The Influence of Housing Status and Food Insecurity on a Behavioral HIV/STI Prevention Intervention for Black Women under Community Supervision in New York City: A Moderation Analysis

Abstract

Black women in community supervision programs (CSPs) are disproportionately affected by HIV and other sexually transmitted infections (STIs). A randomized controlled trial of a group intervention titled Empowering African-American Women on the Road to Health (E-WORTH) demonstrated effectiveness in reducing sexual risk behaviors and STI incidence among Black women in CSPs. This secondary analysis aimed to assess the moderating effects of housing status and food security on E-WORTH effectiveness in reducing sexual risk behaviors and cumulative incidence of STIs over a 12-month period which were found significant in the original trial among a sample of 351 Black women in CSPs in New York City who use drugs and/or engage in binge drinking who reported engaging in HIV risk behaviors or testing positive for HIV. We examined the moderating effects of housing stability, housing independence, and food insecurity on reducing cumulative STI incidence and number of unprotected sex acts using mixed-effects negative binomial regression and logistic regression models that controlled for age, high school education, employment status, and marital status. Findings indicate that the intervention effect was moderated by housing stability, but not housing independence or food security. Compared to the control group, E-WORTH participants who were housing insecure had 63% fewer acts of condomless sex. Our findings highlight the importance of interventions designed for women in CSPs that account for upstream determinants of health and include service linkages to basic needs provisions. Further research is needed to unpack the cumulative impacts of multiple experiences of poverty faced by this population.

Introduction

In the United States, 3.89 million adults are in community supervision programs (CSPs) [1], which encompass parole, probation, and alternative-to-incarceration (ATI) programs. Black women, who are disproportionately represented in community supervision due to structural racism in the criminal legal system (e.g., racialized drug laws, disparities in sentencing) [2–4], have higher rates of sexually transmitted infections (STIs), including HIV [5]. Despite this, there is a lack of effective HIV interventions for populations under community supervision. While Black women in CSPs report high levels of risky sexual behavior and low levels of condom use [6], this population also experiences higher levels of poverty and unmet basic needs, observed through high rates of homelessness, unemployment, and food insecurity [7]. Research shows that lack of housing stability is associated with more drug and sexual risk behaviors [8] and may contribute independently to HIV/STI risk [9].

Housing status and food security are also known as social determinants of health (SDoH), or conditions relating to how people live and the wider set of forces and systems shaping the livability of places [10]. Upstream SDoH are “those overarching factors that are largely beyond the control of the individual and which have significant spill-over effects on other proximal–or downstream–determinants of health” [11]. While there is some research examining the moderating role of various factors (e.g., interpersonal violence, social support, treatment frequency) on behavioral HIV interventions [12–14], there is little to no research examining how SDoH affect HIV interventions for people living under community supervision despite the high rates of housing instability, and poverty documented among this population.

A randomized control trial, funded by the National Institutes of Drug Abuse (NIDA), examined the effect of the “Empowering African-American Women on the Road to Health” intervention, also known as E-WORTH, among 351 Black women in CSPs in New York City (NYC). Black women are disproportionately represented in the criminal legal system of NYC: studies estimate that Black women comprise 37% of the population under community supervision in NYC, even though they make up only 14% of the overall population of NYC [4]. The overrepresentation of Black women in CSPs is an example of structural racism at play, with Black women unfairly targeted for arrest, prosecution, and incarceration. E-WORTH was implemented at a nonprofit multisite organization in NYC, which provides housing, mental health and social services to individuals in CSPs. The study, which ran from 2015 to 2019, found that the EWORTH intervention is successful at reducing unprotected sex and increasing condom use, and results in lower STI incidence among participants [6]. The study also found that participants assigned to the E-WORTH intervention, versus the control condition, had 54% lower odds of testing positive for any STI at the 12-month follow up, and 38% fewer acts of condomless vaginal or anal intercourse during the 12-month period [6]. E-WORTH participants had high group session attendance (69% of participants in the treatment arm) and high retention rates at all follow-up assessments, ranging between 79 and 86% [22]. The E-WORTH intervention has been named by the Centers for Disease Control and Prevention (CDC) as a Best Practice for HIV Prevention [15].

