Introduction

In the United States, an estimated 20–40 % of women at high risk of HIV infection reported having sex in exchange for money or drugs within the past year [1–3]. Female sex workers (FSW) in the US have been identified as a high-risk ‘bridge’ population who can acquire and transmit HIV and other STIs via engagement in multiple risk behaviors that often occur simultaneously: inconsistent condom use, sex with partners of unknown HIV status, concurrent sexual partnerships with risky sexual partners, and engagement in poly-drug use including injection drugs and crack [4, 5]. However, the criminalized nature of sex work in the US makes this population particularly difficult to identify and reach. As a result, FSW may be less likely to receive HIV/STI services and to access programs that enable long-term behavior change [6].

The stigmatization and marginalization of FSW in the US also makes it difficult to reliably quantify HIV-risk in this population. Conservative estimates suggest that women who exchange sex for money or drugs have a higher risk of HIV/STI infection than not only the general population, but also other similarly high-risk women who do not engage in sex work [7–10]. According to 2010 national surveillance estimates for high-risk heterosexuals, HIV prevalence among men and women who reported ≥1 one exchange partner in the past year was 3.7 %, compared with 2.1 % among individuals with no exchange partners [7, 11]. Findings from smaller, single-location studies that used similar venue-based sampling methods suggest that FSW are likewise at increased risk of HIV infection when compared to their non-sex-working counterparts [1, 8].

However, it is likely that population-based studies underestimate the true risk among FSW. Incarcerated FSW, who are typically excluded from larger studies, appear to be at higher risk for HIV/STI. Among newly incarcerated FSW in New York City, more than 10 % tested positive for HIV and 14 % tested positive for STIs [12]. Moreover, HIV/STI risk appears to be variable among FSW, and is greater among FSW who have high numbers of exchange partners [1]. In additional samples of high-risk FSW, studies have documented HIV prevalence above 20 % [13, 14].

Further, there are a number of behavioral and structural risk factors that place FSW at increased risk of acquiring HIV and other STIs [9, 15–23]. FSW are more likely to engage in unprotected sex, have multiple, high-risk sex partners, and inject drugs than other at-risk women [1]. They are likely to abuse other substances, including alcohol, marijuana, crack/cocaine, and prescription drugs, which increase their risk of contracting HIV via lowered inhibitions, reduced ability to negotiate condom use, and a higher likelihood of continued engagement in transactional sex as a method of obtaining drugs or money [4, 24, 25]. Violence, including intimate partner violence, is a common experience of many FSW and has been associated with HIV/STI risky sexual behaviors, including inconsistent condom use, multiple partners, and an earlier sexual debut [21, 26, 27]. Structural gender inequality may prevent FSW from feeling in control of condom use during transactional sex [14]. A confluence of other structural vulnerabilities—e.g., homelessness or unstable housing, incarceration, poverty, unemployment—acting synergistically increases FSW’s likelihood of engaging in HIV-related sex- and drug-risk behaviors [28–30].

Purpose of Review

Despite the risk of HIV/STI acquisition & transmission among FSW and their vulnerability to multiple risk factors, studies of risk-reduction efforts among FSW have been missing from the US-based HIV prevention literature. There is a critical need to develop efficacious interventions to reduce sex and drug injection risk behaviors among this marginalized population. To our knowledge, this is the first systematic review to identify and discuss HIV/STI prevention interventions conducted in the US, though many have been conducted internationally [31–37].

Methods

Since 1996, the Centers for Disease Control and Prevention (CDC) has supported the Prevention Research Synthesis (PRS) Project to systematically review the HIV/STI intervention research literature to understand the state of the science, identify evidence-based HIV prevention interventions, and make evidence-based recommendations [38]. Additional information about the PRS risk-reduction efficacy review methods is available via the PRS website (http://www.cdc.gov/hiv/dhap/prb/prs/efficacy/rr/criteria/review_methods.html).

