Prediction of HIV Sexual Risk Behaviors among Disadvantaged African American Adults using a Syndemic Conceptual Framework

Abstract

The focus of this paper is on HIV sexual risk taking among a community-based sample of disadvantaged African American adults. The objective is to examine multiple factors associated with sexual HIV risk behaviors within a syndemic conceptual framework. Face-to-face, computer-assisted, structured interviews were conducted with 1,535 individuals in Atlanta, Georgia. Bivariate analyses indicated a high level of relationships among the HIV sexual risks and other factors. Results from multivariate models indicated that gender, sexual orientation, relationship status, self-esteem, condom use self-efficacy, sex while the respondent was high, and sex while the partner was high were significant predictors of condomless sex. Additionally, a multivariate additive model of risk behaviors indicated that the number of health risks significantly increased the risk of condomless sex. This intersection of HIV sexual risk behaviors and their associations with various other behavioral, socio-demographics, and psychological functioning factors helps explain HIV risk-taking among this sample of African American adults and highlights the need for research and practice that accounts for multiple health behaviors and problems.

INTRODUCTION

The HIV/AIDS epidemic remains a significant global public health concern, including in the U.S. Here we note the disproportionate impact of the HIV/AIDS epidemic on African Americans, especially those who live in disadvantaged inner-city neighborhoods and who themselves are vulnerable as well. Recent data show the incidence rate for HIV among African Americans to be 8.6 times greater than the rate among whites.(1) Currently, African Americans also have the highest lifetime rates of being diagnosed with HIV, being diagnosed as stage 3 (AIDS), and mortality from HIV/AIDS of all ethnic groups.(1) The highest rates of new infections of HIV occur in the southeastern United States. The state of Georgia has the fourth highest rate of new HIV infections in this region, with an incidence rate that is 50.2% higher than for the Southeast as a whole.(1) Within the state of Georgia, African Americans comprised 69.6% of HIV incidence. Incidence rates for HIV within Atlanta mirror those for the state as a whole, and cluster in Atlanta’s disadvantaged neighborhoods.(1–3)

Studies have shown that condomless sex currently is the dominant mode of HIV transmission among African Americans.(4) Other HIV sexual risk behaviors include having multiple sex partners, sex with at-risk partners, men who have sex with other men, the lack of proper or consistent condom use, and sex in high-risk settings or while under the influence of alcohol or other drugs.(5–13) HIV sexual risk behaviors also vary by type of partner and setting;(14–18) with well-known differences between main and casual, including paying partners.(19, 20) Findings from studies among women show condomless sex to be more common with main partners,(21) which is contrary to findings among men where condomless sex seems more common with casual partners.(22, 23) Individual factors (e.g., unaware of partner’s HIV status, condom attitudes, concurrent sexual partnerships);(24), sex partner characteristics (e.g., partner’s HIV status, partners’ drug use), and contextual issues (e.g., stigma, discrimination, access to HIV testing) all contribute to the need for programs promoting condom use.(25) However, regardless of the correlates, what may be the most important prevention strategy to reduce HIV transmission among is the promotion of consistent condom use.(26) In spite of progress toward other means of HIV prevention, condom use currently remains the primary tool for effective prevention of sexual HIV transmission.(27–31) Considering this, research has focused on factors that promote or inhibit correct condom use among at-risk African Americans.(32, 33)

Research has identified several factors that influence condom use among at-risk African American women, including partner communication, training in the negotiation of safe sex,(32) fear of partner communication because of perceptions of possible partner violence and coercion,(34–36) feelings of self-control and negotiation of condom use,(37, 38) beliefs about long-term relationships and monogamy,(39) and social norms about condom use.(40–42) Among at-risk African American men different factors were found to influence condom use. Namely, communication and partner communication factors seem less important and other concepts such as machismo and structural factors such as poverty, health care disparities, and incarceration may be more important.(33)

Due to the increasing recognition of the complexity of many public health and social problems, more comprehensive approaches to understanding the underlying dynamics are gaining support. The main thesis in the syndemic conceptual framework is that multiple epidemics co-occur, that various risk factors interact with one another, and that each of these worsens the effects of the others.(43, 44) That is, health problems may be linked so they interact and contribute to the overall disease burden in populations in which they co-occur.(44–49) This conceptualization of disease burden also recognizes that disease does not occur in isolation from other conditions, health behaviors, and social problems. These intersections often have magnifying effects on health problems in both individuals and communities.(49)

