Exploring Psychosocial and Structural Syndemic Effects as Predictors of HIV Risk Behaviors Among Black Women (HPTN 064)

Abstract

Background:

Syndemic models have been used in previous studies exploring HIV-related outcomes; however, these models do not fully consider intersecting psychosocial (e.g., substance use, depressive symptoms) and structural factors (unstable housing, concentrated housing vacancy) that influence the lived experiences of women. Therefore, there is a need to explore the syndemic effects of psychosocial and structural factors on HIV risk behaviors to better explain the multilevel factors shaping HIV disparities among black women.

Methods:

This analysis uses baseline data (May 2009–August 2010) from non-Hispanic black women enrolled in the HIV Prevention Trials Network 064 Women’s Seroincidence Study (HPTN 064) and the American Community Survey 5-year estimates from 2007 to 2011. Three parameterizations of syndemic factors were applied in this analysis a cumulative syndemic index, three syndemic groups reflecting the level of influence (psychosocial syndemic group, participant-level structural syndemic group, and a neighborhood-level structural syndemic group), and syndemic factor groups. Clustered mixed effects log-binomial analyses measured the relationship of each syndemic parameterization on HIV risk behaviors in 1,347 black women enrolled in HPTN 064.

Results:

A higher syndemic score was significantly associated with increased prevalence of unknown HIV status of the last male sex partner (adjusted prevalence ratio (aPR) = 1.07, 95% confidence interval or CI 1.04–1.10), involvement in exchange sex (aPR = 1.17, 95% CI: 1.14–1.20), and multiple sex partners (aPR = 1.07, 95% CI: 1.06–1.09) in the last 6 months. A dose–response relationship was observed between the number of syndemic groups and HIV risk behaviors, therefore, being in multiple syndemic groups was significantly associated with increased prevalence of reporting HIV risk behaviors compared with being in one syndemic group. In addition, being in all three syndemic groups was associated with increased prevalence of unknown HIV status of the last male sex partner (aPR = 1.67, 95% CI: 1.43–1.95) and multiple sex partners (aPR = 1.53, 95% CI: 1.36–1.72).

Conclusions:

Findings highlight syndemic factors influence the lived experiences of black women.

Introduction

HIV continues to disproportionately affect black women in the United States, who in 2021 had an estimated HIV incidence rate of 17.5 per 100,000 women. This rate was almost eight times higher than the estimated HIV incidence rate for white women (1.8 per 100,000) and almost four times more than the estimated HIV incidence rate for Hispanic/Latino women (4.6 per 100,000).¹

Existing evidence provides support that psychosocial factors (e.g., intimate partner abuse, drug use, depression), structural factors (e.g., unstable housing, food insecurity, and neighborhood poverty), and HIV risk behaviors intersect to create a syndemic that can further increase the risk of HIV.^2–6 This evidence conveys how critical it is to explore factors operating on multiple levels that may influence HIV risk among black women and serve as targets for future interventions.

Psychosocial factors have been found to be significantly associated with HIV risk behaviors. The presence of depressive symptoms is associated with less condom use, as well as having multiple sex partners among groups at risk for poor sexual health outcomes.^7,8 Furthermore, previous studies have shown that women who reported violent victimization experiences such as childhood sexual abuse and intimate partner violence were more likely to report early sexual debut,^9,10 multiple sex partners,^9–11 substance use,^9,10,12 involvement in exchange sex,¹⁰ history of sexually transmitted infections,^11,13 and condomless sex.^11,14 In addition, structural factors have been associated with HIV risk. HIV surveillance data examining HIV diagnoses and social determinants of health convey that HIV diagnosis rates among black women were highest in census tracts with the highest levels of poverty, income inequality, and percentage of residents without health insurance; and lowest levels of education attainment and median household income.¹⁵ In the HIV Prevention Trials Network 064 Women’s Seroincidence Study (HPTN 064), where one of the eligibility criteria included living in a census tract or zip code with higher ranked HIV prevalence and poverty, estimated HIV incidence was five times greater than the national estimate of HIV incidence among black women.¹⁶ In addition, in a study exploring HIV prevention among women with a history of incarceration, history of incarceration was found to be significantly associated with less condom use, involvement in exchange sex, and increased risk of HIV acquisition.¹⁷ Despite the extensive evidence that shows psychosocial and structural factors are independently associated with HIV risk behaviors, little is known about how these factors converge and create a syndemic that may increase HIV risk behaviors among black women.

