The epidemiology of sexually transmitted disease (STD) in the United States is characterized by immense racial, social, economic, and geographic inequality. In the 2010 annual summary of STD surveillance, the Centers for Disease Control and Prevention (CDC) reported that national rate of Neisseria gonorrhoeae infection was approximately 19 times higher in blacks than in whites.1 The residential segregation of black populations, often in areas of high-economic disadvantage and low social status, may play a crucial role in these observed racial inequities.2,3 Despite significant social changes over the last half century, most metropolitan cities in the United States remain extremely segregated along racial and economic lines.4–6
Residential segregation, especially on the dimensions of concentration and isolation, is believed to be conducive to the spread of infectious diseases.7,8 Sexually transmitted diseases such as gonorrhea are inherently social diseases, surviving and proliferating through continued interactions between individuals in a social group.9 Residential segregation creates distinct social networks with little crossover among them. Although this might inhibit the spread of disease outside these networks, it may also perpetuate the persistence of endemically high rates within them. Therefore, residential segregation is likely a key component in the endemically high rates of STDs observed among socially disadvantaged black populations.
A few previous ecological studies have indicated that racial segregation is associated with rates of syphilis, gonorrhea, and chlamydia.10–12 However, there are several limitations to the existing literature regarding segregation and STDs. Most previous studies have used relatively simple measures of residential segregation, such as racial composition.11,12 Such measures are generally considered a poor proxy for segregation because they fail to fully capture the spatial distribution of racial settlement patterns.13,14 More sophisticated measures such as the isolation index are generally believed to be better at measuring the actual degree of segregation across a defined geographic area.15 In addition, many previous studies have assessed segregation measures at the county or city level.10,11 This may introduce bias and lead to mis-estimation of segregation effects because important trends in segregation may exist only when counties are considered in context. Residential segregation tends to be a metropolitan area phenomenon, with the largest disparities in income and racial residential patterns observed between central cities and their outlying suburbs.16,17 For example, in most metropolitan areas, blacks are substantially more likely than whites to reside in the central city than in the surrounding suburban counties. Therefore, the metropolitan statistical area (MSA) is arguably a better geographic context for the study of segregation effects on health.
Finally, research considering the interaction between economic and racial segregation is still largely lacking, despite evidence that both are important predictors of neighborhood settlement patterns and likely of STD rates.17,18 The purpose of this study was therefore to address these 3 issues by performing an ecological analysis to examine the influence of racial and economic residential segregation, both independently and in combination, on gonorrhea rates in US metropolitan areas. Our hypothesis was that higher degrees of residential racial and economic segregation are independently associated with higher gonorrhea rates in metropolitan areas and that MSAs that are highly segregated across both of these dimensions experience higher rates of gonorrhea than do areas segregated along only one.
MATERIALS AND METHODS
This was a cross-sectional, ecological study comprising US MSAs with total populations of 100,000 or more and black populations of at least 5000. In addition to ensuring sufficient data per unit for analysis, this strategy for selecting MSAs was comparable with that applied by previous studies of black segregation and health.19,20 Metropolitan statistical areas are constructed by the Office of Management and Budget to define counties (i.e., suburbs) clustered around a central city.21
Data on annual rates of gonorrhea from 2005 to 2009 for each MSA were obtained from the CDC. The CDC receives morbidity data for notifiable infectious diseases through regular reporting from state and local disease surveillance systems.1 The primary outcome measure of interest was the 5-year (2005–2009) average gonorrhea incidence rate at the MSA level per 100,000 person-years.
Measures of racial and economic residential segregation for each MSA were extracted from the US Census Bureau’s American Community Survey 2005–2009 5-year estimates.22 Racial segregation was assessed using the isolation index, whereas economic segregation was estimated using the Gini index of household income inequality. The isolation index measures between-group contact, or the extent to which members of a minority population are exposed to other members of the same minority population.23 Specifically, the degree of black isolation relative to whites was the focus because black isolation may be better suited to capturing patterns in infectious disease risk than many other measures of distribution.7,10,14 That is, black individuals may be more likely to be exposed to disease in the course of normal sexual behavior because of the high burden of disease in racially segregated social networks. Both the black isolation index and the Gini index of household income inequality range from 0 to 1.0, with higher values indicating a greater degree of black isolation and greater income disparity (as opposed to all households having equal shares of income), respectively.24,25
Metropolitan statistical areas were classified as either high or low segregation based on the median distributions of the data. This categorization was made without regard to the outcome measure to avoid biasing the resulting estimates.26 Dichotomizing our exposure measures allowed us to examine the distribution of gonorrhea rates among areas with high versus low levels of segregation and permitted us to maximize the contrast between these 2 exposure groups while conserving statistical power.