The focus of this paper is on a conceptual framework to predict HIV sexual risk behaviors with other behaviors including illicit drug use, binge drinking, and depressive symptoms. These behaviors, along with socio-demographics, childhood maltreatment, and psychological functioning factors are hypothesized to intersect and have an additive effect on the degree to which African American adults from disadvantaged communities engage in HIV sexual risk behaviors. Often these conditions have been found to occur in populations that have faced health inequities. Singer (2006) argued that syndemic features may be key factors in racial health disparities in the U.S.(45) Singer and his team (2003; 2009) also pointed out that health behaviors have the potential to interact at multiple levels of influence (e.g., individual, interpersonal, community) to decrease health status and increase morbidity and mortality.(44–46) There are several types of syndemic conceptual frameworks described in the literature that have related specifically to HIV/AIDS risks. Most informative and appropriate for the current study are the substance abuse, violence, AIDS (SAVA) (43, 50), mental health and HIV/AIDS,(51) and drug use and crack cocaine use syndemic conceptual frameworks.(52) Across each of these hypothesized frameworks there is much commonality in that the effects of each of the key factors (e.g. substance use, context, other health behaviors, and HIV/AIDS) intertwine and combine to increase disease burden.(43) The impact of a syndemic is likely to be more profound among vulnerable groups such as those without access to healthcare, those living in poverty, or racial and ethnic minorities.(44, 53) Syndemic conceptual frameworks have been shown to provide a better understanding of HIV risk behaviors such as sexual compulsivity, childhood sexual abuse, depressive symptoms and intimate partner violence.(54–56)

Increasingly, researchers have applied syndemic conceptual frameworks to investigate HIV risk among men-who-have-sex-with-men (MSM).(49, 51, 56–59) Inquiries using these frameworks have commonly explored co-occurring substance abuse, depressive symptoms, and victimization experiences and their relationship to HIV sexual risk behaviors.(48, 56) This research has identified risk factors such as alcohol use, illicit drug use, intimate partner violence, childhood maltreatment, sexual risks such as unprotected anal sex, and sex with HIV-positive partners as inter-related and additive.(49, 51, 56–59) Most specific to this paper, Klein (2011) found a syndemic conceptual framework to be important in describing the complex relationships between behavioral and psychosocial factors which influence the proportion of sex acts that involved the use of protection.(49) Across these studies, those reporting higher numbers of these HIV risk behaviors and negative psychosocial characteristics have been associated with greater odds of HIV seropositivity.(51, 56, 58, 60)

Singer has examined the syndemic conceptual framework among his various studies including reviews of the literature and primary data collections of disadvantaged African Americans. For instance, a study of young African American heterosexual men found that condom use is often related to disparities in social conditions and intersections with other health problems within the individual and community.(45) Likewise, a review of the literature pertaining to African American women and their HIV risk revealed intersecting and additive behaviors and conditions which contribute to HIV vulnerability among this group.(61) Most specific to this study, a recent paper with a majority African American sample recruited from an urban sexually transmitted infection (STI) clinic examined relations among childhood sexual abuse, depressive symptoms, binge drinking, marijuana use, intimate partner violence, sexual risk behavior, and STI diagnosis.(47) Results indicated interrelations among HIV risk behaviors in the sample, and that persons with greater numbers of HIV risk behaviors were more likely to have multiple sex partners and to be diagnosed with an STI.(47) However, because participants for this study were recruited from a single STI clinic, results may have limited generalizability to other African Americans recruited from other venues, particularly nonclinical sites.

In this study we examine the associations between HIV sexual risk behaviors and binge drinking, depressive symptoms, socio-demographics, childhood maltreatment, and psychological functioning factors, their co-occurrence, and how a combination of these may correspond with an increase in HIV sexual risk behaviors. In addition to investigating HIV sexual risk behaviors through the study of these particular outcome measures, we also contribute to the literature by applying a syndemic conceptual framework based upon domains from the literature which are hypothesized to affect engagement in risk behaviors: socio-demographic characteristics, childhood maltreatment experiences, and psychosocial functioning. The HIV sexual risk behaviors measures are included in this study to assess their intersections and to determine if the associations are consistent across domains. Based on previous research linking HIV sexual risk behavior outcomes within the syndemic conceptual framework, which emphasizes the interconnected nature of these behaviors, we hypothesize that the HIV sexual risk behaviors under consideration will exhibit interconnectedness so that increasing the frequency of these risk factors will in turn be associated with increases in condomless sex.

