Heterosexual contact is the second most common route of HIV transmission in the United States, after male-to-male sexual contact.1 The US Centers for Disease Control and Prevention (CDC) defines high-risk heterosexual contact as “heterosexual contact with individuals known to be HIV positive or at high risk of HIV infection,” such as, people having sex with injection drug users or women having sex with men who also have sex with men. In 2007, African Americans comprised 49% of the adult and adolescent HIV cases in the United States, but only 12% of the US population.2 Among African Americans, 66% of HIV cases in 2006 were attributed to high-risk heterosexual contact.3 Based on these data, there is an urgent need to clarify the factors that put African Americans at high risk of HIV infection, specifically those at risk through heterosexual sex.
Structural factors, such as poverty and neighborhood conditions, have been strongly linked to increased HIV and sexually transmitted infection (STI) rates globally and in the United States.4-8 In the United States, individuals living in neighborhoods with high levels of poverty have an increased risk of HIV and STI acquisition.2,9 Furthermore, poverty, as HIV, is not equally distributed throughout the United States. That African Americans are disproportionately affected by poverty is well-documented, but because US public health surveillance systems do not collect income information, our understanding of how poverty relates to HIV/STI in the context of racial disparities is limited.4
The population of Baltimore, MD, may provide insights into the factors that disproportionately place low income, African American populations at risk of HIV infection. The Baltimore-Towson Metropolitan Statistical Area (MSA) was ranked fifth highest in US MSA AIDS case report rates in 20071 and the overall rate of incident HIV diagnoses for Baltimore City was 157.0 per 100,000 in 2006,10 which is approximately 7 times higher than the national estimate for the United States per year from 2003 to 2006.11 About 64% of Baltimore City's population is African American, of whom 26% live below the poverty level.2
In the current study, we analyze and present data collected from the Baltimore site of the National HIV Behavioral Surveillance System (NHBS),12 a multisite study sponsored by the CDC. The data were collected during the cycle targeting high-risk heterosexuals. One of the goals of NHBS is to estimate HIV prevalence for each high-risk group. Because they are considered a hard-to-reach group, studies targeting high-risk heterosexuals often recruit them through sexually transmitted disease (STD) clinics or other settings where at-risk individuals seek STI treatment.13-16 To reduce selection biases in the estimation of HIV prevalence of high-risk heterosexuals, alternative sampling methods were used in NHBS, such as venue-based sampling (VBS) or respondent-driven sampling.17 In addition to a review of the sampling methodology, we present data on sexual risk behaviors and HIV prevalence of individuals living in high poverty areas of Baltimore City, and as results of an analysis looking at factors associated with recent unprotected intercourse (UI) with casual or exchange partners.
Sampling and Recruitment
The NHBS system was created by the CDC to collect risk behavior data from 3 HIV high-risk groups: men who have sex with men (NHBS-MSM), injection drug users (NHBS), and heterosexuals in areas with high poverty and HIV prevalence (NHBS-HET).12 Data for this survey were collected in 25 metropolitan statistical areas (core urban areas containing a minimum of 50,000 population18), including the Baltimore-Towson MSA, during the first NHBS-HET cycle.
The data collection period of the Baltimore site of NHBS-HET, also known as the BEhavioral SUrveillance REsearch (BESURE) Study, took place from July to October 2007. VBS was used at the Baltimore site.17,19,20 The target population of heterosexuals at risk of HIV infection was defined by the NHBS as being adult men and women with a “physical connection” to a high-risk area (HRA) with high rates of poverty and HIV/AIDS, and who had sex with at least 1 opposite-sex partner in the past year. Physical connection to an area refers to living or socializing in a defined HRA.
Standardized morbidity rates for heterosexually transmitted AIDS cases and poverty were calculated for each census tract within the Baltimore-Towson MSA, standardizing each indicator to the rate for Maryland as a whole. Eligible census tracts were those that were categorized as being in the top 20% of census tracts most affected by poverty and heterosexually transmitted AIDS within the Baltimore-Towson MSA (N = 118). HRAs were dropped from this list if (1) they contained institutionalized populations as such individuals would not be eligible for recruitment; (2) they were considered unsafe for field work, as determined by direct observation; (3) they did not meet the CDC's criteria for feasible recruitment; or (4) they underwent substantial physical or population changes after initial data collection. The result of this elimination was 9 remaining tracts, which became the study's recruitment areas. An additional recruitment area was added during a special recruitment event, defined as a 1-time event with a potentially high attendance of the target population, for a total of 10 recruitment areas.
