Background
Global epidemiological surveillance data indicate that in 2015 about 36.7 million people were living with HIV. Of these, 2.1 million were new HIV infections in 2015 and about half of the newly HIV infected people were adolescent girls and young women [
1,
2]. In the last decade, the incidence of HIV has decreased in several developed countries [
3,
4]. However, the spread of HIV among high-risk groups such as men who have sex with men (MSM), female sex workers (FSWs), and people who inject drugs (PWID) is relatively high (44%), especially in developing countries [
5,
6]. According to the World Health Organization report in 2015, about 74% of new HIV diagnoses were due to sexual transmission, 4% to injecting drug use, and for about 20% of the new diagnoses the transmission mode was reported to be unknown [
7]. The UNAIDS 2016–2021 Strategy, with the aims to reach zero infections, absence of discrimination, and zero AIDS-related deaths, highlights the need for effective HIV prevention strategies for key populations [
8].
HIV risk behaviors of injecting drug use and risky sexual behaviors are multidimensional and occur based on biological, individual, and structural factors. While individual attributes (including sex, age, education, occupation, and ethnicity) may influence a person’s attitudes, beliefs, and behaviors, various macro-level and social context characteristics can also contribute to engagement in, and continuation of, HIV risk behaviors [
9].
A body of literature emphasizes the importance of social networks in HIV transmission and prevention [
10‐
17]. As one of the first pieces of evidence on the key role of social networks, using data from 40 MSM with AIDS, Auerbach and colleagues reported in 1984 that HIV could be transmitted through sexual contacts and that having multiple sexual partners increases the probability of HIV transmission [
18].
Since that time, numerous studies have shown that interpersonal interactions occurring in social networks, as well as network characteristics, are critical to understanding HIV risk behaviors and spread of infectious diseases more generally [
11‐
16,
19,
20]. It has also been shown that social network approaches may be helpful for HIV intervention and to allocate resources more efficiently for preventive strategies [
21].
A social network is a set of ties among people who have some common interests or interactions [
22]. Family, friends, neighbors, coworkers, and sex or drug partners may be members of the social network that influence HIV-related behaviors. A network can affect the members’ behaviors and health outcomes; this may be based on structural network characteristics such as size (the number of members of the network), density (the extent to which network members are connected to each other), degree (an individual’s number of direct ties), betweenness (frequency of ties with which an individual is on the shortest path connecting pairs of others in the network), centrality (extent to which an individual has a central position in the network), and homogeneity (similarity between network members) [
9,
23‐
26]. Rothenberg et al., in a study of in sexual transmission of syphilis among teenagers in rural Georgia, showed that structural characteristics of the network position of individuals such as degree, betweenness, and information centrality facilitated the transmission of syphilis. Participants with syphilis had a higher degree (on average 7.4 sexual partners) compared to those without syphilis (2.4 sexual partners). Similarly, the participants with syphilis had an average betweenness of 4.1, which was higher than the average betweenness of 1.7 for those without syphilis. This network parameter indicates that participants with syphilis were more central in the network than those without syphilis [
27].
According to the literature, larger networks provide more opportunities for exposure to a variety of risks, health information, and practices affecting health behaviors and outcomes of network members [
9,
28]. Furthermore, HIV risk behaviors often occur in the context of a dense social network where risk behaviors are normalized, and information can pass easily and frequently between individuals [
9,
29]. Some studies have shown that larger sexual networks are associated with increased reporting of unsafe sex among MSM and of syringe sharing among PWID [
25,
28,
30]. Also, social networks may influence risk and health behaviors through various psychosocial mechanisms and tie characteristics such as frequency of contact, tie duration, social influence, social norms, close contacts, provision of social support, and social capital [
31]. One study among PWID in India showed that the PWID who had more than 10 PWID in their drug network were 1.65 times (95% CI: 1.12 to 2.42) more likely to have shared a syringe at the last injection compared to those who had 0 or 1 member in their networks. These authors found also that participants with the largest injection drug network size were 31% (95% CI for relative proportion: 0.53 to 0.90) less likely to be virally suppressed compared to those with the smallest network size [
32]. A study among 385 male migrants in China showed that condom use norms of the core network were significantly associated with the participants’ condom use. Participants with one or more network members who always used condoms were 12 times (AOR: 11.9, 95% CI: 2.4–59.0) more likely to consistently use condoms than participants with no such alters in their sex work networks [
33].