Syndemic theory is a theoretical framework that has been the foundation of studies exploring multiple co-occurring factors that are hypothesized to increase disease vulnerability.^18–20 A syndemic has been defined as “two or more epidemics, interacting synergistically and contributing, as a result of their interaction, to excess burden of disease in a population.”^18–20 Studies that have used the syndemic framework to evaluate HIV-related outcomes among black populations have provided valuable insights on the importance of exploring the synergistic effects of individual-level factors on HIV-related outcomes.²¹ However, there continues to be a need to explore the synergistic role of neighborhood-level factors on HIV risk behaviors to better explain the multilevel factors shaping HIV disparities among black women. Previous studies that have examined syndemics among women have applied the Substance Abuse, Violence and AIDS (SAVA) syndemic, which refers to the clustering effect of substance abuse, violence, and HIV/AIDS;^22–25 however, psychosocial and structural factors at the individual and neighborhood levels were not included as factors to be considered within the SAVA syndemic framework. Analytical approaches to modeling syndemics have included the sum score approach to estimate the additive effects of variables, factor analyses to investigate the correlations between syndemic variables, and latent class analyses to identify distinct syndemic risk classes or clusters. Although there are multiple ways to measure syndemics, the sum score approach remains the dominant methodology used in research and continues to show that the number of adverse events have a dose–response relationship with health outcomes.^3,22,26–28

In this analysis, we aim to measure the additive effects of multiple psychosocial and structural factors on HIV risk behaviors using three syndemic parameterizations, a method used in a previous study investigating syndemic factors among women at risk for HIV.³ The development of a novel syndemic profile that includes psychosocial and structural factors can serve as a useful construct for understanding HIV risk in black women. Using data from the HPTN 064, the aims of this study included:

• 1.Measuring the cross-sectional relationship between a cumulative syndemic of psychosocial and structural factors and HIV risk behaviors among black women.
• 2.Categorizing each psychosocial and structural syndemic factor into one of three syndemic groups reflecting the level of influence: psychosocial, participant-level structural, and neighborhood-level structural which allowed us to test the relationship between each syndemic group and HIV risk behaviors.
• 3.Measuring the magnitude of the relationship between experiencing none, one, two, or three syndemic groups and HIV risk behaviors to assess whether experiencing multiple syndemic groups is associated with a higher prevalence of reporting HIV risk behaviors.

Methods

Participants

This analysis utilizes a cross-sectional study design, sampling from the HPTN 064. HPTN 064 was a large multisite, longitudinal study conducted to estimate the overall HIV-1 incidence rate among US women living in areas characterized by higher ranked HIV prevalence rates and poverty.¹⁵ The study was conducted in Atlanta, GA; Baltimore, MD; New York, NY; Newark, NJ; Raleigh/Durham, NC; and Washington, DC. Using venue-based, time–space sampling methods, women living in census tracts or zip codes that ranked in the top 30th percentile of HIV prevalence and with more than 25% of residents living below the U.S. federal poverty threshold, as defined by the 2008 U.S. Census Bureau were recruited to participate in the study.^16,29

Eligible participants were self-identified as women, aged between 18 and 44 years at study enrollment, willing to take an HIV test and receive results, resided in a study location, reported at least one instance of condomless vaginal or anal sex with a man in the past 6 months, reported at least one additional high-risk characteristic (e.g., illicit drug use, history of sexually transmitted infections) that she or a partner engaged in the past 6 months; and provided informed consent.³⁰

HPTN 064 study participants were enrolled between May 2009 and July 2010, completing an enrollment visit at baseline, as well as 6-month and 12-month follow-up visits. During the enrollment, 6-month, and 12-month follow-up visits, study participants received HIV rapid testing and completed assessments using audio computer-assisted self-interview (ACASI) technology. In addition, study participant’s residential addresses were obtained at baseline and were linked to a census tract using the U.S. Census Bureau online geocoder. Institutional Review Boards at each of the HPTN 064 study sites and collaborating institutions were approved, and a certificate of confidentiality was obtained.

To create the neighborhood-level structural variables, study participant data was merged at the census tract level with 5-year estimates from 2007 to 2011 from the U.S. Census Bureau’s American Community Survey (ACS). The ACS is an ongoing nationwide survey that aims to provide a more statistically accurate picture of the demographic, social, economic, and housing characteristics of the communities we live in.³¹ Data from 170 black women were excluded in this analysis due to the inability to match participants’ geocoded addresses with ACS census-tract data. In addition, data from 230 black women were excluded due to participants’ addresses being geocoded at the zip code level rather than the census-tract level.

Measures

Outcome variables

The outcome variables for this analysis were the HIV risk behaviors reported at baseline and assessed as occurring in the past 6 months including condomless vaginal sex, condomless anal sex, unknown HIV status of last male sex partner, exchange sex for commodities such as money, drugs, and goods, and multiple sex partners.

Condomless vaginal sex was defined as reporting any vaginal intercourse with a man without a condom in the past 6 months. Condomless anal sex was defined as reporting any anal intercourse with a man without a condom in the past 6 months. Unknown HIV status of last male sex partner was determined by reporting not knowing the HIV status of the last male sex partner. Exchange sex was defined as reporting sexual intercourse with any man in the last 6 months for commodities such as money, drugs, and goods. Multiple sex partners were defined by reporting more than one male sex partner in the past 6 months.