Several measures pertaining to MSA-level sociodemographic and economic indicators were also extracted from the 2005 to 2009 American Community Survey data and were assessed for potentially confounding effects on the association between segregation and gonorrhea rates. These included the following: median age, racial composition (percentage of the population that is black), education (percentage of population aged ≥25 years with less than high school education), population density (people per square kilometer), poverty (percentage of people living below the federal poverty level), median annual household income, unemployment (percentage of civilian labor force ≥16 years currently unemployed), marital status (percentage of population ≥15 currently married), and the percent of female-headed households. Similar contextual covariates have been assessed in previous segregation studies and have been associated with both gonorrhea rates and segregation.1,10,12 All continuous variables were dichotomously categorized according to the median value of their distribution. Geographic region (northeast, midwest, south, and west) was also considered a potential confounder because high gonorrhea rates tend to be concentrated in the south, and patterns of residential segregation also vary substantially by region.1,5
We used logistic regression modeling to produce estimates of odds ratios (ORs) and 95% confidence intervals (95% CIs) to estimate the relationship between gonorrhea rates at the MSA level and the 2 segregation measures of interest (black isolation index and Gini index). To assess the potential for confounding, preliminary analyses were conducted to assess bivariate associations between the 2 segregation indices, other contextual variables, and gonorrhea rates. Confounding was evaluated using an iterative process evaluating the impact of covariates on the estimates of effect. Covariates whose addition changed the estimate of effect more than 10% were retained in the model. The results of these preliminary analyses were used to inform the model-building process and determine which selection of variables to consider for inclusion in adjusted models. Initially, separate regression models were run for each determinant to examine whether each was independently associated with gonorrhea rates after adjusting for other MSA-level socioeconomic factors identified as potential confounders.
Effect measure modification was assessed by evaluating departures from additivity and multiplicativity.26,27 First, we included 3 dummy variables in the model to capture the combined black isolation and Gini indices with low black isolation and low Gini indices as the reference category. From this model, we estimated the relative excess risk due to interaction for the adjusted OR (RERI-OR).27 Next we included a product term for the interaction between the black isolation and Gini indices and assessed the resulting coefficient to determine departure from a multiplicative model. We hypothesized that the combination of racial and economic segregation would result in a positive departure from additivity and multiplicativity. That is, we expected that income segregation modifies the relationship between racial segregation and STD rates, and vice versa, such that areas that are highly segregated both racially and economically experience significantly higher gonorrhea incidence than do areas that are highly segregated along just one of these dimensions.
To supplement the validity of this categorical approach, we also conducted an analysis of these data using both Poisson and linear regression to examine the association between our segregation measures and gonorrhea rates while treating all numerical covariates as continuous (covariates with nonnormal distributions were log transformed).
A total of 277 MSAs were included in this analysis, 138 of which were categorized as having high gonorrhea rates (above the median value of 127.6 cases per 100,000) for the 2005 to 2009 period. Gonorrhea rates in all MSAs ranged from 12.3 to 466.4 cases per 100,000 person-years, with higher rates observed among MSAs in the southern and midwestern regions of the United States. The average (SD) black isolation index across all MSAs was 0.30 (0.18), whereas the average (SD) Gini index was 0.45 (0.02).
Among MSAs with high gonorrhea rates, 79.7% and 63.0% were also categorized as having high black isolation indices and Gini indices, respectively. Conversely, among low-gonorrhea-rate MSAs, only 20.9% and 37.4% were categorized as having high isolation and Gini indices. The distribution of other MSA characteristics by low and high gonorrhea rates are summarized in Table 1. Metropolitan statistical areas with high gonorrhea rates had higher proportions of most indicators of socioeconomic disadvantage relative to MSAs with low rates. For example, the proportion of MSAs characterized by more than 24.9% of households headed by a female was 77.5% compared with 23.0% among MSAs with high and low gonorrhea rates, respectively.