METHODS

Data for this paper were collected as part of a larger study, Be Healthy, a longitudinal observational study of people’s perceptions of how their neighborhood impacted their daily lives and actions.(62) This paper focuses on baseline data, collected between January 2010 and October 2011. Be Healthy eligibility criteria included: being 18 years or older, self-identifying as African American or black, and having lived in the study area; a disadvantaged area of Atlanta, GA,(63–65) for at least 12 months prior to the interview phase. Active street outreach (based on ethnographic information and interviews done with key informants) and passive strategies (e.g., posting flyers in public places such as telephone poles, message boards inside of local bars, and message boards inside of some local stores as well as referrals made by previous study participants) were used to recruit 1,535 participants from 61 census block groups (CBGs). The census blocks chosen for inclusion in the study were selected based on neighborhood structural characteristics as reported in the 2000 U.S. Census Data and based on data from the Atlanta Police Department. Consistent with the study’s conceptual framework and previous research findings (66, 67) the neighborhood structural characteristics were selected on the basis of seven specific criteria: (1) the percentage of household incomes that were reported to be more than 20% above or below the federal poverty level, (2) the percentage of adults who had not completed high school or its equivalent, (3) the percentage of female-headed households, (4) the percentage of people who were unemployed or not in the labor force, (5) the percentage of one-unit housing structures, (6) the percentage of owner-occupied households, and (7) the percentage of vacant housing. Within the selected census block groups, a non-probability sampling frame was designed to ensure sufficient variability by gender, age (specifically, persons who were under the age of 35 and those aged 35 or older, to facilitate analytical comparisons based on younger versus older adults), and drug use (i.e., persons who had not used any illegal drugs during the previous 90 days versus those who had). In this context, a “drug user” was defined as someone who had used powder cocaine, crack cocaine, and/or heroin at least once during the previous week and at least four times during the 90 days prior to interview. These criteria were imposed so that people designated as drug users for the purposes of this research were active, ongoing users of the illegal drugs in question, as opposed to one-time experimenters or casual, infrequent users of these substances. Conversely, in order to be considered a nonuser for this study, people had to report no use of these same drugs during the five years prior to interview. The criteria for being a non-drug user were selected based upon previous research that support using the five-year time frame as a proxy for the length of time it takes to develop a drug-free lifestyle (68–70). Beyond ceasing drug use, this also considers a shift away from a drug use-associated lifestyle (e.g., greater involvement in mainstream social roles). In order to be considered eligible for participation, respondents had to self-identify as African American, be at least 18 years of age, and have lived in that same neighborhood or census block group continuously for at least one year. People were considered ineligible for the study if they: (1) were in a drug treatment program or any other institutional setting at the time of recruitment, (2) were intoxicated at the time of consent or interview, or (3) displayed signs of cognitive impairment at the time of consent or interview.

After acquiring informed consent, trained interviewers invited study participants to a local research site, where they completed computer assisted surveys that included questions from various domains, including socio-demographic characteristics, psychological/psychosocial functioning, childhood maltreatment, reproductive health, sexual history and recent sexual activity, and substance use. Participants were compensated $30 for their time (typically one to two hours) and then were offered referrals to local health and/or social service agencies as appropriate/necessary. The Emory Institutional Review Board approved the study protocol.

MEASURES

Socio-demographic Characteristics

These include age (continuous), gender (male versus female), educational attainment (high school or less versus all other categories), sexual orientation (heterosexual versus gay, lesbian, or bisexual), relationship status (“involved” with someone versus not “involved”), and living situation (own or rent a private residence, living in someone else’s residence), and employment status (unemployed, employed part time, employed full time). Yearly income was measured through a series of questions about the amount of income received in the past 30 days from a variety of sources including legal employment, “under the table” income, public assistance, retirement benefits, unemployment benefits, family sources, illegal income, and other sources. A total was calculated and the participant was asked to confirm the amount. For the purposes of this study, we present income as a categorical variable for descriptive reasons, but the square root of income was computed to make the variable conform to a normal distribution.

Dependent Variables

A measure of condom use during vaginal and/or anal sex with steady partners and casual partners was derived from four questions. Initially, participants were asked to indicate the number of times that they engaged in vaginal sex with a steady partner during the 90 days prior to the interview. Then, they were asked to indicate how many of those times they or their partner(s) used a male condom. Comparable questions were asked for vaginal sex with casual partners, anal sex with steady partners, and anal sex with casual partners. For analytical purposes in the present paper, the responses to these questions were dichotomized to indicate condomless sex (i.e., less than 100% condom use, coded as (1) or consistent condom use (i.e., condoms used 100% of the time, coded as (0).