A 2-stage sampling method was used for the construction of the monthly recruitment calendar. In stage 1 sampling, venues corresponding to the number of planned recruitment events were randomly selected from the sampling frame without replacement. In the second stage, the 4-hour day-time periods for each venue were ranked and the period with the highest ranking was selected. These day-time periods and their venues were then scheduled onto the monthly recruitment calendar.
At the recruitment events, team members would enroll eligible individuals into the study. Upon arrival at the venue in vans custom-fitted with spaces for confidential interviewing and phlebotomy, a defined geographic area was established such that people entering this area who appeared to be eligible by age and sex were counted and approached for enrollment. Individuals were eligible if they had not previously participated in the current cycle of NHBS, were between 18 and 50 years of age, resided in Baltimore City, were men or women (not transgender), reported vaginal or anal sex with a person of the opposite sex in the past 12 months, and had the ability to complete the interview in English. During the recruitment period, all individuals appearing to be men or women and between the ages of 18-50 were enumerated to determine selection probabilities.
Eligible individuals were enrolled and consented, and trained interviewers administered the survey. Surveys had 1 component that was the same for all sites, and a section on health and HIV/STI specific to the Baltimore site. Participants completed the surveys anonymously; unique study numbers were sequentially assigned. After completing the 45-minute survey, a Maryland state-certified HIV counselor conducted an HIV/AIDS counseling session and performed phlebotomy. Participants were reimbursed $25 for completing the surveys and another $25 for HIV testing. They were instructed to call a phone number in 2 days with their study numbers to obtain their HIV test results, and posttest counseling or referrals, as appropriate.
The NHBS-HET protocol and all local materials were approved by Institutional Review Boards at the CDC, the Johns Hopkins Bloomberg School of Public Health, and the Maryland Department of Health and Mental Hygiene.
The NHBS questionnaire collected information on demographic characteristics, sexual behaviors, drug use behaviors, STI history, HIV testing behaviors, and health insurance status of the study population. For a majority of behavioral questions, a 12-month recall period was used. Race and ethnicity categories were compiled using Census 2000 classifications and participants were allowed to choose more than 1 race. Because a vast majority of participants in Baltimore reported themselves as being African American, we collapsed race categories into African American, African American mixed, and other, which included white only and other races that did not also identify as African American. Ethnicity was not taken into consideration when creating race/ethnicity categories because the number of individuals reporting Hispanic ethnicity was small. Using annual household income and number of dependents, we created a poverty indicator using the federal poverty guidelines.21 Two participants who did not know their annual household income were assigned a value indicating that they were not living below poverty level, as we considered it to be the more conservative value. Age was categorized into quartiles. Marital status was categorized as (1) never married, (2) divorced or separated, or (3) married or living together as married. Education was categorized as having graduated from high school or received their General Education Development certificate or not.
Participants were asked about their HIV risk behaviors during the past 12 months. Information on lifetime injection of drugs was also collected. Sexual risk behaviors included the number of opposite-sex partners; types of sex partners, such as main (“man/woman you have sex with and who you feel committed to above anyone else. This is a partner you would call your boy/girlfriend, husband/wife, significant other, or life partner”), casual (“man/woman you have sex with but do not feel committed to or don't know very well”), or exchange ("man/woman you have sex with in exchange for things like money or drugs); having had UI and with what types of partners, same sex partners, concurrent sexual partners, and having been diagnosed with an STI. UI was defined as any act of vaginal or anal sex during which a condom was not used, and was examined by sex partner type. Substance use behaviors included injection of drugs, the use of noninjection drugs and binge drinking. Binge drinking was assessed over the past year and during the past 30 days and was defined as 5 or more drinks in 1 sitting for men and as 4 or more drinks in 1 sitting for women.
