To assess feasibility, the proportion of seeds who successfully recruited at least 1 contact was compared between groups using a χ2 test. The mean number of contacts recruited per seed was compared using analysis of variance.
To assess effectiveness, the prevalence of previous HIV testing, sexual behaviors, STIs, and HIV were compared between groups. Prevalence ratios (PR) and 95% confidence intervals (CI) were calculated using generalized estimating equations with a binomial distribution, log link, and exchangeable correlation matrix to account for clustering by seed. We also explored whether the prevalence of these behaviors and infections was different between the contacts of patients with established and acute HIV infection.
To assess efficiency, we calculated the proportion of contacts newly tested for HIV and compared proportions between groups using generalized estimating equations with a binomial distribution, log link, and an exchangeable correlation matrix. The number of contacts needed to test to identify 1 new case of HIV or any STI syndrome was also calculated using generalized estimating equations with a log link, Poisson distribution, and exchangeable correlation matrix.
Permission for collecting these data was granted by the Malawi National Health Science Research Committee and the School of Medicine Institutional Review Board at the University of North Carolina at Chapel Hill. All seeds and contacts provided written consent to participate. Information about seeds was not shared with contacts and vice versa.
Seed Participant Characteristics
Of 245 randomly selected clinic participants, 118 were eligible. The most common reasons for noneligibility were known HIV-positive status (N = 57, 45%), not meeting age or catchment area requirements (N = 13, 10%), being the sex partner of an STI patient (N = 15, 12%), or not having an STI (N = 12, 9%). Of the 118 eligible clinic participants, 76% consented (N = 90). In the community, 108 locations were visited to recruit 45 community seeds. Some coordinates did not lead to residences (N = 22), some led to residences with nobody home (N = 10), and some led to residences where no one met the eligibility criteria (N = 25). Of the 48 residences with an eligible person present, 93% were willing to participate. As specified by the protocol, the seed population included 45 newly diagnosed HIV-infected patients, 45 HIV-uninfected patients, and 45 community controls. Twelve HIV-infected seeds had AHI.
Among all seeds, 45% were male (Table 1), the mean age was 27.6 years, and most (61%) were married. Most (71%) had not used a condom in any of their last 5 sex acts. In the last 3 months, 12% exchanged sex for money and 16% had 2 or more sex partners, although proportions were higher among clinic-based seeds. Almost all seeds (80%) had been tested for HIV at least once, although this was lowest (60%) among HIV-infected seeds. Among HIV-infected clinic seeds, 27% had recently been diagnosed with AHI. Among community seeds, 1 (2%) had HIV. All clinic seeds had an STI except for 8 of the seeds with AHI.
Overall, the 135 seeds recruited 244 contacts (36% of the maximum number possible). The proportion recruiting at least 1 contact was somewhat higher among community seeds (69%) than among HIV-infected clinic seeds (47%) or HIV-uninfected clinic seeds (53%) (P = 0.09) (Table 2). However, among seeds recruiting at least 1 contact, the mean number of contacts was the same between the 3 groups: HIV-infected seeds: 2.9, HIV-uninfected seeds: 3.4, and community seeds: 3.3 (P = 0.5).
Among HIV-infected seeds, 39% of those with established HIV infection and 67% of those with AHI recruited at least 1 contact. The mean number of contacts recruited per seed was higher among those with AHI (mean = 2.0) than established HIV infection (mean = 1.1). This difference may have been due to additional counseling provided through CHAVI 001.
Social Contact Characteristics
Of the 244 contacts recruited, 228 (93%) participated. Of those who did not, most were ineligible due to being older than 45 years. Of the participating contacts, 62% were friends or neighbors of the seed, 18% were family members (primarily siblings and cousins), 11% were sexual contacts (primarily spouses), and 8% had another relationship. Most had known the recruiting seed for ≥1 year (79%), reported knowing the seed very well (87%), saw the seed several times each week (93%), interacted with the seed primarily at a home (81%), and described conversation as their primary activity together (89%).
Among contacts, 46% were male (Table 3), the mean age was 27.5 years, and most (59%) were married. Most (78%) reported at least on HIV test before the study. Most (76%) had not used a condom during any of the last 5 sex acts. In the last 3 months, 19% exchanged sex for money and 8% had ≥2 sex partners.
