Introduction

We are currently in an unprecedented era as policymakers apply the latest advances in HIV treatment and prevention science to end the HIV epidemic in the United States (U.S.). During the 2019 State of the Union Address, the President announced a plan to end the U.S. HIV epidemic by reducing new infections by 75% in five years and 90% in 10 years [1]. To achieve these goals, the plan is organized around four key pillars: diagnose, treat, protect, and respond. The first, diagnose, focuses on improving early and timely detection of HIV cases. Treat stresses rapid linkage to HIV care, and initiation of antiretroviral therapy to achieve viral suppression, thus eliminating onward transmission [2, 3]. Protect emphasizes protecting those at risk for HIV from becoming infected using novel prevention methods such as pre-exposure prophylaxis [4,5,6,7]. Lastly, respond highlights a rapid response to growing HIV infection clusters and prevention of new ones [8, 9]. Ultimately, the success of the plan hinges on effective strategies to promote HIV testing, the first step in the HIV treatment and prevention cascades [10,11,12].

Of the estimated 1.2 million adults and youth living with HIV in the U.S., approximately one out of seven individuals do not know their status, and 45% aged 13–24 years are unaware of their status [13]. The Centers for Disease Control & Prevention (CDC) currently recommend routine testing during clinical encounters [8] and testing through several non-clinical settings and approaches [9]. These include voluntary counseling and testing sites hosted by trusted community-based organizations as well as venue-based testing, such as at gay Pride events. In addition, public health disease intervention specialists interview those recently diagnosed with HIV to notify sexual partners and encourage them to pursue HIV testing [14,15,16]. Moreover, self-testing at home has emerged as an important strategy, gaining even greater traction during the COVID-19 pandemic given limits to in-person visits [17,18,19]. The majority of tests performed in non-clinical settings in the U.S. occur in the context of CDC-supported HIV counseling, referral and testing (CRT) services with over 3.2 million tests conducted annually, yielding an overall test positivity of 1% [10]. The efficiency and acceptability of any non-clinical testing approach are of particular importance to jurisdictions seeking to invest limited resources in methods with higher case detection rates, especially for key populations at higher risk for HIV acquisition, such as men who have sex with men (MSM), who are encouraged to test at least once annually or more frequently [20].

Recent reviews have pointed to the promise of social network strategy (SNS) to efficiently reach key populations for HIV testing [21]. SNS builds on over 40 years of epidemiologic and interventional studies that have leveraged social networks for participant recruitment, including snowball sampling, respondent-driven sampling (RDS), and long-chain peer referral [22,23,24]. SNS is grounded in the idea that members of a social network share the same or similar risks for HIV, tend to trust each other, and may be more willing to adopt behaviors endorsed by members of their network. SNS enlists an initial group of persons at elevated risk or living with HIV as “seeds.” These seeds are then tasked with recruiting other persons within their social networks (i.e., network associates) to test for HIV and engage in prevention or treatment services. Seeds receive training and education to help them identify network associates and motivate others to pursue testing, and they often receive incentives to support their recruitment efforts. This method has been shown to effectively detect new HIV positive cases at rates of 5% or higher [25].

There is a strong theoretical underpinning for how social networks might optimize HIV testing. Social Network Theory studies the relationships and interactions of social groups, communities, and their various networks [26]. Centrality, which identifies how densely connected an individual is to others in their network, is fundamental to the success of the strategy [27] and prioritizes recruiters who are better connected to their social networks. Egocentric networks are tightly connected to one individual, who knows many others, whereas socio-centric networks connect multiple people in a network who, in turn, may be connected to numerous others [28]. Egocentricity is important in the selection of initial seeds, and successful propagation to subsequent waves requires sufficient socio-centricity. In addition, SNS applies the Theory of Planned Behavior, which identifies social norms and pressures as levers in influencing attitudes toward testing, testing intentions, and perceived control of the behavior [29].

While prior studies have documented the efficiency of SNS for HIV case detection, little is known about the facilitators and barriers to SNS implementation or what factors may influence SNS programs' operational success. To speed the translation of evidence to public health practice, we conducted a systematic review of the SNS literature to identify these characteristics and offer recommendations for community-based organizations and public health agencies considering this approach.

Methods

We conducted a systematic review of the published literature using PubMed and Web of Science databases aligned with the PRISMA criteria [30]. We used a combination of the following terms: “social” and “network” and “strategy”; and “HIV” or “human immunodeficiency virus;” and “United;” and “States”. The eligibility criteria for inclusion in the review were as follows: included key populations, such as MSM, person using intravenous drugs, and racial or ethnic minorities; were peer-reviewed, empirical evaluations; were based in the U.S., and focused on SNS specifically applied to HIV testing. We included publications dated from 1981, the start of the social network literature, through June 2020.