All blood specimens with sufficient volume were tested for HIV-1 antibodies with a licensed enzyme immunoassay (Sanofi Diagnostics Pasteur, Chaska, MN). We confirmed repeatedly reactive samples using Western blot (Bio-Rad, Hercules, CA or Epitope Inc, Organon-Teknika Corporation, Durham, NC). An HIV-seropositive individual was defined as having a reactive enzyme immunoassay with a positive Western blot confirmation.
The original outcome of interest was unprotected anal or vaginal intercourse during the past 12 months. However, using an outcome of any reported UI in the past 12 months led to small cell sizes, as 84% of participants reported UI in the past 12 months with any type of sex partner, whereas 72% reported UI with main partners in the past 12 months. Based on these findings in the data, we created a 3-level composite UI variable as a potential outcome: no UI in the past 12 months, UI with main partners only, and UI with any casual or exchange partners. To examine whether this composite partner-specific UI variable was an appropriate proxy measure for HIV risk, we examined univariate associations between levels of the composite UI variable and reported history of any STD diagnosis (a disease proxy of HIV risk). To provide this particular analysis with more power, we included all participants in the study (N = 323), and 9 individuals who were referred to us as sexual partners of study participants, for a total of 332 participants. Table 1 shows the results of a Fisher Exact test on those 332 participants comparing levels of the composite UI variable and their history of any STD diagnosis. Among those individuals who reported either no UI in the past 12 months or UI with main partners only (N = 205) there was no statistical difference between those 2 groups with regard to their history of self-reported STD diagnosis. Therefore, we collapsed the 2 categories together, resulting in a dichotomous outcome: reported UI with casual or exchange partners in the past 12 months or not.
The chi-squared statistic was used to examine univariate associations between HIV sexual risk and demographic, sexual risk, drug use, and other risk variables. Associations are presented as unadjusted prevalence ratios (PRs). Multivariate analysis was used to assess correlates of HIV sexual risk. Variables were selected based on statistical significance upon univariate analysis (P < 0.05) and/or scientific merit, including sociodemographic characteristics, sexual and other risk behaviors, and health-related factors. For the composite sexual risk variable, the reference category was no reported UI with casual or exchange partners during the past 12 months. Using multivariate log-binomial regression, covariates were adjusted for age and race, and presented as an adjusted prevalence ratio (APR) with 95% confidence intervals (CIs). All statistical procedures were performed using SAS version 9.1.3. (Cary, NC).
The Baltimore City site of NHBS-HET began recruitment in July 2007 and ended October 2007. Over the course of the 49 recruitment events, 5339 adult men and women were enumerated. Of the 363 people who were screened for eligibility, 323 (89%) met the eligibility criteria and were enrolled into the study. The final sample was comprised of the 301 (93% of eligible) participants who completed surveys and had HIV test results.
HIV prevalence in our sample was 3% (n = 10). The median age of participants (N = 301) in this study was 33 years and 47% were men. Nearly all (92%) of our sample identified themselves as being only African American or African American of mixed race. Seventy-five percent of this sample reported their current marital status as never having married. Over one-third (36%) of participants had less than a high school education.
Most participants (84%) reported having unprotected sex during the past 12 months, whereas 19% of participants reported sex with exchange partners. Among participants who reported casual partners, a lower percentage reported UI with those partners (63%), compared with the percentage who reported UI among those who had main partners (84%). Although 57% of participants reported using noninjection nonprescribed drugs (including heroin, crack, and cocaine) in the past 12 months, only 12% reported injecting drugs. Twelve percent reported ever having been diagnosed with an STD.
Tables 2 and 3 present the results of univariate analysis of sociodemographic and risk variables with the outcome UI with casual or exchange partners in the past 12 months. Univariate analysis indicated that those individuals who were currently homeless, older, and men were more likely to report UI with a casual or exchange partner. In terms of sexual behaviors, those reporting 2, 3, or 4 or more sex partners (compared to just 1), same sex partners, or concurrent partners during the past year were also more likely to report UI with a casual or exchange partner. Substance abuse was also associated with reporting UI with a casual or exchange partner, including the report of injecting drugs during the past year, the use of noninjection nonprescribed drugs during the past year, and binge drinking during the past 30 days. Those with a history of diagnosis of any STD or arrest by the police during the past 12 months were more likely to report UI with a casual or exchange partner.