Although the primary purpose of the PRS project is to evaluate intervention efficacy, the purpose of this review was to characterize all interventions for FSW described in the published literature, regardless of whether they were evaluated for intervention efficacy. The PRS database was used to identify eligible studies.

Database and Search Strategy

As part of the PRS project, a cumulative database of the HIV/STI prevention literature was developed using a systematic search procedure, including both automated and manual search strategies [38]. Briefly, the automated component uses combinations of keywords and MESH terms and searches multiple electronic databases—AIDSLINE (1988 to discontinuation in December 2000), EMBASE (OVID), MEDLINE (OVID), PsycINFO (OVID), and Sociological Abstracts (PROQUEST)—to identify relevant literature published between 1988 and 2012. The full search strategy is available from the authors. The automated search component is repeated annually to update the PRS database and the last update prior to this review was completed in January 2013. A manual search component consists of quarterly hand searches of 38 pertinent HIV/AIDS journals and review of reference lists of relevant articles and conference abstracts. The last quarterly hand search prior to this review was completed in February 2013.

The PRS database was searched in May 2013 to identify all eligible HIV/STI prevention intervention studies for this review, using variations of keywords related to female sex work (e.g., commercial sex, sex trade, prostitution, paying partners). The applicable Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed in this review (Moher et al. 2009). A protocol for this review is not available.

Study Selection

A study was eligible for this review if it described or evaluated an HIV/STI behavioral intervention, was conducted in the United States, was an English-language study published in a peer-reviewed journal, was published between 1988 and May 2013, and met one of the following three criteria: (1) focused on or targeted FSW, (2) stratified data by FSW if other groups were also targeted, or (3) included a study sample that consisted of more than 50 % (a majority) FSW. As there is not a substantial literature documenting or quantifying the various types of difference between FSW, this review utilizes the definition of a sex worker as a woman or a girl who trades sex for money, drugs or goods [39, 40]. Studies were eligible for this review whether or not they contained evaluation data, and all study outcomes were abstracted, particularly sex and drug-relevant outcomes.

Data Abstraction

Data were coded for study information and study methodology (e.g., location, allocation method), target population characteristics (i.e., whether the study specifically targeted FSW, stratified results by FSW, or contained greater than 50 % FSW), participant demographics (e.g., age, race/ethnicity, education, housing, and income), risk behavior at baseline (e.g., substance abuse and sexual behavior), and intervention content and characteristics (e.g., HIV prevention, substance abuse prevention, mental health, and economic resources) by two independent reviewers. For one-group designs, within-group outcomes were considered, but for studies with a comparison group, only significant between-group outcomes were considered.

Due to the limited number and heterogeneous nature of the outcomes reported, we were unable to perform a meta-analysis on abstracted data. We summarized the results by stratifying studies in three categories: those that exclusively focused on FSW (“targeted”), those that stratified baseline or outcome data by FSW (“stratified”), and those in which FSW comprised greater than 50 % of the sample (“majority”).

Results

Two thousand and six hundred and eighty four citations were identified through a search of the PRS database and 53 citations were identified through ad hoc searches for inclusion in this review (Fig. 1). Among those citations, 2,737 citations were screened at title and abstract, and 149 were assessed at full report for eligibility. After linking citations that describe the same study, 18 unique studies that either described or evaluated 19 unique interventions met the inclusion criteria and were included in this review. One study, reported by Sterk et al. [41] evaluated two different interventions: the motivation intervention and the negotiation intervention. Another intervention, the female condom study [42], was implemented and evaluated in two cities, St. Louis, MO and San Antonio, TX. Though this intervention was counted as one unique intervention and coded once for content, the baseline and outcome data from each city are reported separately in Tables 1 and 2. Reasons for excluding citations are provided in Fig. 1.

Fig. 1
figure 1

Study flow

Full size image
Table 1 Baseline study population characteristics of female sex workers
Full size table
Table 2 Intervention characteristics
Full size table

The final set of 19 interventions [number of studies (k) = 18] included five that exclusively targeted FSW, two that stratified data by FSW, and 12 that included greater than 50 % FSW.