Metropolitan statistical areas with a high black isolation index had an approximately 15-fold increased odds of high gonorrhea rates (crude OR, 14.90; 95% CI, 8.32–26.69). The crude odds of high gonorrhea rates were 2.85 times greater for MSAs with high Gini index compared with those with a low Gini index (95% CI, 1.75–4.65). As part of the preliminary analysis, logistic regression models were run separately with black isolation index and Gini index as the primary predictors of high gonorrhea rates, adjusting for potential confounders (Table 2). After adjustment, the strength of the association between black isolation index and gonorrhea rates was reduced, dropping from a crude OR of 14.90 to an adjusted OR (AOR) of 3.37 (95% CI, 1.23–9.21). The largest contributors to this dilution of effect were the following 3 measures: percent of the population that was black, percent of female-headed households, and the percent living under the federal poverty level. Adjustment for potential confounders only slightly impacted the association between the Gini index and gonorrhea (AOR, 1.54; 95% CI, 0.74–3.24), but the association was not statically significant after adjustment.
Among MSAs with a combination of high black isolation and high Gini indices, the AOR of high gonorrhea rates was 5.46 (95% CI, 1.72–17.31) compared with those with a combination of low levels of black isolation and low Gini index (Table 3). The adjusted odds of high gonorrhea rates among MSAs with a high black isolation and low Gini index were 3.12 (95% CI, 0.98–9.98), whereas that for MSAs with a low black isolation and high Gini index were 1.25 (95% CI, 0.42–3.77). We did not find evidence of departure from additivity of effects: the joint effect (adjusted RERI-OR, 2.09; 95% CI, −3.30 to 7.48) was on the order of the effect of a high Gini index alone (AOR, 1.44). In contrast, the adjusted RERI-OR was smaller than the AOR of 3.11 when looking at black isolation alone.
Similarly, when we included a product term in the model representing the interaction between black isolation and Gini indices, there was no evidence of departure from a multiplicative model (coefficient = −0.17, Wald χ 2 = 0.19, P = 0.66). That is, MSAs with a combination of racially segregated residential patterns and high levels of income inequality had marginally increased likelihoods of also having high gonorrhea rates, but this association was not significant. The expected AOR for the joint effect of high isolation and high Gini indices, assuming no interaction effect, was 3.90, slightly lower than the observed AOR of 5.46. However, a high black isolation index remained the strongest predictor of high gonorrhea rates.
The substantive findings reported previously held when model covariates were treated as continuous. Using a Poisson regression approach, higher black isolation was associated with higher gonorrhea rates (adjusted rate ratio, 1.17; 95% CI, 1.13–1.21), as was higher household income inequality (adjusted rate ratio, 1.04; 95% CI, 1.02–1.06). Similarly, linear regression modeling indicated that the relationship between the segregation indices and gonorrhea rates remained positively linear. For example, black isolation accounted for 75% of the variance in gonorrhea rates (adjusted coefficient = 143.4, P < 0.001). There remained no evidence of interaction between the 2 segregation measures when an interaction term was included in the adjusted regression models (P = 0.87).
This study sought to examine the influence of racial and economic residential segregation on gonorrhea rates in US metropolitan areas. We hypothesized that higher degrees of residential racial and economic segregation were associated with higher gonorrhea rates in metropolitan areas and that these 2 phenomena may interact such that the combined influence of high levels of segregation across multiple dimensions would be greater than the influence of any one dimension alone. The results of our analysis only partially supported this hypothesis. After adjustment to control for the potentially confounding contribution of various social structures related to race and social class, we found an association between high black isolation and Gini indices with high gonorrhea rates in MSAs, when each was considered individually, but we did not find evidence of additive or multiplicative effect due to the combination of these 2 indices. A high black isolation index was the strongest predictor of high gonorrhea rates.
The percent of a population that is black is strongly predictive of area gonorrhea rates, primarily because rates of many STDs are so much higher among black populations.1,28 However, we must still elucidate the factors that contribute to high rates among black populations. Residential segregation is one of the mechanisms that may contribute to creating and sustaining endemically high rates of disease among black populations. Our study supports previous work indicating that segregation is a more important predictor of high rates of disease than the percentage of blacks in a population.10 That is, it is not necessarily an MSA’s racial composition that leads to higher rates of disease but, rather, the degree to which black populations are inequitably distributed within larger urban areas.