Secondary HIV sexual risk behavior measures included: (1) multiple sex partners in the past 90 days, (2) any sex while the respondent was high, (3) and any sex with a high partner. To assess multiple sex partners in the past 90 days participants were asked the number of men and the number of women they had sex with in the past 90 days. Responses were combined for male and female partners and then dichotomized into a single measure indicating had versus had not had multiple sex partners. Participants were also asked to report the frequency of which they had sex while high and had sex with a high partner in the past 30 days. Responses to these items used a five-point ordinal range, including “never,” “about once per month,” “a few times per month,” “about once per month,” “several times per week” or “daily.” Responses were dichotomized and then used to create two measures indicating and/no sex while high during the previous 30 days and any/no sex with a high partner during the previous 30 days.

Predictors

Childhood Trauma

Items from the brief screening version of the Childhood Trauma Questionnaire (71) were used to examine childhood maltreatment experiences, which asked participants to report on the experiences before the age of 18. Five items were used to assess the level of childhood sexual abuse experienced (Cronbach’s alpha = .94); five items were used to measure physical abuse (Cronbach’s alpha = .76); and five items were used to assess emotional abuse (Cronbach’s alpha = .81). Responses to these items used a five-point ordinal range, including “never,” “rarely,” “sometimes,” “often,” or “very often.” Responses were dichotomized to represent any versus no abuse for each form of childhood maltreatment.

Psychological characteristics

Depressive Symptoms

The Center for Epidemiologic Studies Depression Scale (CESD-20) was used to screen for depressive symptoms.(72) The CES-D consists of 20 self-reported items related to the number of days during the past week during which the participant experienced emotional or behavioral difficulty. As with other research into the validity of the CES-D using different cut-points, those with scores higher than 22 were considered to indicate depressive symptoms.(56, 73, 74) The CESD-20’s reliability among this sample was found to be acceptable (Cronbach’s alpha = .87; mean = 14.5, SD = 10.7).

Self-esteem was assessed using the 10-item Rosenberg Self Esteem scale.(75) Response options ranged from “strongly disagree” (coded 1) to “strongly agree” (coded 5). Scores ranged from 20 to 50, with higher scores representing higher levels of reported self-esteem (Cronbach’s alpha = .82; mean = 39.5; SD = 5.5).

Condom use self-efficacy was assessed using selected items from Brafford and Beck’s (1991) Condom Use Self Efficacy Scale.(76) In this study, we used 13 items to create an overall scale measuring condom use self-efficacy, with individual items scored using a 5-point Likert scale with responses ranging from “strongly disagree” (coded 1) to “strongly agree” (coded 5). The items included measurements of confidence in purchasing condoms, remembering to use condoms in different circumstances, suggesting condom use with partners, persistence in one’s efforts to use condoms, and so forth. Scores ranged from 13 to 52, with higher scores representing higher levels of reported condom use self-efficacy (Cronbach’s alpha = .81; mean = 39.7; SD = 6.2).

Substance use

Illicit drug use

Use of five specific types of illicit drugs in the past 90 days were assessed: crack, cocaine, heroin, ecstasy, and marijuana. For each illicit drug, participants were asked to indicate yes/no to the question: “Have you used ___ in the past 90 days?” Responses were recoded into a single dichotomous variable across drug types to indicate any/no illicit drug use in the past 90 days. Binge Drinking. Participants reported the number of days in the past 90 “How many days did you drink alcohol in the past 90 days?” and the average number of drinks per session “On an average day when you drink, how many drinks do you have?” Using cutoffs of 4 drinks for women per setting and 5 for men,(77) participants were classified as having at least one binge drinking session in the past 90 days versus no binge drinking days.