While some studies have investigated the association of social network structure and function with HIV risk behaviors in teenagers, HIV-at-risk women, PWID, and MSM [
29,
34‐
37], there are few studies that have systematically reviewed the existing literature about this association for FSWs, who are an important group at risk and also hard to reach in many countries [
5]. In addition, they may be a bridge group for HIV transmission to the general population [
38,
39].
Some systematic reviews have assessed networks and health. Perkins et al. focused, in a systematic review in 2013, on how social network structure and influential individuals within a network may reinforce health outcomes and behaviors in low- and middle-income countries [
40]. They found network composition, position, and structure to be related to health outcomes and behaviors. Although these authors considered HIV transmission as one of the health outcomes in the general population, they did not consider HIV risk behaviors specifically among FSWs. Qiao et al. in 2014 conducted a systematic review on the association between social support and HIV-related risk behaviors among groups such as drug users, MSM, adolescents, people living with HIV/AIDS, and FSWs [
41]. This review found 5 studies on FSWs and confirmed the role of social support in reducing HIV risk behaviors.
Despite these interesting findings, these reviews have produced limited information on the role of social networks for HIV risk behaviors among FSWs. Therefore, a systematic review with a focus on functional and structural characteristics of social networks may be valuable to support future interventions for HIV prevention among FSWs. The purpose of the present review was to review and summarize existing quantitative and qualitative studies about network structure and function of FSWs and their association with HIV risk behaviors.
Discussion
In recent decades many countries, especially developing countries, continue to experience a steady increase in the numbers of people living with HIV/AIDS. About 74% of HIV transmission is related to sexual contacts [
2,
7]. Female sex workers are among the most important groups who are at risk of HIV. Given the challenges of prevention of HIV transmission, it is important to focus on HIV risk behaviors among this high-risk population. The present systematic review considered studies of social networks of FSWs and, in particular, the functional and structural characteristics of the networks and their associations with HIV risk behaviors.
Only 19 studies were identified over the period between 1990 and 2016, none of which focused exclusively on the association of social network characteristics and HIV risk behaviors of FSWs. Most of the relevant studies focused only on the role of social support and social capital in condom use among FSWs [
38,
61‐
63,
67,
77]. Only four of the included studies assessed the structural characteristics of FSWs’ social networks and their association with HIV risk behaviors [
68,
71‐
73].
Most of the studies did not include a network name generator and an associated network interpreter to obtain relevant information about the content of the ego-alter relationships [
61,
63,
67,
79]. Although these studies did provide some information about exchanges of social support and other resources between egos and alters in these networks, using roles such as regular partner, client, peer or co-worker, and relative, we do not have sufficient information about quality and quantity of these relationships and of the exchanges of social support and other resources.