Demographics

Baseline study population characteristics of the 18 intervention studies are presented in Table 1, organized by type of intervention. Not all studies reported all demographic variables coded. In total, more than 4,000 participants were included across the 18 studies. Most studies included women who were, on average, in their mid to late thirties (k = 9). The vast majority of included study samples were predominately black/African American (k = 10) or Hispanic/Latino (k = 3); none of the studies were majority white. Women included in these studies tended to have low levels of education (k = 6 had a majority of participants with less than a high school degree), were largely unemployed (k = 6 reported over 75 % unemployed), were often reliant on public assistance (k = 6), friends and family (k = 3), or illegal sources (k = 4) for income, and were frequently homeless or unstably housed (k = 5 reported over 20 % homeless).

HIV/STI Risk Behaviors

Women in the included studies reported high levels of current and past drug use, though the type of drug use varied from study to study. In general, IDU tended to range between 35 and 65 %, while reported crack use was frequently very high (k = 5 studies reported over 75 % crack use). In more than 60 % of the included studies, over 75 % of participants reported trading sex for money, drugs, or another reason. Among the few studies that specifically reported reasons for trading sex, women most frequently reported trading sex for money, followed by trading sex for drugs; only one study reported that women had traded sex for any other reason (Torres et al. [43] counted food as one potential reason). Among studies that tested for HIV, the prevalence of HIV ranged from 0 to 46 % (k = 13).

Childhood Abuse and Adult Interpersonal Violence

Few studies provided information on rates of childhood abuse and adult interpersonal violence among FSW. Wechsberg et al. [44] reported that 30 % of their sample had experienced childhood sexual abuse and 30 % had experienced childhood physical abuse. Similarly, Grella et al. [45] reported that 38 % of their sample had experienced childhood sexual abuse, 40 % childhood physical abuse, and 57 % childhood emotional abuse. Vigalante et al. [46] reported that 31 % of their sample had experienced abuse by an intimate partner, and 50 % had a history of rape. Gollub et al. [47] and Torres et al. [43] reported that 25 and 24 %, respectively, of their sample reported that their first sexual experienced was forced. Surratt and Inciardi [48] reported that 23 % of their sample had been physically victimized, and 18 % had been sexually victimized in the past 90 days.

Very few studies reported rates of depression, anxiety, anger, stress, post-traumatic stress disorder (PTSD), or any other mental health indicator at baseline. Grella et al. [45] stated that 66 % of their sample reported depression, 61 % reported suicidal ideation, 43 % were taking psychiatric medication, 39 % had previously attempted suicide, and 21 % had been hospitalized for a psychiatric problem. Wechsberg et al. [44] reported moderate rates of depression [mean 12.9; standard deviation (SD) 4.9] and anxiety (mean 11.2; SD 6.2) on the Drug Abuse Treatment Assessment and Research scale (range 0–28), and trauma (mean 18.2; SD 10.4) on the Global Appraisal of Individual Needs scale (range 0–48).

Intervention Characteristics & Study Methods

A summary of intervention characteristics and study methods are presented in Table 2, organized by type of intervention. Study locations included 14 states and Puerto Rico; Miami, FL and cities in Southern California were the most common study locations. Study participants were primarily recruited from street strolls or other street-based locations where FSW were known to solicit clients (e.g. back alleys, and empty lots). Some studies also recruited participants from non-profit and health care organizations, correctional facilities, HIV/AIDS clinics, and via fliers and word of mouth. Fifteen studies targeted women who used drugs at recruitment.

Eight interventions (2 target and 5 majority) reported conducting formative research prior to implementing the interventions. Two interventions asked FSW for input in designing the intervention, typically in the form of focus group. The median number of intervention sessions was 4, with a range of 1–6. The median total time per intervention was 145 min, with a range of 30–900 min. Nine interventions were individual-level, seven were group-level, and two were community-level.