Table 4 presents the results of a multivariate log-binomial regression of factors associated with UI with casual or exchange partners in the past 12 months, which was reported by nearly 40% of participants. Multivariate log-binomial regression revealed the following variables to be independently associated with reported UI with a casual or exchange partner during the past 12 months: current homelessness (APR 1.17; 95% CI [1.05-1.31]), increasing age (25-33 years old, APR 1.62; 95% CI [1.06-2.48], 34-41 years old, APR 1.92; 95% CI [1.28-2.90], 42-50 years old, APR 2.19; 95% CI [1.45-3.31]), 4 or more sex partners during the past 12 months (APR 3.50; 95% CI [1.46-6.39]), reporting concurrent sex partners in the past 12 months (APR 2.14; 95% CI [1.07-4.25]), binge drinking during the past 30 days (APR 1.18; 95% CI [1.04-1.33]), and lifetime history of STD diagnoses (APR 1.18; 95% CI [1.06-1.32]).
Three percent of the study population was found to be HIV infected, a prevalence which is approximately 7 times higher than the CDC's HIV prevalence estimate for the United States at the end of 2006.22 It should be emphasized that this population was recruited using VBS, a sampling method originally developed to target MSM.19 Unlike the types of venues used to recruit MSM, these venues were not attended for the sole purpose of socializing or finding sexual partners. Venues included retail stores, fast-food restaurants, and street corners close to grocery stores, libraries, and other common neighborhood establishments. Our HIV testing results indicate that there is a generalized HIV epidemic occurring among the population having heterosexual sex and residing or socializing in low income neighborhoods in Baltimore, MD. Sexual risk behaviors were prevalent in this population, suggesting that additional HIV prevention efforts in these geographic areas are needed. Furthermore, sociodemographic characteristics indicated that a substantial proportion of the study population lived below the poverty line and had less than a high school education. The implications of this are that design of an HIV prevention intervention targeting this population should be done keeping in mind the structural barriers normally encountered by individuals in high poverty areas, including limited access to care and services23 and a lack of insurance.24
Recent binge drinking, concurrent sexual partnerships, and having 4 or more sex partners in the past 12 months were independently associated with reported UI with a casual or exchange partner during the past 12 months, suggesting that in a high poverty and high HIV prevalence area, high-risk sexual behaviors often co-occur. Though our study enrolled people from all races, 92% identified themselves as being of African American descent. In a study of African American men recruited from urban health clinics in Boston, MA, recent binge drinking was found to be independently associated with UI with nonregular female partners, exchanging sex, and recent HIV/STI diagnosis.25 The authors of the Boston study proposed that the bundling of these risk factors results in the creation of “higher risk sexual situations,” and that having UI in such situations may be driving the disproportionately high rate of HIV/STI among African Americans.
Similar to other studies, condom use with nonregular partners was more common than condom use with main partners, as individuals may perceive nonregular partners to be riskier.26-28 However, with the majority of both men and women reporting concurrent partnerships, the lack of condom use with main partners could be equally risky, especially in a setting of high HIV prevalence. Although a majority of the study population reported seeing a physician or health care professional in the past year (79%) and more than half of these individuals reported being offered an HIV test at those visits, a high proportion of study participants still report UI (84%), concurrent sexual partnerships (59%), and sex with exchange partners (19%). The data presented here indicate that although most participants are utilizing the healthcare system and that during medical visits sexual health issues are often addressed, risky sexual behaviors are still highly prevalent, suggesting that these are missed opportunities for HIV prevention for a population that is highly burdened with HIV infection.