Based on these criteria, we identified a total of 979 papers from PubMed and Web of Science to review. We conducted our systematic review of these articles separately for each database. We did not pool databases and remove duplicates at the onset, as we used it as a screening quality metric to assess overlap in our screening between the databases. Therefore, the numbers presented hereafter may include duplicates (from Web of Science and PubMed). Based on the inclusion criteria, we removed a total of 411 studies that were not based in the U.S, 351 papers because they were not focused on HIV testing, and 173 papers because they were not SNS-specific (e.g., they instead focused on respondent-driven or snowball sampling). We excluded an additional eight studies because they were not empirical studies and removed five more because they were not SNS and were missing HIV testing as an outcome. We explicitly included studies that discussed facilitators and barriers of this approach. After combining the two sets of reviews (n = 31 studies), we removed 17 duplicates, leaving a total of 14 unique studies to include in our analysis (Fig. 1). Given that 55% of the studies were present after reviewing both databases, we believe this supports the quality of our screening and review processes. The papers left for inclusion were published between 2009 and 2018.

Fig. 1
figure 1

Diagram of systematic review search and excluded research articles

Full size image

Analysis of Papers

We used thematic analysis to analyze the key factors associated with successful implementation of SNS. We identified themes to understand who SNS reaches for testing and the facilitators and challenges to successful SNS implementation [31]. First, the first author (KS) began to familiarize himself with the findings and main conclusions. Next, KS analyzed the methodological approaches, results, and discussions to understand which priority populations were of primary interest, the studies' locations, and risk of bias. We assessed the risk of bias by exploring potential threats to studies’ internal and external validity. For example, we evaluated study designs (e.g., cross-sectional, cohort), presence of comparison groups, and confounding analyses.

KS then extracted the quantitative metrics that typically accompany SNS, including (1) total number of recruiters; (2) the total number of network associates (recruits) recruited; (3) network indices, defined as the number of network associates recruited divided by the total number of recruiters; and (4) new HIV positivity rates (i.e., number of new cases of HIV detected). New cases were determined by reviewing epidemiological surveillance data in some studies, others were cohort studies, and others relied on self-reported knowledge of serostatus. We incorporated this variation in the results. We developed the key themes, both a priori and posteriori. For the a priori themes, we were guided by SNS theory and its critical components, including working with recruiters, incentives, and trust and confidentiality. Posteriori factors were determined by the thematic review itself, including real-world implementation factors, the collaboration required to implement SNS, and the strategy's sustainability. The key themes included: (1) social network and recruiter characteristics; (2) strategies for and effectiveness of recruiting key populations; (3) use of and types of incentives; (4) trust, confidentiality, and stigma concerns; and (5) implementation plans and real-world guidance. KS also identified several subthemes under these main themes to expand upon the findings.

Results

SNS and HIV Detection Rates

Half of all the studies were cohort studies [32,33,34,35,36,37,38], and the other half were cross-sectional [39,40,41,42,43,44,45]. The majority, nine (out of 14) studies, detected an HIV positivity rate over 1%. Of these nine studies, five were cohort studies and four were cross-sectional studies. In the five studies that did not demonstrate HIV rates above 1%, one study was a cohort study in a low-prevalence area (0.49% positivity rate) [33]. In another, SNS was implemented by an infectious disease clinic and emergency department in a cohort study (1% positivity rate) [35]. Another two were cross-sectional studies in larger geographic areas and focused on Latinx communities (positivity rates of 0.26, 0.37% respectively); these two studies did not achieve their desired sample size [40, 41]. The last was a cross-sectional study that had a 0% positivity after confirming diagnoses with the health department [42]. Six studies (five cohort, one cross-sectional), out of 14, utilized clinical or health department data to validate the positivity rates [33,34,35, 37, 38, 42]. Table 1 describes the relevant studies, including the study location, study metrics, and the key populations reached.

Table 1 Characteristics of U.S. Studies Reviewed, Positivity Rates of New Infections and Network Metrics, 2009–2018
Full size table

Factors That Influence the Implementation of SNS

While most published studies document the ability of SNS to uncover undiagnosed HIV cases, various factors promote successful implementation (Table 2).

Table 2 Qualitative and quantitative social network strategy study results and risk of bias assessments (n = 14), 2009–2018
Full size table

The thematic analysis of the 14 studies identified five major areas related to SNS implementation: (1) social network and recruiter characteristics; (2) strategies for and effectiveness of recruiting key populations; (3) use of and types of incentives; (4) trust, confidentiality, and stigma concerns, and (5) implementation plans and real-world guidance. These and the respective subthemes are summarized in Table 3.