Contacts of the HIV-infected clinic seeds were more likely to be HIV-infected (31%) than contacts of community seeds (11%). HIV prevalence was 3.2 times higher (95% CI: 1.3 to 7.8) among contacts of HIV-infected clinic seeds than among contacts of community seeds (Table 4). Contacts of the HIV-uninfected clinic seeds were not more likely to be HIV-infected (10% established HIV infection and 1% AHI) than contacts of community seeds (PR: 1.1, 95% CI: 0.4 to 3.3). When analyses were restricted to nonsexual contacts, these PR estimates were similar: 3.0 and 1.4, respectively. When analyses were adjusted for seed sexual behavior (condom use and number of partners), PR estimates were also similar: 3.4 and 1.0, respectively.
The contacts of the HIV-infected and HIV-uninfected clinic seeds were more likely to have an STI syndrome (29% and 19%, respectively) than the contacts of the community seeds (9%). STI syndrome prevalence was 2.0 times higher (95% CI: 0.8 to 5.3) among contacts of HIV-infected clinic seeds and 3.2 times higher (95% CI: 1.4 to 7.2) among contacts of HIV-uninfected clinic seeds compared with contacts of community seeds. When analyses were restricted to nonsexual contacts PR estimates were similar: 1.9 and 3.3, respectively. When analyses were adjusted for seed sexual behavior, PR estimates were also similar: 2.1 and 3.2, respectively.
Contacts of seeds with established HIV infection and AHI were compared. The prevalence of HIV was nearly the same among contacts of seeds with established HIV infection (32%) and contacts of seeds with AHI (30%). Most contacts were not assessed for AHI. The prevalence of an STI was higher (24%) among the contacts of clinic seeds with established HIV infection than among contacts of clinic seeds with AHI (10%).
Of the 180 HIV-uninfected contacts, few (19%) were tested for HIV for the first time through the study. Of 35 contacts with HIV infection, 7 (20%) were being tested for HIV the first time through the study, 13 (37%) had tested HIV-negative previously and seroconverted afterward, and 15 (43%) already knew that they were HIV infected. Of the 20 contacts who learned their HIV-positive status through the study, 7 were recruited by HIV-infected seeds, 7 by HIV-uninfected seeds, and 6 by community seeds.
To identify 1 new case of HIV, 8.1 contacts of HIV-infected clinic seeds, 9.7 contacts of HIV-uninfected clinic seeds, and 17.5 contacts of community seeds were screened. To identify 1 new case of an STI, 5.5 contacts of HIV-infected clinic seeds, 3.5 contacts of HIV-uninfected clinic seeds, and 11.4 contacts of community seeds were screened. To identify 1 new case of an STI or HIV, 3.7 contacts of HIV-infected clinic seeds, 2.8 contacts of HIV-uninfected clinic seeds, and 7.3 contacts of community seeds were screened.
Asking STI patients to recruit their social contacts was a feasible, effective, and efficient way of diagnosing new HIV cases in a generalized HIV epidemic. Half of the clinic seeds in our study were able to successfully recruit at least 1 contact, and contacts of HIV-infected clinic seeds had a higher HIV prevalence than contacts of community seeds. To identify 1 new case of HIV infection, only 8–10 contacts of clinic seeds needed to be tested for HIV, much better efficiency than random testing in the population.
High-risk persons tend to associate with other persons who engage in similar high-risk activities. However, this relationship typically has been assessed in concentrated HIV epidemics,20,24,25 with fewer assessments in generalized epidemics.21 In contrast, we used social contact recruitment in a generalized epidemic among persons with biological evidence of risk—a newly diagnosed case of HIV and/or an STI. By using a well-designed community-based comparison group, we were able to demonstrate effectiveness. Even in the context of a generalized HIV epidemic, STI and HIV risk was not evenly distributed, but rather clustered in social networks.
Understanding the reasons for social contact recruitment effectiveness is important. One possible explanation is that members of the same social networks have similar risk behaviors. Formal exploration of this possibility is being assessed in a separate analysis. However, informal comparisons of sexual behavior between seeds and corresponding contacts suggest this explanation alone does not account for these results. An alternative explanation for the observed HIV disease clustering is that the social network itself is a risk factor. Contacts of HIV-infected seeds may be part of sexual networks with a higher HIV prevalence. In other words, the network population may be a more salient exposure than the behaviors within that network, an observation that has been made in concentrated epidemics25,26 and other generalized27,28 HIV epidemic settings. Sexual relationships between seeds and contacts are not the primary reason for the observed clustering, although a high prevalence of HIV concordance has been observed among couples in this setting.29 When analyses were restricted to only nonsexual contacts, elevated HIV prevalence persisted.