Exposure variables

Psychosocial variables

Ongoing abuse was measured by self-reporting at least one experience of emotional abuse, physical violence, or forced sex in the past 6 months. History of childhood abuse was defined as reporting being abused physically, emotionally, or sexually before the age of 18 years old. Depressive symptoms in the past 6 months were determined by using a shortened eight-item version of the Center for Epidemiologic Studies Depression (CES-D) scale.³² The eight-item CES-D scale demonstrated high internal consistency (Cronbach’s α = 0.91) in this sample. The shortened version of the CES-D scale has been used in previous studies that assess depressive symptoms among young black women.³³ A score of 7 or greater was indicative of depressive symptoms. Posttraumatic stress disorder (PTSD) symptoms in the past 6 months were assessed using the four-item Primary Care PTSD Screening tool.³⁴ The four-item measure demonstrated high internal consistency (Cronbach’s α = 0.79). The presence of PTSD symptoms was indicative of a score of 3 or greater.^35–37 Heavy alcohol use and illicit drug use were determined by using a modified version of the World Health Organization Alcohol, Smoking and Substance Involvement Screening Test scale.³⁸ Heavy alcohol use was defined as consuming four or more drinks on one occasion at least two times per week in the past 6 months. Illicit drug use was determined by reporting weekly to daily or almost daily use of amphetamines, cocaine, hallucinogens, inhalants, opioids, and sedatives and sleeping pills in the past 6 months.

Participant-level structural variables

Participant-level structural variables included incarceration, food insecurity, lack of health insurance, and unstable housing. Incarceration was measured by self-reporting incarceration in jail or prison at least once in the past 6 years. Food insecurity was determined by reporting being concerned about having enough food in the past 6 months. Lack of health insurance was defined as not having Medicaid, Medicare, health maintenance organizations, private or commercial health insurance in the past 6 months. Unstable housing was determined whether participants reported living at a friend’s house or apartment, a halfway house or treatment center, a homeless shelter, a motel/hotel, or on the street in the past 6 months.

Neighborhood-level structural variables

Neighborhood-level structural variables were constructed by linking participant’s geocoded addresses to census tracts using the ACS 5-year estimates from 2007 to 2011. Concentrated rental-burdened households were operationalized as living in a census tract where 50% or more of residents are paying 30% or more of income on rent.³⁹ Neighborhood unemployment was operationalized as living in a census tract where 15% or more of residents, 16 years and older, were unemployed. Neighborhood low educational attainment was operationalized as living in a census tract where 25% or more of residents had less than a high school diploma. Concentrated housing vacancy was defined as living in a census tract where 15% or more of vacant residential housing units are present. Neighborhood black racial concentration was defined as living in a census tract where 50% or more of residents were black/African American. Neighborhood black racial concentration was used as a proxy for black racial segregation.⁴⁰ Concentrated income inequality was defined as living in a census tract with a Gini coefficient of 0.44. Gini coefficients range from 0 to 1, with 0 indicating complete equality and 1 indicating complete inequality. High-income inequality has been identified as 0.44.⁴⁰

Syndemic parameterizations

Similar to a previous study,³ three syndemic parameterizations were applied in this analysis. First, a cumulative syndemic index was created by summing the psychosocial and structural variables, with a range of 0–16. The binary coding structure of these variables (e.g., responses are within a range of 0 to 1) allowed for standardization, making it so that no one factor was worth more than the others. Second, the psychosocial and structural syndemic factors were categorized into three cumulative syndemic groups reflecting the level of influence: a psychosocial syndemic group, which included ongoing abuse, history of childhood abuse, depressive symptoms, PTSD symptoms, illicit drug use, and heavy alcohol use, and ranged from 0 to 6; a participant-level structural syndemic group, which included lack of health insurance, incarceration, unstable housing, and food insecurity, and ranged from 0 to 4; and a neighborhood-level structural syndemic group, which included concentrated rental-burdened households, neighborhood unemployment, neighborhood low educational attainment, concentrated housing vacancy, neighborhood black racial concentration, and concentrated income inequality, and ranged from 0 to 6. Finally, we examined syndemic factor groups, that is, the exposure to none, one, two, or three syndemic groups.

Statistical analysis

The prevalence of psychosocial and structural syndemic variables and outcome variables reported at baseline were summarized by calculating frequencies and percentages. The mean and standard deviation were calculated for the cumulative syndemic index and the three syndemic group. In addition, relationships between each psychosocial and structural factor were assessed by generating a relationship matrix using unadjusted log-binomial regression models.

Unadjusted and adjusted log-binomial regression models were used to estimate prevalence ratios, measuring the associations between each HIV risk behavior outcome and syndemic parameterization. Covariates including study site, age, education, and household income were controlled for in adjusted models. These covariates were chosen a priori based on previous literature that included these potential confounders in models evaluating HIV risk behavior outcomes.^3,41–44 Sensitivity analyses were conducted to examine whether household income was reflected in the neighborhood-level structural variables. In the sensitivity analyses, household income was not controlled for in the multivariate cumulative syndemic index model and the multivariate neighborhood-level structural syndemic model. The results did not meaningfully change, and our conclusions remained the same.