This analysis aims to examine how housing stability and food security moderate the effectiveness of the E-WORTH intervention on the primary outcomes of reducing number of condomless sex acts and cumulative HIV incidence. We hypothesize that the E-WORTH intervention would be more effective for individuals who report that they were housing secure or food secure at baseline, because their levels of participation in the intervention would be higher if their basic needs were met [16]. Prior research suggests that housing insecurity is associated with more drug and sexual risk behaviors and contributes independently to HIV risk [17]. We conceptualize these potential interactions using Bronfenbrenner’s social ecological model, which posits interaction effects between micro-, meso-, and macro-level factors [18]. In this study, we posit that meso-level factors such as housing stability and food security positively influence micro-level outcomes, such as behavior change and health outcomes. Our analysis investigates if the effect of the EWORTH intervention on behavioral (e.g., counts of unprotected sex) and biological outcomes (e.g., STI incidence) is attenuated among individuals who reported that they were housing insecure or food insecure at baseline.

Methods

The study received Institutional Review Board (IRB) approval from the Columbia University Institutional Review Board (protocol #: AAAN8409) [6]. The randomized control trial began on November 18, 2015, and ended on August 20, 2019. Study protocols have also been described in detail in previously published articles regarding this study [19].

Recruitment

Various recruitment strategies were employed as described elsewhere [19]. Black female Columbia University research assistants (RAs) directly approached prospective participants at CSPs located in all five NYC boroughs and handed out flyers. Project leadership personnel conducted presentations at each designated recruitment location. Eligibility criteria includes identifying as Black and female, living in NYC, under community supervision, reporting sexual risk behavior in the past 90 days, and reporting substance use risk behavior in the past 6 months.

Study Setting and Design

All post-recruitment intervention and control arm activities were conducted at the Manhattan and Queens locations of the non-governmental provider of community reentry services for individuals in CSPs in NYC with whom Columbia University (CU) contracted for the study. The choice to collaborate with this organization was influenced by their fifty-year history of delivering a comprehensive range of culturally congruent re-entry services to individuals who have been involved with the legal system. CU research assistants enrolled all study participants while intervention sessions were delivered by Black women who were non-governmental organization (NGO) staff and trained by Columbia University to deliver the intervention. Streamlined HIV testing, testing for other sexually transmitted infections (chlamydia, gonorrhea, and trichomoniasis) and a single enrollment session was delivered to control arm participants. Intervention participants also received these items plus five E-WORTH intervention sessions delivered by NGO personnel.

Participants in both arms were provided with an extensive array of case management and linkage services post enrollment and invited to become fully integrated into the NGO during- and post-intervention participation. Services were offered to study participants by the NGO personnel who delivered the intervention. Support offered included housing referrals, a hot meal, food bank referrals and access to an onsite food distribution program, clothing referrals and access to an onsite clothing closet, medical, mental health, and behavioral health referrals, vocational counseling and employment services. Participants who tested positive for HIV and/or other STIs under either condition were also provided with HIV and other posttest counseling and linked to appropriate treatment consistent with CDC guidelines [20].

Study Incentives

In addition to NGO services and referrals, cash incentives were provided to participants for completion of key engagement activities (e.g., screening, enrollment session, intervention sessions, longitudinal surveys completed at baseline, and at 3-, 6-, and 12-months post baseline). Individuals were granted $5 for screening and upon confirmation of eligibility, a payment of $55 was provided for baseline enrollment and survey. A sum of $55, $60, and $65 respectively were given for the 3-, 6-, and 12-month survey participation. Incentives were also provided for transportation to and from intervention locations and for uncovered baby-sitting costs if/as needed.