Consistent with past research by Qiao [
41], we found that social support and social capital as functional characteristics of social networks were significantly associated with HIV risk behaviors, especially condom use. The broader focus of the current review compared to that by Qiao led to finding a greater number of studies (19 vs. 5), more information about name generators and name interpreters, and more extensive results in terms of functional and especially structural characteristic of FSWs’ social networks and their association with HIV risk behaviors. Given the findings of our review, peers and gatekeepers appear to have a key role in the social network of FSWs and can affect their condom use [
38,
61‐
63,
68,
70,
75]. Most of the included studies showed that social support, trust, intimacy, and solidarity with peers and gatekeeper are positively associated with condom use among FSWs [
38,
61‐
63,
68,
70,
75]. When frequency of contact, trust, and social support are high, peers and co-workers can be effective in educating FSWs about prevention of HIV risk behaviors [
74,
82]. Peers and co-workers can facilitate the dissemination of messages about protective health behaviors, teach each other, and improve each other’s power of negotiation with clients or sexual partners about condom use. A lack of social support, by contrast, may increase social isolation and reduce the motivation for learning safe sex behaviors from peers, disclosing HIV status, and insisting on protective behaviors, such as condom use [
64,
66,
75,
83]. Further, some of the included studies showed that gatekeepers (pimps or establishment owners) who manage sex workers have an important role in condom use of FSWs and their clients [
61,
62]. Gatekeepers can provide a supportive environment for safe sex behaviors via their educational messages, determine condom use rules in the workplace, and enforce client condoms use. However, one of the studies found that gatekeepers may be a barrier for promoting safe sex among FSW due to the conflict of interest between financial benefits and the FSW’s health [
67]. Our findings suggest that interventions for promoting condom use among FSWs should consider the role of gatekeepers and peers, and of contextual factors such as contact frequency, trust, intimacy, and social support in the social network of FSWs. Interventions should aim to improve trustful and supportive relationships with peers. Furthermore, condom use messages should be designed to be easily disseminated through the peers or gatekeepers in the FSWs’ social network. Valente et al. in their study on the association between social networks and contraceptive use among women in Cameroon found that the women’s contraceptive use was associated with their perception of their network partners’ support for contraception and with their network partners’ encouragement for contraceptive use [
20].
These findings can be used to develop and implement relevant behavior change interventions. Potential intervention strategies include training programs for two types of network members. The first approach would be to train peers to diffuse safe sex information and skills in condom use negotiations with clients and regular partners [
84]. A second approach would be to train gatekeepers in creating supportive norms of condom use, by providing educational programs and condom use skill trainings, enacting a mandatory condom use rule with penalties for the rule’s violation, promoting condom use negotiations with clients, and by providing FSWs with free condoms [
85]. For example, a peer education program intervention for HIV prevention among FSWs in Bangladesh, which assessed the effects of social support provided by peer educators to FSWs, found that the FSWs who received more informational support or emotional support from their peer educators reported a higher rate of using condoms, more self-efficacy, as well as lower self-reported STI symptoms at follow-up [
84].
In addition, the review findings show that structural characteristics of FSWs’ social networks are associated with HIV risk behaviors [
71,
72]. However, only four of the relevant studies assessed the structural characteristics of FSWs’ social networks and their association with HIV risk behaviors [
68,
71‐
73]. The results of the longitudinal articles showed that the network size, density, network position, centrality, and stability might have a role in HIV transmission [
71‐
73]. According to this study, a large dense network is more likely to have members who share HIV risk behaviors with each other. This finding supports the perspective that dense networks can provide more pathways along which behaviors, as well as diseases, may flow [
9]. Also, based on the findings, FSWs and their clients who are HIV positive and are connected directly or indirectly in the sexual network cannot only transmit the infection to each other but also act as a bridge for HIV transmission to other networks and populations. The authors suggested that positions of persons in networks with different structural characteristics may have a different effect on the rate of HIV transmission. In a sexual network with low density and centrality, HIV positive persons who are clients and occupy a peripheral or isolated position may be less likely to transmit HIV to the network. By contrast, a densely connected network structure in which HIV positive persons are highly central may facilitate transmission of HIV [
72]. These findings also show that changes of the network might be crucial for understanding the dynamics of HIV transmission.
Only one of the included qualitative studies focused on both the structural and functional characteristics of FSWs’ social networks and their association with HIV risk behaviors [
68]. This study showed that structural and functional characteristics of FSWs’ different social networks (family and workplace network) might influence condom use of FSWs, be it negatively or positively. The authors found that support from peers and pimps in FSWs’ work network may promote their condom use. By contrast, family members, especially the presence of children in the social network may exacerbate the need to make more money and have a negative influence on FSWs’ safe sex behavior [
68]. This study also found that FSWs who had a larger work network, with higher density, and more frequent and supportive contacts with peers, had safer sex than those who were isolated or had smaller networks with low density and fewer contacts with peers. Despite these interesting findings, the qualitative results of this study do not provide strong evidence about the association of social network characteristics with HIV risk behaviors.