Table 3 Themes and subthemes related to implementation of SNS, 2009–2018
Full size table

Social Network and Recruiter Characteristics

SNS relies on recruiters to engage with their social networks and persuade persons to test for HIV. From a program’s inception, implementers must clearly define the priority populations, learn about the connectivity of networks, and appreciate recruiters’ centrality in their networks. From the reviewed papers, there was a range of network indices, a standard SNS measurement. The network index is defined as the number of network associates recruited divided by the total number of recruiters. Across all the studies, the network index ranged from 0.8 to 10.6 (Table 1). The wide range of indices reflects the variability in recruiters’ centrality within the network and success in recruiting network associates. For example, in one study, 32% of recruiters accounted for 91% of linked network associates [44].

SNS assumes people will have similar HIV statuses or associated risks, and sociodemographics. Many of the reviewed papers, 12 out of the 14, indicated that network associates who tested comprised of key populations, including MSM, those having condomless sex, and persons who have not tested before [32,33,34,35, 38,39,40,41,42,43,44, 46]. Two papers indicated that recruiters’ demographics, such as race, ethnicity, and gender, were not associated with the demographics of those recruited [44, 46]. Three studies did show that PLHIV recruiters were more likely to recruit network associates that tested positive for HIV [40, 44, 46]. To optimize HIV testing efficiency, the review of papers underscores the importance of working with recruiters who intimately know their networks, can foster trusting relationships, and have similar risk factors.

Strategies for and Effectiveness of Recruiting Key Populations

Many of the studies discussed the facilitators and barriers to effective recruiting. One theme that arose was the need to understand the risk factors of the network. One of the studies described the need to explicitly examine the sexual risks of the social network before beginning to recruit [40]. A cohort study in an CTR designated emergency department in a low prevalence area found that many recruiters brought in family members and acquaintances, which indicates no identifiable risk [35]. Relatedly, recruiter comfort with discussing HIV and risks is important. In a cross-sectional study, recruiters that recruited more than two network associates found that recruiters who indicated that telling girlfriends about knowing HIV status and the high rates of HIV in their community was associated with successful recruitment [40]. Barriers in this study included lack of time, difficulty in speaking about HIV, concerns about network associate believing recruiter was HIV positive, and girlfriends were afraid to know their HIV status [40]. The review of studies indicates that having the “correct” recruiter is important to the success of SNS.

Another theme was that SNS implemented in collaboration with community organizations were better equipped to find effective recruiters. In total, eight studies worked with community-based organizations (CBOs) for recruitment [32,33,34, 36, 39, 42,43,44]. Anecdotally, one agency in a cohort study, indicated they used social media to recruit network associates; this agency had the largest network index (number of network associates / number of recruiters) [34]. A study by Kimbrough noted that partnership with CBOs was important for successful and effective recruitment [44]. In a cross-sectional study, the recruiters were identified by the health department, which may have recruited different seeds and networks, as compared to community organizations [42]. In another study they described a more expansive strategy for locating recruiters including through local support groups, local gay bars, word of mouth, and through CBOs [46]. Most studies found that partnerships with CBOs supported finding effective recruiters.

Lastly, the use of social peers helped to improve the effectiveness of SNS to get people to test. In a cross-sectional study, 65% of SNS participants agreed that encouragement from peers was a facilitator to testing, as compared to 42% in AVT (p < 0.0001) [41]. In a qualitative study, peer recruiters reported positive experiences with distributing HIV test kits, with most stating the training prepared them and that they were motivated to help their community to test [46]. Studies that did explore the importance of peers indicated how peers help to improve HIV testing in SNS.

Use of Incentives

Incentives were used in 13 out of the 14 studies; however, there was a wide range of incentives offered, their purpose, and to whom the incentives were given. The first type of incentive was for the recruiters. For recruiters, in 10 studies, the range of incentives was $10–25 per network-associate recruited [33,34,35, 38,39,40,41,42,43,44]. In four of the studies, the network associates had to complete their HIV test for the recruiter to receive the incentive [33,34,35, 38]. Four studies also gave recruiters separate incentives for agreeing to recruit, which ranged from $10 [35] in one study, $20 in another study [38], $35–60 in another [41], and $100 in a study when the recruiter finished training [46]. Network associates also received an incentive when they tested, which ranged from $5–25 [33,34,35, 38, 40, 43, 44, 46]. The types of incentives offered varied, for example cash, Visa and Amazon gift cards, and transportation vouchers.