Seven of the included interventions utilized a randomized control trial (RCT) design, four utilized a non-RCT two-group design, seven were tested with a one-group post-design only, and one intervention did not report study design or evaluation information.

Intervention Content

Table 3 describes intervention content for “target” and “majority” studies, including guiding behavioral theories, content addressing issues facing FSW, HIV prevention information, HIV services, substance abuse, and skills building techniques. The two stratified studies [49, 50] did not report sufficient information to make meaningful comparisons with other interventions.

Table 3 Comparison of intervention content between interventions that targeted FSW or enrolled > 50 % female sex workers (j = 17)
Full size table

Two target interventions tailored content for issues facing FSW and few reported using behavioral theory to guide intervention development. At least half of the target HIV/STI prevention interventions for FSW included general HIV/STI and substance abuse prevention information. Nearly all target interventions referred participants to social service programs while few offered general health care, mental health, psychosocial, or victimization-relevant services within the intervention.

Similar to target interventions, less than half of the majority interventions included content that specifically addressed issues faced by FSW and few reported using behavioral theory to guide intervention development. Nearly all of the majority interventions included general HIV/STI and substance abuse prevention information, and several offered HIV counseling and testing services. Few interventions offered HIV/STI-related medical care or substance abuse treatment services. Majority interventions were more likely to focus on skill building within the intervention, particularly proper syringe/needle cleaning techniques and male/female condom demonstrations than target interventions. More majority than target interventions included psychosocial content, including gender norms, empowerment, motivation to reduce risk behavior, risk-reduction attitudes, and self-esteem, although less than half of the interventions included these components. Fewer majority interventions referred participants to social service programs than target interventions, while a greater number of majority than target interventions created individualized risk-reduction plans, typically centered on substance use reduction.

Outcome Findings

One target intervention reported a significant reduction in STI incidence among FSW [51]. Six out of the ten interventions that reported a sex-risk reduction outcome, and five out of the ten interventions that reported a drug-related risk reduction outcome significantly reduced HIV risk-taking behaviors. Five interventions reported a sex-work-related outcome; two of these were successful in decreasing sex work. Of the nine interventions that reported other outcomes (e.g. mental health, violence/abuse, homelessness, employment, etc.), four reported significant intervention effects.

Three target interventions reported a significant sex or drug risk reduction [48, 52, 53]. Sherman et al. [52] microenterprise intervention reported reductions in the greatest number of risk behavior outcomes spanning multiple categories such as increased condom use, decreased number of sex trade partners (sex-risk behavior); decreased injection drug use, daily general drug use, daily crack use, money spent on drugs per day, income from selling drugs (drug-risk behavior); and decreased income from sex work (sex work-related risk behavior). Unfortunately, this microenterprise intervention did not include a comparison group to evaluate efficacy [52].

Many majority interventions reported significant reductions in several substance use behaviors, such as general or specific drug use (e.g. heroin, crack), as well as other outcomes, including homelessness, employment, and recidivism. Bowser et al. [54] harm-reduction-based drug treatment intervention reported the greatest number of significant effects, including a reduction in the number of days a participant used drugs and spent time in jail, and an increase in the proportion of participants who found housing and employment. Wechsberg et al. [44] Women’s Co-Op, a gender and culturally sensitive intervention created to reduce sex- and drug-risk behaviors among high-risk crack-using African American women, also reported a number of significant intervention effects, including reductions in unprotected sex, the number of days a participant smoked crack, sex trading, homelessness, and unemployment.

Discussion

Our systematic review of US-based HIV/STI prevention intervention studies of FSW highlights important factors that have been addressed by current efforts as well as gaps that should be addressed in future research. The overall findings of our review demonstrate that few rigorously implemented or evaluated HIV/STI behavioral prevention interventions exist that address the needs of FSW in the US.