This review highlights the heterogeneity of approaches to measurement used to assess social network characteristics of FSWs. This heterogeneity is related to different study designs, different definitions of the concepts of social support and social capital, and the use of a variety of instruments for measuring social networks and their complex properties. Some limitations of the measures in the related studies were the lack of a theoretical or conceptual framework with respect to potential effects of social network characteristics on HIV risk behaviors of FSWs, insufficient measurement of social network characteristics, measurement of HIV risk behaviors only by self-reports, recall bias, and a lack of information regarding reliability and validity of instruments.
The present systematic review hopes to provide insights into understanding the social network characteristics of FSWs, especially the role of structural characteristics of these networks for HIV risk behaviors. This review provides evidence about the positive association of social support with condom use among FSWs. This information may help researchers and public health planners to develop HIV prevention intervention for FSWs. However, due to the heterogeneity of approaches to define and measure social support, we could not combine the results and generalize across all included studies. Despite the findings regarding the role of network structure for HIV transmission and risk behaviors, this evidence, based on just one quantitative and one qualitative study, is not sufficient to provide a reliable conclusion about the role of structural characteristics of FWS’s social network on HIV risk behaviors. Therefore, to address the question regarding which structural characteristics of FWS’s social network may affect HIV risk behaviors, it is necessary to conduct additional research.
Limitations of the present systematic review are the following. First, non-English and unpublished studies were not included. Second, the search strategy used was broad, but still some articles may have been missed. Third, collection of demographic variables and social network characteristics was not consistent across the studies; because of this diversity, the findings could not be combined in a meta-analysis. Fourth, most of the studies included were cross-sectional, so that it is impossible to draw any causal inferences between social network characteristics and HIV risk behaviors.
Despite these limitations, the findings of the present systematic review have important suggestions for future studies and interventions. First, future studies need to pay attention to methodological and measurement issues. For example, future studies should be guided by theoretical frameworks to examine the mechanisms expressing how social network characteristics may affect HIV-related risk behaviors of FSWs. In addition, we suggest that future social network studies use types of network inventory (name generator and interpreter) that are frequently used in personal network studies [
78], to provide sufficient information about quality and quantity of relationships between ego and alters in a network. The number of existing longitudinal studies was very limited, consisting of only one study which had three published articles. Longitudinal data are necessary to provide stronger information on causal relationships between social networks and HIV-related risk behaviors.
Second, further studies should consider structural characteristics of FSWs’ social network.
Social networks with different structural characteristics may have a different effect on HIV risk behaviors and HIV transmission among FSWs. For example, FSWs in a sexual network with high density, where more clients know each other and where the centrality of HIV-positive clients is high, may be more affected by HIV than FSWs who are engaged in a sexual network with a low density in which HIV-positive clients occupy a peripheral or isolated position [
72].
Information about structural characteristics of FSWs’ social networks such as density, degree, betweenness, and centrality that can facilitate diffusion of behaviors, information, disease transmission, is necessary to develop an effective HIV prevention intervention among FSWs. Such information will aid in the design of network interventions among FSWs and help policymakers to allocate resources for HIV prevention programs.
Third, to provide a more complete picture of FSW’s social networks, future studies should examine both structural and functional characteristics of social networks and their association with HIV risk behaviors among FSWs to provide sufficient information about the structural and psychosocial mechanisms through which the relationships between network members may affect health-related behaviors and outcomes of FSWs. Only one qualitative study in China considered both structural and functional characteristics of social networks and their association with HIV risk behaviors among FSWs.
Fourth, further studies should focus on the quantity and quality of ties among peers as well as ties with gatekeepers and sexual partners (clients and regular partners), and the dynamics of these relationships within the social networks of the target population. The HIV risk behaviors of FSWs may be embedded in power dynamics between FSWs, gatekeepers, sexual partners, and peers. For example, FSWs with smaller size peer network or low frequency of contact with peers who can support safe sex behaviors of FSWs may be more dependent on their partners and maintain the emotional intimacy with their partners even through unsafe sex [
66,
86].