There were also challenges with incentives. One study, with many repeat testers, indicated that interest in the incentives might have been the prime motivating factor for recruiters and network associates [33]. In this study, some recruiters exceeded the 20 contacts specified by their protocol (range 1–63). In two other studies, one cross-sectional and one cohort, there were concerns that the incentives inadvertently resulted in a high number of individuals already living with HIV [38, 43]. In another cohort study, after feedback from participants, the incentive amounts were increased to align with other local testing services. This increase appeared to improve participation [38]. In one cohort study, researchers did not provide incentives, yet reached a 2.1% positivity rate [32]. In total, 13 studies that offered incentives to promote successful recruitment indicated the potential utility of incentives, which was particularly useful when the incentives were aligned with participants’ needs [33,34,35,36,37,38,39,40,41,42,43,44, 46]. Overall, the majority of studies used incentives. Many described their benefits, but additional examination is needed to better understand their utility and potential pitfalls.

Trust, Confidentiality and Stigma Concerns

SNS leverages the trust between recruiters and network associates to encourage testing. However, stigma and marginalization can impede the utility of SNS to reach key populations. One study that focused on testing among Latinx women found that stigma remained a barrier to testing [40]. In four studies, SNS recruited heterosexual-identifying MSM, a highly stigmatized key population [39, 41,42,43]. Another study used SNS to distribute HIV self-testing kits to further reduce barriers to testing and concerns with confidentiality [46]. Qualitative results from this study indicated high levels of acceptability to test at home, as compared to the clinic, because of the opportunity for additional privacy and anonymity. However, there was no comparison group or adjusted analysis for this study. The review of studies supports SNS as a strategy that can reduce barriers to HIV testing by leveraging trust within networks; however, stigma associated with HIV testing remains.

Implementation Plans and Real-World Considerations

The papers in this review highlighted the importance of SNS program organizers to engage with relevant stakeholders prior to implementation. All but one of the SNS programs reviewed represented a collaboration between CBOs and/or health departments. Health departments were able to reference surveillance records to reconcile whether the positivity rates were incident or prevalent cases. Of the 14 papers, eight studies (six cross-sectional and two cohort) relied on self-report to “confirm” an incident diagnosis rather than health department or clinical records, which makes the studies subject to recall bias [32, 37, 39,40,41, 43, 44, 46]. Health departments are a crucial collaborating partner for SNS to cross check HIV surveillance data when assessing positivity rates.

Implementation of an SNS program requires thoughtful consideration and planning for how to balance SNS requirements with the organization’s existing policies and programs. Four studies highlighted that engagement with organizational leadership and staff is key to the success of SNS programs [33, 38, 42, 44]. The four studies found that staff described difficulty taking on additional SNS-associated job duties such as tracking referrals, linking recruiters to network associates, training and supporting recruiters and supplying incentives. Also, SNS training and coaching of recruiters may be time-intensive [38, 44]. In another study, at four different sites, it was reported that SNS was least familiar to staff, it required more training, and implementation was more time consuming, as compared to AVT or targeted outreach [32]. However, SNS, when implemented by CBOs, reduces the number of persons who need to be recruited to find a undiagnosed case of HIV, as compared to other testing strategies [44]. Another cross-sectional study indicated that staffing changes and other logistical challenges hampered SNS efforts [41]. The review of papers elucidated the importance of dedicating staff and resources in order to support successful and efficient implementation of SNS. Relatedly, in the review, two of the studies conducted cost analyses. One of the studies, which had a limited number of HIV diagnoses (two positive cases out of 24 tested), showed that SNS was cost-saving for one site in the study, as compared to venue-based and voluntary counseling and testing [37]. The other study included retrospective cost data and matched unavailable cost data (e.g., mobile van costs, staff wages, and time spent on counseling and testing activities) to other similar jurisdictions. In their analyses, 72–85% of the fixed costs were related to program management, start-up costs, facilities, and utilities, which they anticipated reducing as the program matures [36]. The largest variable cost was on identifying and training recruiters. There could be other potential costs, such as home-testing kits, depending on the testing strategy used, and various incentive costs.