ACASI Computerized Surveys Utilized to Elicit Sensitive Narratives

All surveys were administered with the aid of audio computer-assisted self-interview (ACASI) software, a highly effective data collection approach that facilitates the collection of sensitive life histories (e.g., intimate partner violence, survival sex) from participants with extensive trauma experiences [21].

Intervention Design and Study Procedures

“Project E-WORTH” was adapted to address the need for culturally tailored HIV/STI prevention and risk reduction among Black women in community supervision [19]. The intervention is comprised of five group sessions structured to enhance HIV/STI knowledge, skills and self-efficacy to reduce associated risks (see Appendix A). Sessions focus on HIV education and testing, consistent condom use and taking pre-exposure prophylaxis (PrEP). The first session of the intervention is an orientation to E-WORTH with the video-assisted, streamlined HIV testing and knowledge (described in the control condition section) and introduction to goal setting and referrals for ancillary services. The following four weekly 90-minute group sessions utilize an in-person Black female facilitator who guides the opening and closing activities and who introduces the individualized computerized interactive activities with video versions of a Black female guide and additional modeling characters. Participants who missed a group session were able to complete an individualized makeup session to complete individualized computerized interactive activities. Core components include raising awareness about HIV/STI risks, condom use technical skills, PrEP and HIV Medication Adherence, sexual negotiation skills, risk reduction goal setting, increasing social support and linkage to services, intimate partner violence (IPV) screening, safety planning, and referral to IPV services.

The control condition was a streamlined HIV testing intervention which consisted of a single 30-minute individual HIV testing session included: (1) viewing a 5-minute HIV Rapid testing information video from OraSure; (2) reviewing the OraSure testing pamphlet on transmission risks and strategies for reducing risks; (3) taking the rapid OraQuick HIV test; and (4) receiving the test results and a service manual for ancillary services. Free condoms and safer sex kits were regularly distributed for both conditions at both CSP sites. Additional information about the core elements and the intervention adaptation process [19] is reported elsewhere. All study activities related to the E-WORTH intervention implementation were approved by the IRB at Columbia University.

Measurement of Exposure and Outcome Variables

Survey data completed by participants at baseline, 3-month follow-up, 6-month follow-up and 12-month follow-ups were used for this analysis, along with biological data testing for three STIs (chlamydia, trichomoniasis, and gonorrhea) conducted at baseline and 12-month follow-up. Sociodemographic information (e.g., age, completion of high school education, employment status, and marital status) included in the analysis was measured at baseline.

The moderating variables of interest were housing status and food security. Housing status was estimated with two variables: housing stability and housing independence. Housing stability was measured at baseline with a survey item with a yes/no response option asking, “Are you currently homeless?” Response options were yes and no. Housing independence was measured at baseline with a survey item asking participants, “In what type of place do you currently live?” Responses were recoded to a new dichotomous variable, with participants classified as housing independent if they answered that they owned or rented their own home, and housing dependent if they answered that they stayed at someone else’s home (e.g., family member(s), friend(s) or other), a group home, in a rented room, boarding/halfway house, a shelter, or other. Food security was estimated with a one-question survey item with a yes/no response option measured at baseline. Participants were asked, “In the past 3 months, have you ALWAYS had enough money to buy food?”

Behavioral outcome variables, estimated by survey questions in both the baseline and the 12-month follow-up surveys, include: (1) the total number of acts of condomless vaginal and/or anal intercourse across all partners in the prior 90 days; and (2) the total number of acts of condomless vaginal or anal sex with the main partner in the past 90 days. Biological outcome variables, including presence of any STI and presence of a new STI, were measured at baseline and the 12-month follow-up assessment. Participants provided vaginal swab specimens that were tested for presence of the three STIs. Participants were also tested for HIV using Oraquick swabs. If someone tested positive for an STI, they were treated by a medical practitioner and were asked to provide proof of treatment by a prescription bottle or note.