Statistical Analysis

The objective of this study was to examine the associations between HIV sexual risk behaviors and binge drinking, depressive symptoms, socio-demographics, childhood maltreatment, and psychological functioning factors, their co-occurrence, and how a combination of these may correspond with an increase in HIV sexual risk behaviors. Analyses to address these research aims were undertaken in steps that mirror the approach taken by Parson, Grov, & Golub (2012) and others.(51, 56, 58) First, descriptive statistics were used to characterize personal socio-demographics, reports of condomless sex, multiple sex partners, sex while the respondent was high, and sex with a high partner, and illicit drug use, binge drinking, and depressive symptoms. Second, Pearson’s r correlations were calculated across pairings of behaviors to determine any possible intersections among these behaviors and establish the associations between pairs of HIV sexual risk behaviors. Third, multivariate logistic regression models were used to determine the associations between the predictors and HIV sexual risk behaviors. Predictors were selected for this model to represent the each of the hypothesized domains as tested in the literature.(49, 51, 56–59) Inter-predictor correlations and variance inflation factor (VIF) values were produced to check the assumption of multicollinearity among these predictors. Models were also computed predicting reporting of multiple sex partners, sex while the respondent was high, and sex with a high partner for comparison to results for condomless sex. Finally, reports of self-reported condomless sex were also calculated for the number of positive responses to HIV risk behaviors including: any illicit drug use, any binge drinking, depressive symptoms, multiple sex partners, any sex while high, and any sex while partner high categorized into 0, 1, 2, and 3 or more. A logistic regression model including the count of HIV risk behaviors as a predictor was used to test the strength and direction of the association. Logistic regressions with pairwise interactions of the variables were computed using grand mean centered variables to determine if these interactions predicted greater condomless sex.

RESULTS

Sample Description

As shown in Table I, among the 1,535 participants, the average age was 37.5 (SD = 13.2), 57.3% were male, 58.9% had an income of less than $8,500 per year, 77.1% had a high school education or less, 92.7% were heterosexual, 56.3% reported being partnered in a relationship, only 8.4% owned their own home, and 74.4% were unemployed. Also, 57.9% of participants reported childhood emotional abuse, 26.8% reported childhood sexual abuse, and 72.4% reported childhood physical abuse. A total of 67.8% reported condomless sex with sexual partners in the past 90 days, 39.9% reported multiple sexual partners, 55.0% reported having sex while high, and 36.3% reported sex with a high partner in the past month. A high prevalence other HIV risk behaviors and conditions were also reported by the participants; 36.3% reported illicit drug use and 28.2% reported binge drinking in the past 90 days, and 19.7% reported depressive symptoms.

Table I.

Study demographic variable descriptive statistics (n=1,535).

	Overall
	Mean (SD)
Age	37.5 (13.2)
	n (%)
Yearly Income
Up to $2,500	321 (20.9)
$2,501 – $5,500	301 (19.6)
$5,501 – $8,500	282 (18.4)
$8,501 – $14,500	342 (22.3)
More than $14,500	287 (18.7)
Gender
Male	879 (57.3)
Female	656 (42.7)
Educational Attainment
Less than high school	585 (38.1)
High school graduate/GED	599 (39.0)
At least some college	351 (22.9)
Sexual orientation
Heterosexual	1410 (92.7)
Gay, lesbian, or bisexual	111 (7.3)
Relationship Status
Not partnered	670 (43.7)
Partnered	864 (56.3)
Employment Status
Unemployed	1063 (74.4)
Employed, part-time	233 (16.3)
Employed, full-time	133 (9.3)
Any Childhood Trauma
Emotional abuse	887 (57.9)
Sexual abuse	410 (26.8)
Physical abuse	1108 (72.4)

Bivariate Associations and Odds Ratios among HIV Risk Behaviors

Table II presents Pearson’s r correlations among for the HIV risk behaviors. As shown in Table II, thirteen of the fifteen correlations that were computed among the other HIV risk behaviors were statistically significant and all except one of these associations were positive. For example, condomless sex was associated with a greater likelihood of binge drinking (r = .08, p<.01), a greater likelihood of experiencing depressive symptoms (r = .06, p<.05), a greater chance of having sex while the respondent was high (r = .11, p<.001), a greater chance of having sex with a partner who was high (r = .09, p<.001), and a diminished likelihood of recently having had more than one sex partner (r = −.06, p<.05). In general, the size of these relationships were found to be what could be described as small effect sizes.(78) Additionally, the results suggest that relationships between those who participated in illicit drug and binge drinking was .15 (p < .001), depressive symptoms was .08 (p < .01), having multiple sexual partners was .26 (p < .001), sex while the respondent was high in the past month was .42 (p < .001), and sex with a high partner in the past month was .30 (p < .001); suggesting that as illicit drug use increases, the other HIV risk behaviors also increase. The size of these relationships were found to be what could be described as small to medium effect sizes.(78) Overall, there were consistent statistically significant and positive prediction of increasing levels of HIV risk behaviors associated with participation in other behaviors.

Table II.