Furthermore, we found that racial segregation, as measured by black isolation, was a larger driver of differences in gonorrhea rates than was income inequality. Although alternate measures of income segregation, such as black-white income disparity, might yield different results, it is also possible that racial segregation is more relevant to perpetuating disparities in STD rates than is income inequality. If such is the case, reform within the health care system may not be adequate to effectively reduce inequities in disease rates because it does not address the social conditions and structures that place some populations at higher risk for infection. However, an understanding of the role of isolation in the perpetuation of high gonorrhea rates could be combined with targeted, sustained screening and treatment of disease to lower the overall burden of disease in socially isolated sexual networks.
These findings do not mean that economic conditions are not relevant to STD epidemiology. The proportion of female-headed households and proportion of the population living below the poverty level were notable confounders in the association between racial segregation and gonorrhea rates, and their individual contribution to gonorrhea rates should not be ignored. However, our results support previous work indicating that a broad approach addressing both economic and social determinants is necessary to alleviate disease disparities. Tackling such geographically based social disparities is not a simple proposition, but it may be necessary to effect significant change.
This study addresses many issues inherent in the previous literature regarding segregation and STDs, including using MSAs as the unit of analysis and the assessment of racial residential segregation using the black isolation index, which studies suggest may better at capturing patterns of unhealthy exposures and infectious disease risk than other measures of racial residential distribution.7,10,14 The ecological nature of this study is also a strength in the context of the phenomenon under study because the objective was to examine population-level determinants of disease rates. In general, individual-level analysis may be of limited use in the study of STD epidemiology because the risk of infection often depends more on the characteristics of populations and the social environment, rather than individual behaviors or characteristics.29–31
Although our study has several methodological strengths, our findings must be interpreted with several limitations in mind. Our dichotomization of both outcome and determinant measures, although done in such a manner as to avoid bias, may have obscured the true association between these factors. However, the results of sensitivity analyses indicated that neither applying alternative parameterizations or modeling techniques nor performing the analysis treating variables as continuous covariates altered our substantive findings. In addition, the Gini index of income inequality may not be the best measure to capture the impact of income segregation on STD rates. Although concentrated economic disadvantage and poverty are known predictors of high STD rates, there is evidence that the degree of black-white income inequality may be pertinent in addition to the distribution of income within a given area.10 Future studies should assess the use of such a measure.
Our findings lend further credence to the theory that residential segregation may have an important role in perpetuating racial inequities in gonorrhea rates. Nonetheless, segregation is inextricably tied to disparities in other social determinants of health, none of which are easily addressed within the context of normal STD prevention programs. Nonetheless, reduction in these social disparities is key to reducing STD disparities.3 Addressing these issues necessitates collaboration with other health promotion campaigns, including both infectious and chronic diseases. There is substantial evidence that residential segregation impacts not only STDs but a variety of other indicators of health and well-being, as well.14 Therefore, to a substantial degree, the success of efforts to alleviate the inequitable burden of STDs is likely to depend on the effectiveness of policy-level measures aimed at decreasing segregation, in addition to concentrated efforts to diagnose and treat disease, as well as ensure proper patient education, within socially segregated populations.
1. Centers for Disease Control and Prevention. Sexually Transmitted Disease Surveillance 2010. Atlanta, GA: U.S. Department of Health and Human Services, 2011.
2. Williams DR, Collins C. Racial residential segregation: A fundamental cause of racial disparities in health. Public Health Rep 2001; 116: 404–416.
3. Hogben M, Leichliter JS. Social determinants and sexually transmitted disease disparities. Sex Transm Dis 2008; 35 (12 suppl): S13–S18.
4. Fischer CS, Stockmayer G, Stiles J, et al. Distinguishing the geographic levels and social dimensions of U.S. metropolitan segregation, 1960–2000. Demography 2004; 41: 37–59.
5. 5. Iceland J, Weinberg DH, Steinmetz E. Racial and ethnic residential segregation in the United States: 1980–2000. U.S. Census Bureau. Washington, DC: U.S. Government Printing Office, 2002: 59–76.