Outliers were assessed by measuring Cook’s distance, leverages, and studentized residuals; there was no evidence of overly influential observations. Variance inflation and multicollinearity were assessed by measuring the variance inflation factor (VIF) values; the results indicated that there was no evidence of variance inflation (VIF > 5). Missing data were addressed with multiple imputations using the Markov chain Monte Carlo method assuming an arbitrary missing pattern. Values that were imputed were truncated to fit within the range of possible values and were rounded to the nearest whole number to have the binary coding structure.

All analyses were conducted using log-binomial regression models with generalized estimating equations and an independent correlation structure to account for clustering by census tract. Models generated prevalence ratios (PRs) with 95% confidence intervals (CIs). SAS Studio (SAS Institute, Inc., Cary, NC) was used to conduct all analyses.

Results

A total of 2,099 participants were enrolled in the HPTN 064 study. Of those, 1,747 were identified as non-Hispanic, cis-gender black women. Data from 400 study participants were excluded due to the inability to match geocoded addresses with ACS census-tract data (n = 170) and geocoded addresses were at the zip code level rather than the census-tract level (n = 230). As a result, our final analytic sample included data from 1,347 study participants.

Of the 1,347 participants included in this analysis, the majority (n = 859, 64%) were less than 35 years old. Most participants had an annual household income of less than $10,000 (n = 631, 72.0%) and were high school graduates (n = 542, 40.2%). At baseline, a large proportion of participants were living in census tracts characterized by concentrated rental-burdened households (n = 912, 67.7%), neighborhood unemployment (n = 1,031, 76.5%), concentrated housing vacancy (n = 1,028, 76.3%), neighborhood black racial concentration (n = 1,223, 90.8%), and concentrated income inequality (n = 835, 62.0%). On average, participants experienced seven syndemic factors overall (M = 7.4, SD = 2.5); and two psychosocial syndemic factors (M = 1.8, SD = 1.6), 1 participant-level syndemic factor (M = 1.3; SD = 1.0), and 4 neighborhood-level syndemic factors (M = 4.2, SD = 1.2). In addition, 63.4% (n = 854) of participants experienced syndemics in at least two syndemic groups (Table 1).

Table 1.

Syndemic and HIV Risk Behavior Characteristics Reported at Baseline Among non-Hispanic Black Women in the HPTN 064 Cohort Study

	n (%)
Demographic characteristics
18–24 years	408 (30.3)
25–34 years	451 (33.5)
35–44 years	486 (36.1)
45+ years	2 (0.2)
Annual household income (n = 877)
<$10,000	631 (72.0)
$10,000–$20,000	131 (14.9)
$20,001–$40,000	84 (9.6)
$40,001–$60,000	21 (2.4)
$60,001–$80,00	8 (0.9)
>$80,000	2 (0.2)
Education status (n = 1,347)
Less than HS diploma	484 (35.9)
HS graduate	542 (40.2)
Some college	237 (17.6)
Vocational school	64 (4.8)
College graduate and above	20 (1.5)
Variables reported in past 6 months
Psychosocial characteristics
Ongoing abuse (n = 1,331)
Yes	495 (37.2)
No	836 (62.8)
History of childhood abuse (n = 1,323)
Yes	588 (44.4)
No	735 (55.6)
Depressive symptoms(n = 1,316)
Yes	440 (33.4)
No	876 (66.6)
Posttraumatic stress disorder symptoms (n = 1,331)
Yes	387 (29.1)
No	944 (70.9)
Illicit drug use (n = 688)
Yes	291 (42.3)
No	397 (57.7)
Heavy alcohol use (n = 1,122)
Yes	317 (28.3)
No	805 (71.8)
Structural factors
Participant-level characteristics
Incarceration (in past 5 years) (n = 1,347)
Yes	547 (40.6)
No	800 (59.4)
Food insecurity (n = 1,327)
Yes	607 (45.7)
No	720 (54.3)
Unstable housing (n = 1,324)
Yes	203 (15.3)
No	1,121 (84.7)
Health insurance status (n = 1,292)
Has health insurance	431 (33.4)
Does not have health insurance	861 (66.6)
Neighborhood-level characteristics (n = 1,347)
Concentrated rental-burdened households (living in a census tract where 50% or more of residents are paying 30% or more of income on rent)
Yes	912 (67.7)
No	435 (32.3)
Neighborhood low educational attainment (living in a census tract where 25% or more of residents had less than a high school diploma)
Yes	632 (46.9)
No	715 (53.1)
Neighborhood unemployment (living in a census tract where 15% or more of residents, 16 years and older, were unemployed)
Yes	1031 (76.5)
No	316 (23.5)
Concentrated housing vacancy (living in a census tract where 15% or more of vacant residential housing units are present)
Yes	1,028 (76.3)
No	319 (23.7)
Living in census tracts with neighborhood black racial concentration (living in a census tract where 50% or more of residents were black/African American)
Yes	1,223 (90.8)
No	124 (9.2)
Concentrated income inequality (living in a census tract with a Gini coefficient of 0.44)
Yes	835 (62.0)
No	512 (38.0)
Syndemic parameterizations
Cumulative syndemic score, M (SD)	7.4 (2.5)
Psychosocial syndemic group score, M (SD)	1.8 (1.6)
Participant-level syndemic group score, M (SD)	1.3 (1.0)
Neighborhood-level syndemic group score, M (SD)	4.2 (1.2)
Number of syndemic groups (n = 1,347)
0	8 (0.6)
1	485 (36.0)
2	506 (37.6)
3	348 (25.8)
HIV risk behaviors (reported in past 6 months)
Condomless vaginal intercourse (n = 1,331)
Yes	1,086 (81.6)
No	245 (18.4)
Condomless anal intercourse (n = 541)
Yes	399 (81.3)
No	92 (18.7)
Unknown HIV status of last sexual partner (n = 1,327)
Yes	620 (46.1)
No	724 (53.9)
Exchange sex (n = 1,327)
Yes	522 (39.3)
No	805 (60.7)
Multiple sex partners (two or more partners) (n = 1,324)
Yes	808 (61.0)
No	516 (39.0)