Correlations Among Psychosocial HIV Risk Behaviors and HIV Sexual Risk Behaviors

	Any Illicit Drug Use	Any Binge Drinking	Depressive Symptoms	Multiple Sex Partners	Any Sex while High	Any Sex with High Partner	Condomless Sex
Any Illicit Drug Use		.15***	.08**	.26***	.42***	.30***	0.03
Any Binge Drinking			.11***	.11***	.26***	.15***	.08**
Depressive Symptoms				.06*	.05	.08**	.06*
Multiple Sex Partners					.28***	.35***	−.06*
Any Sex while High						.46***	.11***
Any Sex with High Partner							.09***

Note:

^*= p < .05,

^**= p < .01,

^***= p < .001

Multivariate Logistic Regression Analyses of HIV Sexual Risk Behaviors

This paper also examined the co-occurrence of risk factors and how a combination of these may correspond with an increase in HIV sexual risk behaviors among disadvantaged African American adults. The multivariate model predicting condomless sex with all predictors entered was statistically significant overall x² (14) = 201.57, p < .001, Nagelkerke R² = 0.19. Correlations (e.g. no values above .60) and VIF values (e.g. no values approached 10) indicated no problems with multicollinearity among predictors.

As can be seen in Table III, women (adjusted odds ratio (p < .05), those who reported being in a steady relationship (p < .001), heterosexuals (p < .05), with higher levels of self-esteem (p <.01), those reporting sex while high (p < .01), and sex with a high partner (p < .01) were more likely to report condomless sex. Those with higher condom use self-efficacy were less likely to engage in condomless sex (p < .001).

Table III.

Multivariate Logistic Regressions Among Intersecting HIV Sexual Risk Behaviors

	Dependent Variables
Independent Variables	Multiple Sex Partners AOR (95% CI)	Any Sex while High AOR (95% CI)	Any High Sex Partner AOR (95% CI)	Condomless Sex AOR (95% CI)
Gender (1=female)	0.47 (0.35, 0.62)***	0.58 (0.43, 0.77)***	1.69 (1.26, 2.27)***	1.32 (1.01, 1.73)*
Sexual orientation (1=heterosexual)	0.20 (0.12, 0.35)***	1.43 (0.80, 2.55)	0.49 (0.29, 0.84)**	1.65 (1.02, 2.67)*
Relationship status (1=partnered)	0.32 (0.24, 0.41)***	1.23 (0.92, 1.65)	0.73 (0.55, 0.98)*	2.97 (2.29, 3.85)***
Childhood Sexual Abuse	1.22 (0.89, 1.67)	0.96 (0.69, 1.35)	1.39 (1.01, 1.91)*	1.32 (0.97, 1.81)
Childhood Physical abuse	1.10 (0.81, 1.49)	1.27 (0.93, 1.75)	1.18 (0.86, 1.62)	1.30 (0.97, 1.73)
Childhood Emotional Abuse	0.95 (0.72, 1.26)	1.05 (0.78, 1.41)	0.99 (0.74, 1.31)	0.96 (0.73, 1.25)
Self Esteem	0.97 (0.94, 1.00)*	1.01 (0.98, 1.04)	1.00 (0.97, 1.03)	1.04 (1.01, 1.07)**
Condom Use Self-Efficacy	0.99 (0.97, 1.01)	1.00 (0.98, 1.03)	1.00 (0.98, 1.03)	0.90 (0.88, 0.93)***
Any Illicit drug use	1.88 (1.39, 2.54)***	4.60 (3.47, 6.11)***	1.74 (1.26, 2.40)**	0.91 (0.68, 1.21)
Any Binge drinking	0.99 (0.74, 1.32)	3.23 (2.35, 4.45)***	0.93 (0.69, 1.24)	1.30 (0.97, 1.75)
Depressive Symptoms	0.95 (0.66, 1.36)	1.07 (0.73, 1.58)	1.07 (0.74, 1.55)	1.35 (0.93, 1.95)
Multiple Sex Partners	----	1.80 (1.33, 2.45)***	2.90 (2.16, 3.88)***	0.78 (0.58, 1.04)
Any Sex while High	1.72 (1.27, 2.34)**	----	6.54 (4.73, 9.03)***	1.51 (1.12, 2.05)**
Any Sex while Partner High	2.87 (2.15, 3.84)***	6.69 (4.84, 9.23)***	----	1.59 (1.17, 2.17)**
Condomless Sex	0.81 (0.60, 1.08)	1.49 (1.10, 2.02)**	1.61 (1.19, 2.19)**	----
Model r2 (Nagelkerke)	0.34	0.45	0.38	0.19