Statistical Methods of the Secondary Data Analysis

We fit multilevel mixed-effects models to estimate the intervention effects moderated by whether the participants were housing stable, housing independent, or food secure. To test for moderation, we ran unadjusted and adjusted unweighted models to test if housing stability, housing independence, or food security moderate the relationship between the intervention assignment and biological and behavioral outcomes. We added interaction terms of housing stability, housing independence, and food security at baseline with intervention assignment at baseline to the adjusted unweighted models. Hypothesis testing for the intervention effects was based on Incident Rate Ratios (IRRs) from multilevel mixed-effects negative binomial regression models for the number of condomless acts of vaginal or anal intercourse with main partner and the number of condomless acts of vaginal or anal intercourse with all partners. Also, hypothesis testing for the intervention effects was based on Odds Ratios (ORs) from multilevel mixed-effects logistic regression models for the presence of any sexually transmitted infection (STI) and the presence of a new confirmed STI. All models, consistent with the original E-WORTH intervention effectiveness paper [6], included covariate adjustments for the baseline measures of the outcome variables (e.g., confirmed STI at baseline), age, race/ethnicity, marital status, high school graduation status, and employment status. Estimates were considered statistically significant if p < 0.05. All statistical analyses were performed using SPSS 25.

Results

The demographic characteristics of the sample of participants in the E-WORTH trial are described in a separate paper about the intervention effectiveness [6]. Overall, 80.3% of participants in the study were housing stable, 41.3% of participants in the study were housing independent, but only 36.6% of participants were food secure (see Table 1).

Table 1.

Descriptive statistics of housing stability and food security at baseline among E-WORTH participants in NYC

	Housing status				Food security
	Housing stability		Housing independence
	Yes (%)	No (%)	Yes (%)	No (%)	Yes (%)	No (%)
	N = 282 (80.3)	N = 69 (19.7)	N = 145 (41.3)	N = 206 (58.7)	N = 128 (36.6)	N = 222 (63.4)
Latinx ethnicity
Yes	54 (22.4)	18 (31.6)	32 (22.1)	47 (22.8)	22 (21.8)	50 (25.5)
No	187 (77.6)	39 (68.4)	113 (77.9)	159 (77.2)	79 (78.2)	146 (74.5)
High school graduate or GED
Yes	156 (55.3)	40 (58.0)	73 (55.7)	94 (56.3)	72 (56.3)	123 (55.4)
No	126 (44.7)	29 (42.0)	58 (44.3)	73 (43.7)	56 (43.8)	99 (44.6)
Married (including common law)
Yes	115 (40.8)	29 (42.0)	50 (38.2)	67 (40.1)	56 (43.8)	87 (39.2)
No	167 (59.2)	40 (58.0)	81 (61.8)	100 (59.9)	72 (56.3)	135 (60.8)
Sexual orientation
Heterosexual/straight	158 (65.6)	38 (66.7)	108 (74.5) *	122 (59.2) *	75 (74.3)	120 (61.2)
Bisexual	77 (32.0)	17 (29.8)	32 (22.1) *	77 (37.4) *	25 (24.8)	69 (35.2)
Other	6 (2.5)	2 (3.5)	5 (3.4) *	7 (3.4) *	1 (1.0)	7 (3.6)

^*Significant at p < 0.05 on Chi-square test

Moderation Analyses of Intervention Effect

Results from the multilevel mixed effect models show that the intervention effect was moderated by housing stability, but not housing independence or food security (see Tables 2 and 3). Compared to the participants in the control arm of the intervention over the 12-month study period, E-WORTH participants who were housing insecure had 68% fewer acts of condomless vaginal or anal intercourse with their main partner (IRR = 0.315, 95% CI = 0.177, 0.561, p < 0.001) and 63% fewer acts of condomless vaginal or anal intercourse with all partners (IRR = 0.368, 95% CI = 0.206, 0.658, p < 0.001) (see Table 3).