A large proportion of participants reported condomless vaginal intercourse (n = 1,086, 81.6%), condomless anal intercourse (n = 399, 81.3%), and having multiple male sex partners (n = 808, 61.0%) in the past 6 months. Almost half of the sample reported unknown HIV status of last male sexual partner (n = 620, 46.1%), whereas 39.3% of participants (n = 522) reported engaging in exchange sex in the past 6 months (Table 1).

Table 2 shows the matrix of bivariate associations between each psychosocial and structural syndemic variable reported at baseline. Most psychosocial and structural variables were significantly correlated with each other. Strong associations were observed between history of childhood abuse and ongoing abuse (PR = 2.02, 95% CI: 1.75–2.34), PTSD and ongoing abuse (PR = 2.09, 95% CI: 1.83–2.38), PTSD and depressive symptoms (PR = 3.22, 95% CI: 2.79–3.73), incarceration and illicit drug use (PR = 2.68, 95% CI: 2.17–3.31), food insecurity and PTSD (PR = 2.04, 95% CI: 1.72–2.44), concentrated housing vacancy and neighborhood low educational attainment (PR = 2.30, 95% CI: 1.88–2.83), and neighborhood black racial concentration and neighborhood unemployment (PR = 3.77, 95% CI: 2.70–5.27).

Table 2.

Bivariate Prevalence Ratios between Each Psychosocial and Structural Syndemic Factor at Baseline