Note:

^*= p < .05,

^**= p < .01,

^***= p < .001

In order to further elucidate the findings for condomless sex an additional three models were also computed to predict multiple sex partners, sex while the respondent was high, and sex with a high partner. The intent of these models was to examine the relative stability of predictors across each of these HIV sexual risk behaviors and if they are consistent with our results for condomless sex. Overall models for multiple partners x² (14) = 399.42, p < .001, Nagelkerke R² = 0.34, sex while the respondent was high x² (14) = 552.42, p < .001, Nagelkerke R² = 0.45, and sex with a high partner x² (14) = 437.99, p < .001, Nagelkerke R² = 0.38 were statistically significant. Table III reveals that women reported less instances of having multiple partners (AOR = 0.47, p < .001), less sex while the respondent was high (AOR = 0.58, p < .001), but a higher likelihood of a sex partner that was high (AOR = 1.69, p < .001). Those who reported being heterosexual were less likely to report multiple sex partners (AOR = 0.20, p < .001) and having a sex partner who was high (AOR = 0.49, p < .01). Those who reported having a relationship partner were less likely to report multiple partners (AOR = 0.32, p < .001), and having a high sex partner (AOR = 0.73, p < .05). Illicit drug users were more likely to report multiple sex partners (AOR = 1.88, p < .001), sex while the respondent was high (AOR = 4.60, p < .001), and a high sex partner (AOR = 1.74, p < .001). Last, with the exception of multiple sex partners and condomless sex in their respective models, the HIV sexual risk behaviors were largely statistically significant predictors of each other.

Among the sample 14.1% (n = 217) reported zero, 19.7% (n = 303) one, 19.5% (n = 300) two, and 46.6% (n = 715) three or more HIV risk behaviors. A multivariate logistic regression model was computed to explore the association of an increasing number of HIV risk behaviors on condomless sex. Predictors used in earlier models were also included in these models. The multivariate model predicting condomless sex with all predictors entered was statistically significant overall x² (14) = 181.59, p < .001, Nagelkerke R² = 0.17. Those having three or more HIV risk behaviors AOR 1.72 (1.17, 2.53, p < .01) were more likely to report condomless sex than those who did not report any of the HIV risk behaviors. Additionally, women AOR 1.47 (1.08, 1.83, p < .05), those reporting having a steady partner AOR 3.11 (2.42, 4.01, p < .001), and those who were heterosexual AOR 1.65 (1.03, 2.64, p < .05), had higher levels of self-esteem AOR 1.04 (1.01, 1.06, p < .01), and those with higher condom use self-efficacy were less likely to report condomless sex AOR 0.90 (0.88, 0.93, p < .001). Finally, models were computed that tested interactions between HIV risk behaviors and condomless sex. None of the pairwise interactions between HIV risk behaviors were statistically significant predictors of condomless sex.

DISCUSSION

In this study we examine the associations between HIV sexual risk behaviors and binge drinking, depressive symptoms, socio-demographics, childhood maltreatment, and psychological functioning factors, their co-occurrence, and how a combination of these may correspond with an increase in HIV sexual risk behaviors. High rates of the use of alcohol and other drugs as well as health problems (specifically depressive symptoms) have been identified among residents in disadvantaged urban neighborhoods, many of who tend to be members of racial/ethnic minority population, live below the poverty level, and experience social and health challenges. It appears that the conditions of the participant’s life create an underlying foundation for the co-occurring conditions that form a syndemic. From a public health perspective such a co-occurrence reveals a syndemic.(45) For example, existing research findings show that African American adults who are disadvantaged at the individual as well as the neighborhood level are more likely than those in more advantaged situations to report depressive symptoms,(79) high levels of alcohol use, (80, 81) illicit drug use,(82–84) and engagement in unsafe sex that places them at risk for HIV infection.(16, 24–26, 62) In this paper we found that greater numbers of co-occurring socio-demographic, socio-psychological, and sexual behaviors intensify reports of condomless sex.

Limitations of the present study include its cross-sectional nature, thereby not allowing for making any causal inference. Second, study participants self-reported information on sensitive topics such as childhood maltreatment and sexual behaviors. This may have resulted in bias due to recall and social desirability. Third, is the limitation that the results of the present study have uncertain generalizability to other disadvantaged African American adults. The purposive convenience methods of recruitment also preclude an assessment of the representativeness of the sample. Last, we recognize that the complexity of the HIV epidemic cannot be limited to individual HIV risk behaviors. Future analyses should include structural and socio-contextual factors such as racism and access to healthcare, which likely put urban at-risk African Americans at greater HIV risk.