Overall, the quality of these intervention studies was low, as less than half of the interventions were evaluated using RCTs and others lacked an adequate control or comparison group. It is unclear why so few US-based studies have been rigorously implemented and evaluated with FSW, particularly since other industrialized countries (e.g. Australia, Canada) have a more developed FSW literature [55]. It may be that the stigmatized and often illegal behaviors in which FSW engage may have dissuaded a dedicated investment in the population. Nevertheless, it is possible that increased risk-reduction efforts may result in lowered HIV/STI risk in this important and marginalized population.

Both target and majority HIV/STI prevention interventions included general HIV/STI and substance abuse prevention information; however, few interventions tailored this content to address issues facing FSW. Although majority interventions did not tailor content to FSW, they included more skill building activities and psychosocial content, and were grounded in behavioral theory. The greater robustness of majority interventions may be reflective of the fact that they were typically developed to meet the needs of high-risk women who use drugs, a priority population during the peak of the HIV epidemic. Indeed, the two interventions that reported the greatest number of significant outcomes were both majority interventions that recruited crack-using women, a population especially important to HIV prevention among high-risk women in the 1990s [25, 56].

As a result of prioritizing HIV/STI prevention among high-risk women who use drugs, much of the extant literature regarding FSW has targeted women who are poor, use drugs, or who have sexual contact with multiple partners. Most interventions included in this review specifically recruited FSW who use drugs to participate in the intervention. Due to this bias, there is information on the risk behavior of FSW who do not use drugs and the types of intervention efforts that would lead to greater risk reduction.

Few included interventions addressed psychosocial risk factors such as victimization and poor mental health, and no study reported rates of violence between a FSW and paying partners, police, or other individuals on the street. This was the case despite prior research demonstrating that FSW experience greater psychological distress, report more physical and sexual abuse, and have more frequent encounters with police than non-FSW [9, 21]. Interventionists interested in working with high-risk FSW [57] could adapt principles of cognitive behavioral therapy to deal with prior trauma and current distress, as this technique has been effective in both curbing risk behaviors and reducing psychological distress among other high-risk populations [58, 59]. Stronger linkages can also be made between community organizations that provide mental health services and participants graduating from an intervention program.

Similarly, despite the association between structural issues (e.g. homelessness, access to healthcare, economic resources) and negative outcomes among FSW, few target or majority interventions focused on these content areas. While some interventions referred FSW to social service programs that provided these kinds of resources, few interventions actually offered assistance with these issues or direct linkages to community organizations.

In general, psychosocial and structural factors affecting FSW have been more successfully incorporated in FSW interventions internationally, particularly in sub-Saharan Africa where the greatest HIV burden among FSW has been observed [34, 55, 60]. HIV/STI interventions in this region have included female and male condom promotion, voluntary HIV/STI counseling/testing, peer education, stigma reduction, policy changes, and community empowerment/social support approaches [60]. The current World Health Organization guidelines for HIV prevention among FSW advocate decriminalizing sex work, removing discriminatory laws and regulations, prioritizing the prevention of violence against sex workers, and increasing access to health care and biomedical prevention and treatment options [61].

Two interventions included in this review were replicated among samples of FSW: Wechsberg and colleagues’ Women’s CoOp in Pretoria, South Africa and Sherman and colleagues’ Microenterprise intervention in Chennai, India [62, 63]. These interventions resulted in significant increases in condom use, decreases in the number of paying sexual partners, and increases in the amount of income derived from legal sources (via microenterprise activity). These successful replications of US-based interventions in international contexts suggests that increased communication between international and domestic HIV/STI prevention efforts among FSW may be mutually beneficial to both sides.

Most of the interventions included in this review defined sex work as exchanging sex for money, drugs, or both money and drugs. However, recent reports on the nature of transactional sex in the US suggest that it may be necessary to expand this definition to include other types of sexual relationships. Dunkle et al. [28] reported that of 1,453 randomly selected unmarried women, 13.1 % of African American women and 2.9 % of white women reported having sex because they needed help paying for their housing, groceries, utilities, bills, or child-related expenses. Similarly, 21.6 % of African American women and 10.5 % of white women started a new sexual relationship to receive financial support. It may be the case that relying on an overly narrow definition of female sex work may prevent an accurate understanding of the nature of transactional sex work among women in the US.