Risk of Bias

There were biases with some of the studies that may limit interpretation of results. As described earlier, half the studies (n = 7) were cross-sectional, of which six relied on self-report to determine positivity rates. Recall bias may pose a challenge with validity in the cross-sectional studies. Five of the seven cross-sectional studies included comparison group analyses [39, 41,42,43, 45] and four of the seven cohort studies included comparison groups [32, 33, 35, 37]. Five cohort studies reconciled their diagnoses with health department data, which improves validity of their positivity results [33,34,35, 37, 38]. Twelve out of the 14 studies did not have adjusted statistical analyses, which doesn’t address issues of confounding (Table 2). Only two studies focused on cost, one of which had limited numbers of persons HIV testing making the sample size small. The three mixed-method studies were robust and included quantitative and qualitative assessments that explored not only the metrics and the yield of the strategy, but the processes behind SNS. Many of the studies, especially the cross-sectional ones, were of limited duration (i.e., less than 12 months); therefore, the durability of the response is difficult to assess.

Discussion

HIV status awareness is essential to advance HIV treatment and prevention. Our systematic review of the published literature to evaluate SNS’s role in detecting new HIV cases confirmed, through health department surveillance and cohort studies, that positivity rates exceeded those using standard HIV counseling, referral and testing (1% positivity). Our thematic analysis revealed that successful SNS implementation was fostered by effectively tapping into densely connected socio-centric networks, offering incentives that align with recruiter and network needs, and leveraging strong organizational leadership and buy-in from staff. In many of the studies, SNS was able to reach key populations at heightened risk for HIV, including heterosexual-identifying MSM, persons who have never tested, and persons engaged in sexual and substance use risk behaviors.

Our findings align with other research that synthesized strategies to improve HIV testing, including a review of 15 global studies that used SNS. Campbell and colleagues noted that SNS increased HIV positivity rates from 4 to 31% across the 15 studies, nine of which were in the U.S. [21]. Campbell et al.’s synthesis found that SNS was particularly helpful to organizations and communities that historically had limited success reaching key populations for HIV testing [21]. However, that study did not examine whether positivity rates were from incident or prevalent cases. This was a strength of our analysis, in which we were able to draw from cohort studies and studies that collaborated with health departments to confirm new HIV diagnoses. Additionally, success of SNS may hinge on a small percentage of recruiters. For example, in one study in our review, 34% of recruiters did not recruit any network associates, and 32% of recruiters accounted for 91% of all linked network associates [44]. The wide range of network indices found in the review of studies, 0.8 to 10, indicated that certain recruiters may be better connected to their networks, and thus more successful in recruitment. Understanding the recruiter and their role in their network is important for the success of SNS. More explicit use of social network theories during enlistment of recruiters may be beneficial [26].

Implementation of SNS must also explore its relationship to other real-world factors that may influence its ability to reach key populations for HIV testing. For example, having explicit implementation plans, dedicated staff to SNS, and understanding how SNS overlays onto services is important (e.g., HIV testing hours). Many of the reviewed studies were 12 months or shorter, which contains challenges with understanding the durability of the public health practice. While SNS helps find new cases of HIV, testing is only the first step in the U.S. “End the HIV Epidemic” plan [1]. Linkage to treatment and prevention services are critical to fully leveraging the benefit of SNS. Understanding the underlying systems of care for persons living with HIV and those who are negative is critical to the success of SNS. As uncovered by one of the reviewed studies, 60% of those who tested positive during SNS were still not engaged in care, and it took a month to link the other 40% to care [34]. The underlying systems of HIV care and prevention, including adjacent services, such as substance use, mental health, serve as the foundation for SNS’s success. Research has extensively described how collaboration with diverse stakeholders and inclusion of a multiplicity of services are critical to HIV care and prevention (47).

There are limitations to this systematic review. First, a key question in the field of SNS is defining to what extent concordance in race and ethnicity, gender identity, sexual orientation, among other factors, may be the most relevant to consider when selecting recruiters. Some studies found no variation in network associate’s demographics based on the recruiter’s demographics, whereas others did. Secondly, the first author was the only reviewer of the systematic review. However, the findings and tables used for analysis were shared and discussed among coauthors to ensure the accuracy of the interpretation of findings and the review. Search terms and exported search data was shared with the senior author in order to confirm the interpretation of study inclusion. In addition, 55% of the studies were found in both databases after conducting the review on each database separately. Another limitation is the caution needed to interpret the positivity rates, which is particularly true for the cross-sectional studies that relied on self-report. The cohort studies that confirmed cases with health department data and included comparator groups provide the most robust evidence for the improved reach of SNS. Lastly, many of the studies relied on convenience samples, which creates selection bias and limits the generalizability of the findings outside the study populations reviewed.

Conclusion

SNS is a promising approach to increase case detection that underpins the U.S. plan to end the HIV epidemic by 2030. SNS programs that make use of available HIV surveillance data, engage relevant stakeholders, and dedicate sufficient resources to program staff and meaningful incentives for participants are well positioned to improve HIV testing efficiency with key populations.