	Ongoing abuse	History of childhood abuse	Depressive symptoms	PTSD	Illicit drug use	Heavy alcohol use	Incarceration	Food insecurity	Unstable housing	Lack of health insurance	Concentrated rental cost burdened	Neighborhood low educational attainment	Neighborhood unemployment	Concentrated housing vacancy	Neighborhood black racial concentration
History of childhood abuse	2.02*** (1.75–2.34)	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Depressive symptoms	1.91*** (1.67–2.18)	1.65*** (1.47–1.85)	—	—	—	—	—	—	—	—	—	—	—	—	—
PTSD	2.09*** (1.83–2.38)	1.87*** (1.67–2.09)	3.22*** (2.79–3.73)	—	—	—	—	—	—	—	—	—	—	—	—
Illicit drug use	1.20* (1.03–1.40)	1.15* (1.00–1.31)	1.97*** (1.70–2.27)	1.76*** (1.49–2.08)	—	—	—	—	—	—	—	—	—	—	—
Heavy alcohol use	1.21* (1.04–1.40)	1.17* (1.03–1.34)	1.60*** (1.38–1.87)	1.69*** (1.43–1.99)	1.78*** (1.46–2.18)	—	—	—	—	—	—	—	—	—	—
Incarceration	1.14 (0.99–1.31)	1.08 (0.96–1.22)	1.45*** (1.24–1.68)	1.37** (1.16–1.61)	2.68*** (2.17–3.31)	1.25* (1.03–1.50)	—	—	—		—	—	—	—	—
Food insecurity	1.86*** (1.61–2.14)	1.47*** (1.30–1.66)	1.97*** (1.68–2.31)	2.04*** (1.72–2.44)	1.64*** (1.34–2.01)	1.39** (1.15–1.68)	1.20** (1.05–1.36)	—	—	—	—	—	—	—	—
Unstable housing	1.23* (1.04–1.46)	1.24** (1.07–1.44)	1.24* (1.03–1.50)	1.53*** (1.26–1.85)	1.92*** (1.55–2.37)	1.38** (1.09–1.73)	1.39*** (1.19–1.61)	1.25** (1.08–1.44)	—	—	—	—	—	—	—
Lack of health insurance	1.12 (0.97–1.30)	1.03 (0.91–1.18)	1.22* (1.04–1.42)	1.05 (0.88–1.26)	1.69*** (1.39–2.06)	1.42** (1.18–1.72)	1.35*** (1.19–1.54)	1.04 (0.92–1.18)	1.39* (1.08–1.80)	—	—	—	—	—	—
Concentrated rental cost burdened	1.22* (1.04–1.43)	1.12 (0.98–1.27)	1.08 (0.91–1.27)	0.95 (0.80–1.14)	0.90 (0.73–1.10)	0.95 (0.78–1.16)	1.01 (0.88–1.15)	0.92 (0.81–1.04)	0.81 (0.62–1.05)	1.34** (1.12–1.60)	—	—	—	—	—
Neighborhood low educational attainment	0.94 (0.82–1.09)	1.01 (0.90–1.14)	1.15 (0.99–1.34)	1.05 (0.89–1.24)	1.24* (1.02–1.51)	0.90 (0.75–1.09)	1.13 (0.99–1.28)	0.96 (0.85–1.08)	0.92 (0.71–1.19)	1.09 (0.93–1.27)	0.86*** (0.79–0.92)	—	—	—	—
Neighborhood Unemployment	1.06 (0.90–1.26)	0.87** (0.76–0.99)	0.94 (0.79–1.12)	0.84 (0.70–1.02)	1.01 (0.79–1.28)	0.90 (0.72–1.10)	0.87* (0.75–1.00)	0.96 (0.84–1.10)	0.49*** (0.38–0.63)	0.85 (0.72–1.01)	1.26*** (1.13–1.39)	1.35** (1.15–1.58)	—	—	—
Concentrated housing vacancy	1.02 (0.87–1.21)	1.03 (0.89–1.19)	1.10 (0.91–1.32)	0.98 (0.80–1.19)	1.20 (0.94–1.54)	1.08 (0.86–1.35)	1.12 (0.95–1.31)	0.88 (0.77–1.00)	1.16 (0.85–1.58)	1.22* (1.00–1.49)	1.38*** (1.24–1.55)	2.30*** (1.88–2.83)	0.97 (0.91–1.04)	—	—
Neighborhood black racial concentration	0.94 (0.75–1.19)	0.95 (0.78–1.16)	1.01 (0.77–1.32)	0.93 (0.70–1.23)	1.30 (0.88–1.93)	1.17 (0.82–1.66)	0.97 (0.78–1.20)	1.08 (0.87–1.34)	0.79 (0.53–1.17)	0.71** (0.57–0.88)	1.41** (1.18–1.70)	0.79** (0.67–0.93)	3.77*** (2.70–5.27)	1.20** (1.05–1.37)	—
Concentrated income inequality	0.90 (0.78–1.03)	1.12 (0.99–1.27)	1.16 (0.99–1.37)	1.07 (0.90–1.28)	1.27* (1.02–1.57)	1.13 (0.93–1.38)	1.08 (0.94–1.24)	1.04 (0.92–1.18)	1.27 (0.97–1.67)	1.00 (0.85–1.17)	0.88** (0.82–0.94)	1.17* (1.04–1.33)	1.04 (0.97–1.10)	0.94* (0.88–0.99)	1.11*** (1.06–1.15)

^* p < 0.05. **p < 0.01. ***p < 0.001.

PTSD, posttraumatic stress disorder.

Table 3 shows the prevalence ratios between each syndemic parameterization and HIV risk behavior. When adjusted for study site, age, annual household income, and education, a higher syndemic score was significantly associated with increased prevalence of unknown HIV status of last male sex partner (aPR = 1.07, 95% CI: 1.04–1.10), exchange sex (aPR = 1.17, 95% CI: 1.14–1.20), and having multiple sex partners (aPR = 1.07, 95% CI: 1.06–1.09). This observation persisted for the psychosocial and participant-level structural syndemic groups. A higher syndemic score in the psychosocial syndemic group was associated with increased prevalence of unknown HIV status of last male sex partner (aPR = 1.12, 95% CI: 1.09–1.16), exchange sex (aPR = 1.27, 95% CI: 1.23–1.32), and multiple sex partners (aPR = 1.11, 95% CI: 1.09–1.14). Similarly, a higher syndemic score in the participant-level structural syndemic group was associated with increased prevalence of unknown HIV status of last male sex partner (aPR = 1.15, 95% CI: 1.09–1.22), exchange sex (aPR = 1.36, 95% CI: 1.29–1.44), and multiple sex partners (aPR = 1.17, 95% CI: 1.13–1.22). A dose–response relationship was observed between the number of syndemic groups and HIV risk behaviors (Table 3). For instance, being in two syndemic groups doubled (aPR = 2.10), the prevalence of exchange sex while being in all three groups tripled (aPR = 3.07) the prevalence of exchange sex. In addition, being in all three syndemic groups was associated with increased prevalence of unknown HIV status of the last male sex partner (aPR = 1.67, 95% CI: 1.43–1.95) and multiple sex partners (aPR = 1.53, 95% CI: 1.36–1.72). There were no statistically significant associations observed between syndemic parameterizations and condomless anal and vaginal sex.