Table 2.

Descriptive statistics of E-WORTH outcome variables by moderator variables

	Housing status								Food security
	Housing stability				Housing independence
	Yes (N = 282 (80.3%))		No (N = 69 (19.7%))		Yes (N = 145 (41.3%))		No (N = 206 (58.7%))		Yes (N = 128 (36.6%))		No (N = 222 (63.4%))
	Control	E-WORTH	Control	E-WORTH	Control	E-WORTH	Control	E-WORTH	Control	E-WORTH	Control	E-WORTH
Any STI (chlamydia, gonorrhea, or Trichomonas vaginalis) (Number, %)
Yes	36 (60.0%)	24 (40.0%)	18 (40.9%)	26 (59.1%)	23 (63.9%)	13 (36.1%)	19 (56.0%)	15 (44.1%)	19 (76.0%)	6 (24.0%)	22 (50.0%)	22 (50.0%)
No	87 (49.4%)	89 (50.6%)	6 (60.0%)	4 (40.0%)	53 (57.6%)	39 (42.4%)	52 (40.6%)	76 (59.4%)	33 (44.6%)	41 (55.4%)	72 (49.3%)	74 (50.7%)
Any confirmed new STI, (Number, %)
Yes	19 (70.4%)	8 (29.6%)	3 (75.0%)	1 (25.0%)	10 (62.5%)	6 (37.5%)	12 (80.0%)	3 (20.0%)	9 (90.0%)	1 (10.0%)	13 (61.9%)	8 (38.1%)
No	103 (50.0%)	103 (50.0%)	21 (42.0%)	29 (58.0%)	65 (59.1%)	45 (40.9%)	59 (40.4%)	87 (59.6%)	43 (48.5%)	45 (51.1%)	80 (47.9%)	87 (52.1%)
No. of unprotected acts of vaginal or anal intercourse with main partner (Mean, SD)
	21.5 (37.4)	16.4 (34.0)	23.7 (32.5)	11.5 (25.5)	19.5 (31.1)	10.3 (20.9)	23.3 (40.8)	18.4 (37.3)	27.7 (43.3)	8.72 (17.7)	18.9 (32.2)	18.4 (37.0)
No. of unprotected acts of vaginal or anal intercourse with all partners (Mean, SD)
	19.0 (34.3))	14.2 (29.4)	23.6 (32.5)	11.4 (26.0)	18.5 (29.7)	9.8 (21.0)	22.2 (38.8)	15.6 (32.1)	25.0 (40.3)	7.9 (17.8)	17.2 (29.8)	16.3 (32.4)

Table 3.

Interaction analysis of housing stability and food security / moderation analysis of the effectiveness of the E-WORTH intervention