Social contact recruitment by patients with AHI may also be a promising way of effectively finding the “leading edge” of the HIV epidemic. In this study, we were only able to explore this possibility through enrollment of a few persons with AHI. On average, these patients were willing to recruit 2 social contacts and the HIV prevalence among their contacts was high (30%). However, we were not able to explore whether their contacts had AHI. Exploring AHI in social contacts of AHI patients is a key next step, as these persons may be exposed to networks with elevated HIV incidence.
Social contact recruitment was feasible in all groups but more feasible for community-based seeds. Clinic-based seeds recruited fewer social contacts. Lower recruitment may have been due to stigma or fear of contacts learning their STI or HIV results, factors under exploration in an analysis of acceptability. Despite clinic-based seeds recruiting fewer contacts, half of the seeds were successfully able to recruit at least 1 contact.
Lower feasibility coupled with greater effectiveness led to greater efficiency of social contact recruitment by clinic-based seeds. The total number of newly diagnosed contacts was approximately the same in all 3 groups. However, the number of contacts needed to test to identify 1 new case of HIV was considerably lower among contacts of clinic-based seeds. For routine implementation, screening fewer high-risk contacts is more efficient than more low-risk contacts. Efficiency may be improved further by targeting those seeds most likely to recruit undiagnosed HIV-infected persons. Such targeting could reduce the number of additional persons presenting to a busy clinical setting while simultaneously reaching those with greatest need. Demographic, behavioral, and relationship characteristics associated with recruitment of high-risk contacts is a direction to explore.
Several operational considerations deserve further research. First, in addition to receiving $5 per research visit, all seeds received a $2 incentive for each contact who presented to the clinic. This amount was considered motivational but not coercive by local staff and community advisors. However, because this amount did not vary, we could not assess whether a larger incentive would have improved contact recruitment. Additionally, all seeds were exposed to the same messages regarding contact recruitment. They were encouraged to bring friends who would benefit from the health promotion program. But other messages, such as encouraging recruitment of high-risk contacts, may be more effective. Future studies could randomize whether different incentive amounts and messages are associated with different degrees of feasibility, effectiveness, and efficiency.
Replication in other clinical settings is warranted. STI clinics serve patients with greater biological and behavioral risk for HIV, and these patients were part of social networks with elevated undiagnosed HIV infection. Whether newly diagnosed HIV-infected patients in other settings would also be part of higher risk networks is unknown. However, recruitment by only HIV-infected seeds may result in inadvertent disclosure of seed HIV status. Assessment in other settings, with attention to inadvertent disclosure, is an important direction for future research.
Our findings reflect a novel strategy for addressing a pressing public health need: identifying undiagnosed hard-to-reach cases of HIV infection. We demonstrated that asking STI patients to recruit their social contacts was a feasible, effective, and efficient way of identifying this population. These observations support social contact recruitment extending the reach of the health care screening system. Such an approach could become a powerful way of identifying HIV in hard-to-reach populations earlier.
The authors would like to thank the clinic and community staff and participants for their contributions.
1. Cohen MS, Chen YQ, McCauley M, et al.. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365:493–505.
2. Granich RM, Gilks CF, Dye C, et al.. Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model. Lancet. 2009;373:48–57.
3. WHO, UNAIDS, UNICEF. HIV Testing and Counseling. Towards Universal Access: Scaling up Priority HIV/AIDS Interventions in the Health Sector, Progress Report 2010 Geneva, Switzerland: World Health Organization; 2010.
4. Macro International. Malawi Demographic and Health Survey 2010. Zomba, Malawi and Calverton, Maryland: Macro International; 2011.
5. Marks G, Crepaz N, Senterfitt JW, et al.. Meta-analysis of high-risk sexual behavior in persons aware and unaware they are infected with HIV in the United States: implications for HIV prevention programs. J Acquir Immune Defic Syndr. 2005;39:446–453.
6. Powers KA, Ghani AC, Miller WC, et al.. The role of acute and early HIV infection in the spread of HIV and implications for transmission prevention strategies in Lilongwe, Malawi: a modelling study. Lancet. 2011;378:256–268.
7. Rosenberg NE, Pettifor AE, Bruyn GD, et al.. HIV testing and counseling leads to immediate Consistent condom use among South African stable HIV-discordant couples. J Acquir Immune Defic Syndr. 2013;62:226–233.
8. Roura M, Watson-Jones D, Kahawita TM, et al.. Provider-initiated testing and counselling programmes in sub-Saharan Africa: a systematic review of their operational implementation. AIDS. 2013;27:617–626.
9. Were W, Mermin J, Bunnell R, et al.. Home-based model for HIV voluntary counselling and testing. Lancet. 2003;361:1569.