Due to the high reported levels of unprotected sex in this population, specifically among individuals with main partners, using any unprotected sex in the recent past to assess HIV risk was not a useful research strategy. Consequently, we chose an outcome that was thought to capture more risk in this population, UI with causal or exchange partners. Using this outcome to capture HIV risk is also supported in the literature, as having sex specifically with nonregular partners has been shown to be independently associated with HIV and sexually transmitted infections.29-31
In populations with high proportions of individuals engaging in risky sexual behaviors, the more appropriate questions should be which factors are protective against HIV/STI when condom use is low and concurrency and multiple partnerships are common. Although we did not examine sexual networks, the results of this study strongly suggest that individual level behaviors are not the only or main drivers of HIV infection or STI. Adimora's work with low income African Americans in rural North Carolina provides strong evidence that concurrency and having dense sexual networks are major factors in HIV heterosexual transmission.32-35 Although research among low income African American women has shown that unprotected sex with main male partners is a major risk factor for HIV infection because male partners often have other concurrent sexual partnerships36, in this population, concurrent sex partnerships were highly prevalent (>50%) for both men and women. In our population, having a prior STI diagnosis was moderately associated with unprotected sex with casual or exchange partners, but individuals reporting concurrent sex partners were twice as likely to engage in unprotected sex with casual or exchange partners. Although the cross-sectional nature of this study does not allow the assessment of causal inference, the strength of associations provides some information as to which factors are most relevant.
Our study had several limitations. Temporality of associations cannot be determined in a cross-sectional study. Demographic, health, and HIV risk behavior data were all self-reported, potentially resulting in misreporting due to poor recall or the urge to give socially desirable answers, concerns about stigma, or cultural differences. If the outcome of UI with casual or exchange partners in the past 12 months was subject to the effect of participants wanting to provide socially desirable answers, results could be biased toward the null. A similar effect on results could occur due the exclusion criteria in the selection of HRAs, namely that areas containing institutionalized populations or that were considered unsafe for field work were excluded.
Furthermore, because a vast majority of the population identified themselves as African American, it was not possible to make comparisons between races. The number of HIV-infected individuals in our population was too low to conduct multivariate analyses using HIV serostatus as an outcome. Finally, the results of these studies may not be generalizable to people who infrequently attend the recruitment venues and we may have missed individuals who attend venues we recruited at less frequently. Despite these limitations, this study provides important insights into the HIV epidemic among low income populations, and in particular, low income African Americans in Baltimore City. A complex set of sexual and network dynamics places these individuals at high risk of HIV/STI, compounded by a set of social norms that are barriers to using condoms during sex. In addition, although poverty at the individual level was not found to be associated with unprotected sex with casual or exchange partners, there was an overall lack of variability in poverty and income levels. Future research should focus on understanding the role of neighborhood poverty and the risk of HIV, and other nonindividual level factors, such as sexual network density and concurrency.
1. Centers for Disease Control and Prevention. HIV/AIDS Surveillance Report, 2007
. Vol. 19. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention; 2009.
2. US Census. 2005-2007 American Community Survey 3-Year Estimates
3. Centers for Disease Control and Prevention. HIV/AIDS Surveillance Report, 2006
. Vol. 18. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention; 2008.
4. Aral SO, Adimora AA, Fenton KA. Understanding and responding to disparities in HIV and other sexually transmitted infections in African Americans. Lancet
5. Riley ED, Gandhi M, Hare C, et al. Poverty, unstable housing, and HIV infection among women living in the United States. Curr HIV/AIDS Rep
6. Stratford D, Mizuno Y, Williams K, et al. Addressing poverty as risk for disease: recommendations from CDC's consultation on microenterprise as HIV prevention. Public Health Rep
7. Smith Fawzi MC, Jagannathan P, Cabral J, et al. Limitations in knowledge of HIV transmission among HIV-positive patients accessing case management services in a resource-poor setting. AIDS Care
8. Cohen D, Spear S, Scribner R, et al. “Broken windows” and the risk of gonorrhea. Am J Public Health
9. Krieger N, Chen JT, Waterman PD, et al. Painting a truer picture of US socioeconomic and racial/ethnic health inequalities: the Public Health Disparities Geocoding Project. Am J Public Health
10. AIDS Administration. Maryland HIV/AIDS Epidemiological Profile
. Baltimore, MD: Department of Health and Mental Hygiene; 2007.
11. Hall HI, Song R, Rhodes P, et al. Estimation of HIV incidence in the United States. JAMA
12. Gallagher KM, Sullivan PS, Lansky A, et al. Behavioral surveillance among people at risk for HIV infection in the U.S.: the National HIV Behavioral Surveillance System. Public Health Rep
. 2007;122(Suppl 1):32-38.