Table 3.

Prevalence Ratios and 95% Confidence Intervals for the Associations between Psychosocial and Structural Syndemic Factors and HIV Risk Behaviors Among Black Women

	Condomless anal sex at last intercourse		Condomless vaginal sex at last intercourse		Unknown HIV status of last male sex partner		Exchange sex		Multiple sex partners
Syndemic variables	PR (95% CI)	aPR (95% CI)	PR (95% CI)	aPR (95% CI)	PR (95% CI)	aPR (95% CI)	PR (95% CI)	aPR (95% CI)	PR (95% CI)	aPR (95% CI)
Cumulative syndemic score	1.00 (0.98–1.01)	0.99 (0.98–1.01)	1.00 (1.00–1.02)	1.00 (0.99–1.01)	1.08 (1.06–1.11)	1.07 (1.04–1.10)	1.20 (1.18–1.23)	1.17 (1.14–1.20)	1.07 (1.06–1.09)	1.07 (1.06–1.09)
Syndemic groups
Psychosocial syndemic group score	1.01 (0.98–1.03)	1.00 (0.98–1.03)	1.01 (1.00–1.03)	1.01 (0.99–1.02)	1.14 (1.11–1.18)	1.12 (1.09–1.16)	1.32 (1.28–1.36)	1.27 (1.23–1.32)	1.11 (1.08–1.14)	1.11 (1.09–1.14)
Participant-level structural syndemic group score	0.99 (0.95–1.03)	0.98 (0.94–1.03)	1.00 (0.97–1.02)	0.99 (0.97–1.02)	1.18 (1.12–1.25)	1.15 (1.09–1.22)	1.44 (1.37–1.53)	1.36 (1.29–1.44)	1.17 (1.13–1.22)	1.17 (1.13–1.22)
Neighborhood-level structural syndemic group score	0.98 (0.96–1.01)	0.98 (0.96–1.01)	1.01 (0.99–1.03)	1.00 (0.98–1.02)	1.02 (0.98–1.07)	1.01 (0.96–1.06)	1.01 (0.96–1.07)	1.01 (0.96–1.07)	1.00 (0.97–1.04)	0.99 (0.96–1.03)
Number of syndemic groups
0–1	REF	REF	REF	REF	REF	REF	REF	REF	REF	REF
2	1.02 (0.92–1.13)	1.00 (0.90–1.11)	1.00 (0.94–1.06)	0.99 (0.93–1.05)	1.47 (1.26–1.72)	1.41 (1.21–1.65)	2.28 (1.84–2.82)	2.10 (1.70–2.59)	1.36 (1.21–1.52)	1.37 (1.22–1.53)
3	0.99 (0.89–1.10)	0.96 (0.86–1.08)	1.00 (0.93–1.06)	0.97 (0.91–1.04)	1.77 (1.52–2.06)	1.67 (1.43–1.95)	3.56 (2.91–4.35)	3.07 (2.51–3.77)	1.55 (1.38–1.73)	1.53 (1.36–1.72)

Estimates where p < 0.05 are bolded to facilitate interpretation. Adjusted models are controlled for study site, age, education, and annual household income. Five models were fit for each outcome: one with cumulative syndemic score, one with psychosocial syndemic group score, one with participant-level structural syndemic group score, one with neighborhood-level syndemic group score, and one with syndemic group number.

CI, confidence interval; PR, prevalence ratio; REF, reference group.

Discussion

This analysis found that among black cisgender women living in U.S. census tracts with higher-ranked HIV prevalence and poverty, a higher syndemic score was significantly associated with increased prevalence of unknown HIV status of last male sex partner, exchange sex, and multiple sex partners. There were also significant associations observed between syndemic groups and study outcomes. Being in multiple syndemic groups significantly increased the prevalence of reporting HIV risk behaviors compared with being in one syndemic group. Furthermore, the cumulative effect of these syndemic factors was associated with multiple HIV risk behavior outcomes. Findings from this analysis provide preliminary evidence that black women have compounded burden from multiple individual and neighborhood-level factors that influence health outcomes, including HIV vulnerability.

Black women experienced a range of psychosocial and structural factors, including depression,^7,8,45 incarceration,¹⁷ history of childhood sexual abuse,^10,12,13 and neighborhood disorder⁴⁶ that have been found to be independently associated with HIV risk behaviors. Many of the psychosocial and structural variables were correlated with each other, providing evidence that the study variables are mutually enhancing, thereby forming a syndemic. On average, women in the analysis reported experiencing seven syndemic variables.