		OR or IRR [95% CI]		OR of IRR [95% CI]		OR or IRR [95% CI]
Any STI (chlamydia, gonorrhea, or Trichomonas vaginalis), OR (95% CI)	Housing insecure	0.339 [0.076, 1.514]	Housing dependent	0.336 [0.142, 0.796] *	Food insecure	0.769 [0.366, 1.617]
	Housing secure	0.476 [0.021, 10.757]	Housing independent	0.715 [0.088, 5.784]	Food secure	0.150 [0.019, 1.206]
	Diff.	1.404 [0.277, 7.105]	Diff.	2.128 [0.623, 7.266]	Diff.	0.195 [0.051, 0.746] *
Any confirmed new STI, OR (95% CI)	Housing insecure	0.347 [0.031, 3.839]	Housing dependent	0.152 [0.032, 0.728] *	Food insecure	0.643 [0.233, 1.773]
	Housing secure	0.469 [0.003, 63.090]	Housing independent	0.994 [0.030, 33.479]	Food secure	0.116 [−5.531, 3.436]
	Diff.	1.253 [0.095, 16.434]	Diff.	6.542 [0.931, 45.988]	Diff.	0.181 [0.017, 1.938]
No. of unprotected acts of vaginal or anal intercourse with main partner, IRR (95% CI)	Housing insecure	0.315 [0.177, 0.561] **	Housing dependent	0.712 [0.515, 0.986] *	Food insecure	0.808 [0.599, 1.091]
	Housing secure	0.845 [0.249, 2.861]	Housing independent	0.605 [0.262, 1.397]	Food secure	0.474 [0.207, 1.084]
	Diff.	2.683 [1.411, 5.103] *	Diff.	0.848 [0.509, 1.416]	Diff.	0.587 [0.347, 0.993] *
No. of unprotected acts of vaginal or anal intercourse with all partners, IRR (95% CI)	Housing insecure	0.368 [0.206, 0.658] **	Housing dependent	0.845 [0.611, 1.167]	Food insecure	0.932 [0.693, 1.252]
	Housing secure	0.935 [0.277, 3.155]	Housing independent	0.669 [0.292, 1.530]	Food secure	0.525 [0.233, 1.185]
	Diff.	2.537 [1.342, 4.796] *	Diff.	0.792 [0.478, 1.311]	Diff.	0.564 [0.336, 0.946] *

Covariate adjustment: age, high school education, employment status (employed full-time or part-time vs. not employed), marital status (married or common-law vs. single), any confirmed STI at baseline, and baseline measure of the outcomes

^*p < 0.05

^**p < 0.001

While there was no evidence of the intervention effect being moderated by housing independence, the effect of the intervention was stronger among E-WORTH participants who were housing dependent upon adjustment for the moderator variable. Among the participants who were housing dependent at baseline, and compared to the participants in the control arm of the intervention over the 12-month study period, E-WORTH participants had a 66% lower odds of having an STI (OR = 0.336, 95% CI = 0.142, 0.796, p = 0.013) and an 85% lower odds of having a new STI (OR = 0.152, 95% CI = 0.032, 0.728, p = 0.018) (see Table 3). In addition, among the participants who were housing dependent at baseline, and compared to the participants in the control arm of the intervention over the 12-month study period, E-WORTH participants had 29% fewer acts of unprotected vaginal or anal intercourse with their main partner (IRR = 0.712, 95% CI = 0.515, 0.986, p = 0.041) (see Table 3).

There was evidence of moderation of the intervention effect by food security in unadjusted models; however, the effect of the intervention upon all outcomes became statistically insignificant upon including covariates in the model.

Discussion

This study was conducted to assess if housing status and food security moderated the effectiveness of the E-WORTH intervention in reducing sexual risk behavior and improving health outcomes. The study findings suggest that the E-WORTH intervention was more effective for individuals who were housing insecure 90 days prior to baseline. The results diverged from our original hypothesis, illuminating the complex moderation effects of upstream determinants of health on behavioral interventions and burgeoning the case for continued use of moderation analysis in SDoH research. The intervention design did consider that housing instability can exacerbate sexual risk behaviors and incorporated strategies to address this specific factor through enhanced service linkage to housing services and strengthening social support to help with issues like housing. These study findings show that the moderating effects of upstream SDoH on HIV interventions for economically marginalized women in CSPs are complex and highlight the importance of considering multi-level risk and protective factors that may position some women to take greater advantage of interventions.

