risk rather than simply the racial composition of the entire MSA.
Although this novel study was the first to use national data to directly assess the association of racial residential segregation with sexual risk in the United States, it had limitations. First, gonorrhea infections can be asymptomatic and therefore undiagnosed, and screening practices to detect infections may vary by gender and geography.8,59 Detected cases also may be underreported. For example, compared with White persons, minority persons more often seek STI care in public clinics, which may be more likely to report cases than would private providers.59 If differential diagnosis and reporting are also related to MSA racial residential segregation, then bias could be introduced. Because of our exclusion criteria, our sample may not be representative of all metropolitan areas in the United States. Additionally, this study examined only MSA-level covariates and did not include neighborhood-or individual-level measures of risk, which might help elucidate the mechanisms and account for potential confounding.24 Even though associations were seen even after adjusting for MSA-level variables, we may have underestimated the real effect because some of these variables (e.g., socioeconomic position) may be on the causal pathway.28 Finally, numerous comparisons were made, which could have resulted in significant findings by chance. However, because these findings were relatively consistent across strata of sex and age, these findings, taken as a whole, are not likely the result of chance alone.
Although this study was limited to Black people, the findings suggest that racial residential segregation, a characteristic that differentially affects Black and White persons, may help to explain the large racial disparity in gonorrhea rates, and future studies should aim to directly test this. Additionally, our findings suggest that interventions to reduce sexual risk that account for racial residential segregation may be needed. Interventions may have to be adapted in highly segregated communities to account for the social context that put individuals at risk for these infections. For example, an intervention to reduce STI prevalence among Black individuals living in segregated areas may include a focus on empowering individuals to make systemic changes in their community that aim to reduce racial residential segregation. Although structural interventions that directly address the factors that created or perpetuate racial residential segregation—such as programs that aim to reduce unfair housing policies and discriminatory practices in the housing market and increase affordable housing60,61—are often seen as distal to health outcomes, they could have broad benefit by affecting sexual risk and reducing health disparities as well as other negative health, social, and economic outcomes.30
This work was supported by the National Institute of Mental Health at the National Institutes of Health (grants T32MH020031 and P30MH062294).
Human Participant Protection
This study was granted an exemption by the Human Subjects Committee of Yale University.
About the Authors
At the time of the study, Katie B. Biello, Trace Kershaw, Jeannette Ickovics, and Linda Niccolai were with Yale School of Public Health and Center for Interdisciplinary Research on AIDS, Yale University, New Haven, CT. Robert Nelson and Matthew Hogben are with the Division of STD Prevention, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, Centers for Disease Control and Prevention, Atlanta, GA.
K. B. Biello conceptualized the study, analyzed the data, and was primarily responsible for writing the article. T. Kershaw, J. Ickovics, and L. Niccolai assisted in conceptualizing the study, assisted in interpreting the data, and participated in revising the article. R. Nelson and M. Hogben assisted in obtaining and analyzing the data and participated in revising the article.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Mental Health or the National Institutes of Health.