In the analysis, there was a positive relationship between the cumulative syndemic index score and HIV risk behavior outcomes. The number of syndemic factors reported increased as the prevalence of reporting unknown HIV status of the last male sex partner, exchange sex, and multiple sex partners increased. This finding is consistent with previous studies among urban women,³ Latino men who have sex with men (MSM),⁴⁷ and transgender women.⁴⁸ There was no statistically significant relationship observed between the cumulative syndemic index score and condomless sex. This finding was not consistent with previous studies, which found an increased association between higher number of syndemic factors and reporting condomless sex. However, 85% of participants reported condomless sex at the last intercourse; therefore, this unexpected finding may be the result of a homogenous sample. Homogeneity may explain the reason why a statistically signification relationship between cumulative syndemic index score and condomless sex was not observed because there was minimal variation in the outcome variable.

In addition, the findings highlight the role neighborhood characteristics play in the lived experiences of black women and the complex and interrelated connections these factors share with other syndemic factors that have been found to shape HIV risk. In this analysis, structural factors were explored at the individual and neighborhood levels. Previous studies have found a relationship between psychosocial and individual-level structural syndemic effects and HIV risk behaviors;³ however, there are no studies measuring syndemic effects on HIV risk behaviors that have included neighborhood-level structural characteristics in a syndemic index. When categorized into syndemic groups, the neighborhood structural syndemic group was not statistically significant with any of the HIV risk behavior outcomes; however, a statistically significant dose–response relationship was observed between number of syndemic groups and HIV risk behaviors, except condomless anal and vaginal sex. These findings suggest that the cumulative effect of neighborhood-level variables alone does not influence HIV risk but works in tandem with co-occurring psychosocial and individual-level variables to predict HIV risk behavior outcomes.

This analysis is not without limitations. First, the HPTN 064 Women’s HIV Seroincidence Study specifically recruited women from six urban and peri-urban cities who were behaviorally high risk. This affects the external validity of the findings because findings are unable to be generalized to all black women in the United States. Second, because this analysis only included data collected at baseline, the analysis is cross-sectional in nature and therefore is unable to infer causality and temporality. This is important because without the ability to determine causality and temporality, it is impossible to determine the direction of observed associations. However, the present analysis is one of few analyses that establish an association between syndemic effects across varying levels and HIV risk behaviors. Longitudinal analyses are ideal for further examining the relationship between syndemic effects and HIV risk behaviors over time. Given the opioid overdose epidemic that dates back to 2013 and precedes the HPTN 064 study period, patterns of substance abuse by the study’s target population may have changed and should be taken into consideration when interpreting study results. Also, this analysis focuses on HIV risk behaviors; however, further research is needed to explore the relationship of syndemic effects on new HIV infections. HIV incidence among black women in the analytic sample was 0.7%, thereby limiting the ability to use HIV incidence as a study outcome. Finally, this analysis models the additive effect of syndemic factors on HIV risk behaviors. This additive syndemic model assumes every factor in the model carries the same weight. This is a limitation because there may be different combinations of factors that increase or decrease risk; however, we are limited in our ability to assess this using an additive syndemic model. Further research is needed to identify and define HIV risk profiles across multiple dimensions of risk behaviors among black women.

Conclusion

This analysis provides evidence that psychosocial and structural factors work in tandem and form a syndemic and that the plurality of HIV disparities among black women are disparities stemming from the convergence of several psychosocial and structural factors. The structural factors explored in this analysis provide evidence of a relationship with black women’s vulnerability to HIV. These structural factors extend beyond those used in this analysis and include institutional and policy-related structural factors that would need to be reformed if we are to effectively address the socio-structural issues that drive HIV disparities for black women. Findings emphasized the dire need for macrolevel policy initiatives such as workforce development programs to reduce poverty and increase incomes, and mandated policies aimed at reducing discrimination when hiring individuals who have been previously incarcerated—that is, policy initiatives specifically targeting structural systems that negatively impact health.

Findings demonstrate the need for the application of a syndemic framework in further analyses to better define HIV risk profiles among black women, as well as in the development and implementation of HIV prevention programs. The development of HIV prevention programs has often focused on changing the behavior of the targeted population; however, such strategies may be obsolete when it comes to HIV prevention among black women. HIV prevention models and strategies should be developed with a socio-structural framework in mind, thereby considering the social contexts that intersect with the lived experiences of black women.

Author’s Contribution

T.D.: Conceptualization (supporting); Writing—review and editing (equal). D.H.: Conceptualization (supporting); Writing—review and editing (equal). T.M.: Writing—review and editing (equal). M.M.: Writing—review and editing (equal). Q.N.: Writing—review and editing (equal). R.T.: Methodology (supporting); Writing—review and editing (equal). L.W.: Conceptualization (lead); Data Curation (lead); Formal Analysis (lead); Methodology (lead); Visualization (lead); Writing—original draft preparation (lead); Writing—review and editing (equal).

Author Disclosure Statement

No competing financial interests exist.

Funding Information

Overall support for the HPTN is provided by the National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH), National Institute on Drug Abuse (NIDA), the National Institute of Mental Health (NIMH) under Award Numbers UM1AI068619. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.