10. Menzies N, Abang B, Wanyenze R, et al.. The costs and effectiveness of four HIV counseling and testing strategies in Uganda. AIDS. 2009;23:395–401.
11. Dalal W, Feikin DR, Amolloh M, et al.. Home-based HIV testing and counseling in rural and Urban Kenyan communities. J Acquir Immune Defic Syndr. 2013;62:e47–e54.
12. Tumwesigye E, Wana G, Kasasa S, et al.. High uptake of home-based, district-wide, HIV counseling and testing in Uganda. AIDS Patient Care STDs. 2010;24:735–741.
13. Were WA, Mermin JH, Wamai N, et al.. Undiagnosed HIV infection and couple HIV discordance among household members of HIV-infected people receiving antiretroviral therapy in Uganda. J Acquir Immune Defic Syndr. 2006;43:91–95.
14. Negin J, Wariero J, Mutuo P, et al.. Feasibility, acceptability and cost of home-based HIV testing in rural Kenya. Trop Med Int Health. 2009;14:849–855.
15. Armbruster B, Helleringer S, Kalilani-Phiri L, et al.. Exploring the relative costs of contact tracing for increasing HIV case finding in sub-Saharan countries. J Acquir Immune Defic Syndr. 2011;58:e29–e36.
16. Helleringer S, Mkandawire J, Reniers G, et al.. Should home-based HIV testing and counseling services be offered Periodically in programs of ARV treatment as prevention? A case study in Likoma (Malawi). AIDS Behav. 2012;17:2100–2108.
17. Kimbrough LW, Fisher HE, Jones KT, et al.. Accessing social networks with high rates of undiagnosed HIV infection: the social networks demonstration project. Am J Public Health. 2009;99:1093–1099.
18. McCoy SI, Shiu K, Martz TE, et al.. Improving the efficiency of HIV testing with peer recruitment, financial incentives, and the involvement of persons living with HIV infection. J Acquir Immune Defic Syndr. 2013;63:e56–e63.
19. Golden MR, Gift TL, Brewer DD, et al.. Peer referral for HIV case-finding among men who have sex with men. AIDS. 2006;20:1961–1968.
20. Malekinejad M, Johnston LG, Kendall C, et al.. Using respondent-driven sampling methodology for HIV biological and behavioral surveillance in international settings: a systematic review. AIDS Behav. 2008;12(suppl 4):S105–S130.
21. Townsend L, Zembe Y, Mathews C, et al.. Estimating HIV prevalence and HIV-related risk behaviors among heterosexual women who have multiple sex partners using respondent-driven sampling in a high risk community in South Africa. J Acquir Immune Defic Syndr. 2012;62:457–64.
22. Ssali S, Wagner G, Tumwine C, et al.. HIV clients as agents for prevention: a social network solution. AIDS Res Treat. 2012;2012:815823.
23. Latkin CA, Knowlton AR, Sherman S. Routes of drug administration, differential affiliation, and lifestyle stability among cocaine and opiate users: implications to HIV prevention. J Subst Abuse. 2001;13:89–102.
24. Powers KA, Miller WC, Pilcher CD, et al.. Improved detection of acute HIV-1 infection in sub-Saharan Africa: development of a risk score algorithm. AIDS. 2007;21:2237–2242.
25. Dennis AM, Murillo W, de Maria Hernandez F, et al.. Social network-based recruitment successfully reveals HIV-1 transmission networks among high-risk individuals in El Salvador. J Acquir Immune Defic Syndr. 2013;63:135–141.
26. Latkin C, Donnell D, Liu TY, et al.. The dynamic relationship between social norms and behaviors: the results of an HIV prevention network intervention for injection drug users. Addiction. 2013;108:934–943.
27. Millett GA, Peterson JL, Flores SA, et al.. Comparisons of disparities and risks of HIV infection in black and other men who have sex with men in Canada, UK, and USA: a meta-analysis. Lancet. 2012;380:341–348.
28. Helleringer S, Kohler HP. Sexual network structure and the spread of HIV in Africa: evidence from Likoma Island, Malawi. AIDS. 2007;21:2323–2332.
29. Brown LB, Miller WC, Kamanga G, et al.. HIV partner notification is effective and feasible in sub-Saharan Africa: opportunities for HIV treatment and prevention. J Acquir Immune Defic Syndr. 2011;56:437–442.
Keywords:© 2014 by Lippincott Williams & Wilkins
HIV; social network; sexually transmitted infection; Malawi; HIV counseling and testing; syndromic management