Consistent with past research, a large majority of participants (67%) reported condomless sex.(26, 85) The men and women in this study indicated a similar frequency of substance use, including illicit drugs and binge drinking, that has been reported in studies of African Americans living in disadvantaged areas of Atlanta.(19, 86) In addition, approximately 20% of the respondents reported depressive symptoms and over 70% reported at least one type of negative childhood maltreatment. Based upon our findings, this study provided some evidence supporting the application of a syndemic burden that is linked to elevated levels of HIV sexual behaviors among disadvantaged African American adults. Socio-demographics and an increasing number of socio-psychological challenges increased the likelihood of condomless sex. Hence it important to encourage at-risk African American adults to seek HIV testing so that they may become aware of their HIV status and be connected to available social and health services, including the continuum of care available to HIV infected individuals.

The results illustrate that there is inter-relatedness among HIV risk behaviors for urban African American adults. For instance, although our results did not indicate that illicit drug use was directly associated with condomless sex, our preliminary analyses found bivariate relationships between illicit drug use and each of the other detrimental health behaviors. Additionally, we did find that reporting sex while the respondent was high and sex with a high partner to be strongly related to condomless sex. The prevalence of illicit drug use among African Americans is 11% and this rate has increased significantly since 2002, when it was reported to be 9.7%.(87) Furthermore, the use of crack cocaine has also been shown to be related to HIV risk (88–93) and may be a key risk factor for various infectious diseases,(52) and cocaine, marijuana, and heroin are the dominant drugs of abuse in the Atlanta area.(3) Crack use is often associated with greater involvement in risk behaviors and with HIV/AIDS, especially in disadvantaged neighborhoods.(63, 94–98) Although we used condomless sex as our primary outcome variable, our research corroborates studies using binge drinking, illicit drug use, and the other predictors we explored in our models.

The syndemic conceptual framework used for the main analyses in this study posited intersecting influences hypothesized to affect the extent to which disadvantaged African American adults engage in risk behaviors: socio-demographic characteristics, childhood maltreatment experiences, psychosocial functioning, and substance use. Among the distinct contributions of this study are the nature of the study sample, which represented disadvantaged African American adults who live in disadvantaged neighborhoods. The relationship between social disadvantage and unhealthy behaviors as well as poor health is structural in nature. Krieger (1999) introduced the concept of ecosocial factors.(99) Few studies have used a sampling frame that has considered unhealthy behaviors, in this case condomless sex and other HIV sexual risk behaviors, from an ecosocial (100) or risk environment perspective.(101). Moreover, studies reporting on a community-based, heterosexual sample are limited.(8, 45) The inclusion of childhood maltreatment as a more distal risk factor compared to more recent ones also has not received much investigation. (49, 55)

The findings from the current study reveal the presence of a syndemic or synergistic relationships between distal and current social and social-psychological characteristics and HIV sexual risk behaviors, specifically condomless sex. Moreover, the findings suggest the need for comprehensive and innovative prevention and intervention efforts, those that move beyond isolating HIV sexual risk taking. Condomless sex is triggered by individual perceptions (e.g., the safety of condoms or the sexual experience with a condom), sex partner dynamics (e.g., gender role expectations and the inherent message that proposing condom use may give), and community norms (e.g., regarding condomless sex). It is essential that such interventions consider the realities and underlying causes of social and health disparities, including those at the individual, dyad, network, community-level, and society at large. In addition, it is important to expand these programs not only to cover multiple health conditions, but also consider multi-level population-based approaches to research and care based upon the complex individual, dyadic, familial, and social determinants that put African Americans at higher risk of HIV infection. Furthermore, an enhanced understanding of how disparities are supported and sustained within the cultural and political societal structures and climate is needed. This includes consideration of perceived discrimination and its link to health disparities.(102) Doing so will ensure that prevention and intervention efforts are prepared to address features of a syndemic and move beyond behavioral change into elimination of factors that contribute to social and health disparities. Such structural interventions must address poverty, discrimination, living conditions, employment opportunities, and access to health care as well as changes in policies and legal approaches that result in loss of employment, dismantling families and support networks, among other major disruptions that further push the disadvantages away from mainstream society and its privileges.