6. Sethi R, Somanathan R. Inequality and segregation. J Political Econ 2004; 112: 1296–1321.
7. Acevedo-Garcia D. Residential segregation and the epidemiology of infectious diseases. Soc Sci Med 2000; 51: 1143–1161.
8. Acevedo-Garcia D. Zip code-level risk factors for tuberculosis: Neighborhood environment and residential segregation in New Jersey, 1985–1992. Am J Public Health 2001; 91: 734–741.
9. Adimora AA, Schoenbach VJ. Social context, sexual networks, and racial disparities in rates of sexually transmitted infections. J Infect Dis 2005; 191 (suppl 1): S115–S122.
10. Thomas JC, Gaffield ME. Social structure, race, and gonorrhea rates in the southeastern United States. Ethnic Dis 2003; 13: 362–368.
11. Kilmarx PH, Zaidi AA, Thomas JC, et al. Sociodemographic factors and the variation in syphilis rates among US counties, 1984 through 1993: An ecological analysis. Am J Public Health 1997; 87: 1937–1943.
12. Kaplan MS, Crespo CJ, Huguet N, et al. Ethnic/racial homogeneity and sexually transmitted disease: A study of 77 Chicago community areas. Sex Transm Dis 2009; 36: 108–111.
13. Echenique F, Fryer RF. On the Measurement of Segregation. Labor and Demography, Series 0503006. EconWPA: 2005.
14. Kramer MR, Hogue CR. Is segregation bad for your health? Epidemiol Rev 2009; 31: 178–194.
15. Reardon SF. A conceptual framework for measuring segregation and its association with population outcomes. In: Oakes JM, Kauffman JS, eds. Methods in Social Epidemiology. San Francisco, CA: John Wiley & Sons, Inc, 2006: 169–192.
16. Acevedo-Garcia D, Osypuk TL. Invited commentary: Residential segregation and health—The complexity of modeling separate social contexts. Am J Epidemiol 2008; 68: 1255–1258.
17. 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: 215–221.
18. Fischer MJ. The relative importance of income and race in determining residential outcomes in U.S. urban areas, 1970–2000. Urban Aff Rev 2003; 38: 669–696.
19. Osypuk TL, Acevedo-Garcia D. Are racial disparities in pre-term birth larger in hypersegregated areas? Am J Epidemiol 2008; 167: 1295–1304.
20. Subramanian SV, Acevedo-Garcia D, Osypuk TL. Racial residential segregation and geographic heterogeneity in black/white disparity in poor self-rated health in the US: A multilevel statistical analysis. Soc Sci Med 2005; 60: 1667–1679.
21. Office of Management and Budget. Standards for defining metropolitan and micropolitan statistical areas; notice. Fed Regist 2000; 65: 82228–82238.
22. U.S. Census Bureau. A Compass for Understanding and Using American Community Survey Data: What Researchers Need to Know. Washington, DC, U.S. Government Printing Office, 2009.
23. Massey DS, Denton NA. The dimensions of residential segregation. Soc Forces 1988; 67: 281–315.
24. Weinberg DH. U.S. Neighborhood income inequality in the 2005–2009 period. U.S. Census Bureau American Community Survey Reports. 2011; ACS-16.
25. Bishaw A, Semega A. Income, earnings, and poverty data from the 2007 American Community Survey. U.S. Census Bureau American Community Survey Reports. 2008; ACS-09.
26. Rothman KJ, Greenland S, Lash TL. Modern Epidemiology, 3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins, 2008.
27. Richardson DB, Kaufman JS. Estimation of the relative excess risk due to interaction and associated confidence bounds. Am J Epidemiol 2009; 169: 756–60.
28. Newman L, Berman S. Epidemiology of STD disparities in African American communities. Sex Transm Dis 2008; 35 (suppl 12): S4–S12.
29. Susser M. The logic in ecological: I. The logic of analysis. Am J Public Health 1994; 84: 825–829.
30. Susser M. The logic in ecological: II. The logic of design. Am J Public Health 1994; 84: 830–835.
31. Koopman JS, Longini IM. The ecological effects of individual exposures and nonlinear disease dynamics in populations. Am J Public Health 1994; 84: 836–842.