Additionally, the majority of the interventions included in our review were conducted over 10 years ago, which suggests that efforts to prevent HIV and STIs among FSW should be updated. Of the 19 interventions included in this review, only three met criteria for inclusion in the CDC’s Compendium of Evidence-Based HIV Prevention Interventions: the Women’s CoOp [44] the Negotiation Intervention [41], and Community PROMISE [64]. Inclusion in the CDC Compendium is based on review criteria that include quality of study design and implementation as well as strength of the findings (see: http://www.cdc.gov/hiv/dhap/prb/prs/efficacy/rr/criteria/index.html). Many of the interventions included in this review employed designs too weak to be considered for inclusion in the compendium despite even if they reported positive results.

Due to the outdated nature of the literature, some of the newer biomedical, behavioral, and structural advances in HIV prevention are not being utilized or evaluated among FSW despite high vulnerability to contracting and transmitting HIV. Some studies include condom distribution programs [65] or antiretroviral pre-exposure prophylaxis (PrEP) among heterosexually active women and injection drug users [66]. PrEP, in particular, shows promise in preventing HIV transmission among FSW, but adherence can be reduced by factors such as substance use and frequent geographical displacement [66].

In the time since the database was searched, an additional intervention that targeted FSW was published [67]. This intervention tested the efficacy of a case management framework in which FSW were randomly assigned to a strengths-based intervention led by either a professional case manager or by a peer. Results demonstrated that both interventions equally reduced HIV risk behaviors and increased service utilization. Given that this study sought to specifically address issues relevant to the lives of FSW (e.g. housing, social support, stigma), the observed positive results are expected and in line with the recommendations from this review.

There are a few limitations to this study. One of the primary limitations is that the lack of consistent quantitative information reported in the studies precluded a meta-analysis of the outcomes. We did link content areas to reported outcomes to observe possible qualitative trends, but this effort was also limited due to the fact that some content areas were not adequately included in the interventions (e.g. mental health/victimization, economic resources, and psychosocial variables). Similarly, this review identified a relatively limited number of studies, and only 5 studies that specifically targeted FSW. We believe that the small number of studies highlights the relative dearth of female sex-worker focused intervention studies being conducted within STI or HIV prevention in the US, despite the need for such efforts. It is possible that additional intervention research studies did indeed include FSW in their study population, but did not stratify or report results specific to FSW, thus rendering these studies ineligible for our review. In addition, because our search did not extend to the grey literature, it is possible that additional interventions for FSW exist but have not been scientifically tested or published. Despite these limitations, we believe that this review highlights ways in which HIV/STI prevention efforts may have underserved FSW in the United States. It is our hope that this review will inform future efforts to tailor risk-reduction approaches to address issues facing US-based FSW.

Conclusion

The purpose of this systematic review was to examine HIV/STI behavioral interventions conducted in the US that aim to reduce sexual- or drug-related risk behavior among FSW in order to highlight current gaps and identity potential future directions. We reviewed three types of interventions: interventions that targeted FSW, interventions that stratified data by FSW, and interventions that included a majority of FSW in the intervention without targeting them explicitly. Our findings suggest that majority interventions tended to include more content than targeted interventions, likely due to the prioritization of HIV prevention among high-risk women who used drugs during the height of the HIV epidemic. Our findings also demonstrate that while most interventions typically provided general HIV and substance use prevention information, few interventions tailored content to focus on issues specific to FSW. Existing HIV/STI prevention efforts should be updated to address the unique needs of FSW, including an emphasis on structural and psychosocial risk factors as well as increasing access and adherence to biomedical approaches that can benefit all high risk populations.