13. Hutton HE, McCaul ME, Santora PB, et al. The relationship between recent alcohol use and sexual behaviors: gender differences among sexually transmitted disease clinic patients. Alcohol Clin Exp Res
14. Satterwhite CL, Kamb ML, Metcalf C, et al. Changes in sexual behavior and STD prevalence among heterosexual STD clinic attendees: 1993-1995 versus 1999-2000. Sex Transm Dis
15. Tian LH, Peterman TA, Tao G, et al. Heterosexual anal sex activity in the year after an STD clinic visit. Sex Transm Dis
16. Senn TE, Carey MP, Vanable PA, et al. The male-to-female ratio and multiple sexual partners: multilevel analysis with patients from an STD clinic. AIDS Behav
. 2008. (E-pub ahead of print).
17. Magnani R, Sabin K, Saidel T, et al. Review of sampling hard-to-reach and hidden populations for HIV surveillance. AIDS
. 2005;19(Suppl 2):S67-S72.
19. MacKellar D, Valleroy L, Karon J, et al. The Young Men's Survey: methods for estimating HIV seroprevalence and risk factors among young men who have sex with men. Public Health Rep
. 1996;111(Suppl 1):138-144.
20. Valleroy LA, MacKellar DA, Karon JM, et al. HIV prevalence and associated risks in young men who have sex with men. Young Men's Survey Study Group. JAMA
21. Federal Register. Vol. 72. Office of the Federal Register, National Archives and Records Administration 2007:3147-3148.
22. HIV prevalence estimates-United States, 2006. MMWR Morb Mortal Wkly Rep
23. Reif S, Golin CE, Smith SR. Barriers to accessing HIV/AIDS care in North Carolina: rural and urban differences. AIDS Care
24. DeNavas-Walt C, Proctor BD, Lee CH. Income, Poverty, and Health Insurance Coverage in the United States: 2005
. Washington, DC. 2006.
25. Raj A, Reed E, Santana MC, et al. The associations of binge alcohol use with HIV/STI risk and diagnosis among heterosexual African American men. Drug Alcohol Depend
26. Mercer CH, Copas AJ, Sonnenberg P, et al. Who has sex with whom? Characteristics of heterosexual partnerships reported in a national probability survey and implications for STI risk. Int J Epidemiol
27. Niccolai LM, Ethier KA, Kershaw TS, et al. New sex partner acquisition and sexually transmitted disease risk among adolescent females. J Adolesc Health
28. de Visser RO, Smith AM, Rissel CE, et al. Sex in Australia: safer sex and condom use among a representative sample of adults. Aust N Z J Public Health
29. Dunkle KL, Jewkes RK, Brown HC, et al. Transactional sex among women in Soweto, South Africa: prevalence, risk factors and association with HIV infection. Soc Sci Med
30. Cote AM, Sobela F, Dzokoto A, et al. Transactional sex is the driving force in the dynamics of HIV in Accra, Ghana. AIDS
31. Go VF, Solomon S, Srikrishnan AK, et al. HIV rates and risk behaviors are low in the general population of men in Southern India but high in alcohol venues: results from 2 probability surveys. J Acquir Immune Defic Syndr
32. 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.
33. Adimora AA, Schoenbach VJ, Martinson FE, et al. Concurrent partnerships among rural African Americans with recently reported heterosexually transmitted HIV infection. J Acquir Immune Defic Syndr
34. Adimora AA, Schoenbach VJ, Martinson FE, et al. Heterosexually transmitted HIV infection among African Americans in North Carolina. J Acquir Immune Defic Syndr
35. Adimora AA, Schoenbach VJ, Martinson F, et al. Concurrent sexual partnerships among African Americans in the rural south. Ann Epidemiol
36. HIV transmission among black women-North Carolina, 2004. MMWR Morb Mortal Wkly Rep
Keywords:© 2010 Lippincott Williams & Wilkins, Inc.
HIV prevalence; high-risk heterosexuals; venue-based sampling; National HIV Behavioral Surveillance System