The E-WORTH intervention may be more effective for housing insecure adults versus those who are housing secure due to several factors. First, housing instability exacerbates vulnerability and instability often brings about a range of challenges and stressors, such as precarious living situations, frequent moves, and inadequate access to resources [23]. These factors can contribute to increased vulnerability to engaging in sexual risk behaviors, including condomless sex [17, 24]. The E-WORTH intervention, with its focus on addressing the unique needs of vulnerable individuals, may have been better equipped to target and mitigate these specific risk factors related to unmet basic needs, resulting in a more significant effect. Additionally, tailored elements of the E-WORTH intervention included components specifically designed to address the needs and challenges associated with housing instability. For instance, through our collaboration with the community supervision program, some participants may have had housing support services, such as referrals to stable housing resources or assistance in navigating housing systems. By directly addressing the housing instability experienced by participants, the intervention likely provided a more comprehensive and targeted approach, leading to greater effectiveness among this subgroup. Housing status is often intertwined with various other socioeconomic challenges, such as limited access to healthcare, employment instability, and social marginalization [25]. These contextual factors can compound the risk of engaging in sexual risk behaviors. The E-WORTH intervention may have considered these complex contextual factors and incorporated strategies to mitigate their impact. By addressing the broader context in which housing insecure individuals operate, the intervention may have been better suited to produce positive outcomes.

Housing insecure individuals may be more motivated and readier for change due to the immediate impact of housing insecurity on their daily lives. The experience of recent homelessness, as reported by individuals in the E-WORTH intervention arm at baseline, may have served as a powerful motivator for behavior change. The E-WORTH intervention capitalized on this readiness for change by providing support and resources to facilitate healthier decision-making regarding sexual risk behaviors. Prior studies also document the role that individual- or interpersonal-level interventions aimed at improving health play in improving housing and/or food security [16, 26]. It is important to note that the specific mechanisms underlying the differential effectiveness of the E-WORTH intervention for housing insecure adults cannot be definitively determined without further analysis or additional information. However, the factors mentioned above provide potential explanations for the observed pattern in the results. Alternatively, women who were housing insecure over the intervention and follow-up period may have fewer opportunities to be sexually active and have intimate partnerships due to the lack of privacy. Thus, further research could explore these factors in more detail to gain a deeper understanding of why the intervention had a greater effect on housing insecure adults and how to optimize interventions for different subgroups.

Our research supports prior studies documenting the role that individual- or interpersonal-level interventions have greater effectiveness among participants with a higher level of unmet basic needs, and the role that individual- or inter-personal-level programs aimed at improving health play in improving housing and/or food security [16, 26]. Study findings–and previous literature examining possible links between housing status and sexual risk behavior–lead us to conclude that the effectiveness of E-WORTH in reducing sexual risk behavior may also impact future housing stability. Further longitudinal studies are needed to determine with greater precision the role of episodes of homelessness over time on health. Results from this study highlight the importance of further testing links between housing status and sexual risk behaviors, in order to improve the effectiveness of interventions such as E-WORTH. The E-WORTH intervention specifically targets Black women in CSPs who use drugs or engage in binge drinking and with a focus on addressing their sexual risk behaviors and the unique challenges faced by individuals under community supervision who may historically experience housing instability. These design elements shape the understanding of how the intervention activities lead to the desired impact and inform the strategies implemented within the intervention. However, the E-WORTH intervention did not explicitly include elements targeting levels higher than the interpersonal level. By honing in on the needs of this specific population, the tailored intervention design acknowledges the unique challenges and needs of this group and adapts the program to effectively address those challenges emphasizing the importance of stable housing as a foundational element for reducing sexual risk behaviors and working across the network of health and social service providers in NYC to address the unmet basic needs of this population.

The importance of developing effective HIV interventions for populations under community supervision is paramount, given the legacy of racist drug laws and policies that have resulted in an alarming overrepresentation of Black women and men who remain at high risk of HIV/STIs due to persistent racial health inequities in the HIV epidemic. We analyzed our findings in the context of eco-social theory, which considers “how we literally biologically embody exposures arising from our societal and ecological context, thereby producing population rates and distributions of health” [27]. Scholars of public health and epidemiology use theories such as eco-social theory and intersectionality to contend with the current limitations of the conversations around social determinants of health, and to include factors such as period and duration of exposure, historical context, and other phenomena [27, 28]. In this context, Black women in community biologically embody the inconsistency of not having enough food or housing, and are empowered by tailored interventions such as E-WORTH. Using this theory, we consider that the medium- and long-term health outcomes of E-WORTH participants may have been impacted by the chronicity and periods of exposure to social hardship not only over the course of the intervention study, but over the lifespan [29].

The findings of this study have several implications for future studies, policy development, and direct service provision in addressing sexual risk behaviors and related issues among populations impacted by housing instability and homelessness. Future research can evaluate the effectiveness of basic needs support and service referrals in the E-WORTH intervention, especially as this intervention is further disseminated and implemented (see CDC Compendium of Evidence-Based Interventions and Best Practices for HIV Prevention [15]). Additional studies that replicate these findings in other contexts will establish the robustness of the observed moderation effect of housing instability on behavioral HIV interventions for individuals living in poverty. Replicating the study with larger and more diverse samples will enhance generalizability. Given that food security did not show a significant moderation effect, future studies should explore other potential moderators that may influence the intervention’s outcomes, such as mental health status, substance abuse, or social support. Conducting long-term follow-up studies can provide insights into the sustainability of the intervention’s effects on sexual risk behaviors and the subsequent rates of STIs.

The study highlights the importance of collaboration between city, state, and federal government agencies, housing services, healthcare providers, and the criminal legal system, such as probation, and other community organizations. Collaborative efforts can enhance the delivery of comprehensive and integrated services to address both housing instability and sexual risk behaviors. There is much potential for systems science approaches to facilitate cross-sector and/or interorganizational collaboration needed to implement innovative housing solutions [30, 31]. Group model building has demonstrated much effectiveness with enhancing a community’s ability to identify intervention points, opportunities for assessment and evaluation, and policy implications [32]. Practitioners can also adopt holistic approaches that address the interconnected factors contributing to sexual risk behaviors, including housing instability, mental health, substance abuse, and social support. A comprehensive approach can yield better outcomes in reducing sexual risk behaviors and improving overall well-being.

Limitations & Strengths

There are several limitations to this study. While longitudinal data were collected, the analysis was conducted without a longitudinal structure in place (outcome variables were at the 12-month period and all other variables were collected at baseline). This study design did not consider changes in food or housing status over time resulting from the E-WORTH intervention. In addition, the timeframe specified by different survey questions (e.g., 90 days vs. 3 months) might result in a misclassification bias. The SDoH constructs measured by our variables, such as housing status and food security are also dynamic, and not easily captured by single-item survey questions. This study’s generalizability is limited to Black women in one large multi-site CSP in a dense urban context, resulting in unique elements driving the housing and criminal legal systems and the varying, unique lived experiences of participants navigating unmet basic needs in NYC.

Despite these limitations, this study has several strengths, such as high rates of participant attendance and retention and use of biological STI outcome in addition to self -reported outcome of condomless sex. The study findings are transferable to a range of community supervision settings and contribute to the implementation knowledge of how to improve E-WORTH intervention fidelity and effectiveness. Very little is known about implementation science for evidence-based public health recommendations in other sectors, and contributing to the base of knowledge unpacking the importance and causal complexity of upstream SDoH is novel and merits further research.

Conclusions

Black women in CSPs typically have multiple unmet basic needs and are disproportionately affected by HIV/STIs, largely attributed to structural drivers. The aim of this study was to identify the moderating effect of housing status and food security on the effectiveness of the E-WORTH intervention in reducing sexual risk behavior and improving health outcomes. Study findings reveal that compared to the control group, individuals in the E-WORTH intervention arm who were housing insecure reported fewer acts of condomless sex. Future research efforts include evaluation of service referral components, investigating the effectiveness of replicating the E-WORTH intervention in other contexts, and systems science approaches to addressing upstream SDoH for individuals in CSPs to promote health and mitigate health disparities. The findings from this research are critical to inform policy development to prioritize stable housing and integrated services and guide real-world practitioners in tailoring interventions to address the complex needs of individuals experiencing housing instability and homelessness while addressing sexual risk behaviors.