Secondary Logo

Journal Logo


The Influence of Social Networks on Antiretroviral Therapy Initiation Among HIV-Infected Antiretroviral Therapy–Naive Youth in Rural Kenya and Uganda

Brown, Lillian B. MD, PhDa; Balzer, Laura B. PhDb; Kabami, Jane MPHc; Kwarisiima, Dalsone MBChB, MPHc; Sang, Norton MAd; Ayieko, James MBChB, PhDd; Chen, Yiqun BSe; Chamie, Gabriel MD, PhDa; Charlebois, Edwin D. PhD, MPHa; Camlin, Carol S. PhD, MPHa; Cohen, Craig R. MD, MPHa; Bukusi, Elizabeth MBChB, PhDd; Kamya, Moses R. MD, PhDf; Moody, James PhDg; Havlir, Diane V. MDa; Petersen, Maya L. MD, PhDh

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: January 1, 2020 - Volume 83 - Issue 1 - p 9-15
doi: 10.1097/QAI.0000000000002199



An estimated 2.2 million youth (aged 15–24 years) were living with HIV in sub-Saharan Africa (SSA) in 2015, and youth account for 37% of all new HIV infections in the region.1 The expansion of antiretroviral therapy (ART) across SSA has led to improved quality of life and greater life expectancy and has prevented new infections.2,3 However, youth are half as likely to initiate ART as older adults,4–6 placing them at risk of poor health outcomes and ongoing transmission.

High levels of stigma and lack of social support have been shown to contribute to this age disparity,7–9 suggesting that the social networks of youth may play an important role in determining whether and when they link to HIV care and initiate ART. Social networks, or the connections between and among people, impact health behavior and outcomes through the spread of ideas, social influence, and access to resources10 and may influence an individual's behavior above and beyond the influence of his/her own attributes. Social network analyses have demonstrated the importance of peer behavior to HIV risk behaviors in older adults in SSA, including HIV testing,11 partner concurrency,12 and condom use.13 In addition, data suggest that peer influences are important to HIV testing14 and linkage to care among adolescents.15 However, social network analyses of HIV+ youth in SSA are sparse, and peer effects on ART initiation in adolescents have yet to be demonstrated directly. Therefore, an improved understanding of how the social networks of youth contribute to their engagement in ART care could suggest effective interventions to increase the proportion of HIV-infected youth on ART.

The SEARCH HIV test-and-treat study (NCT01864683) collected social network data on 150,395 adults in 32 communities in rural Kenya and Uganda; of whom, 89% were tested for HIV. We examined whether baseline social network characteristics predicted ART initiation among ART-naive HIV-infected youth. We hypothesized that youth whose immediate social network included contacts who were also HIV infected would be more likely to initiate ART than those without HIV-infected contacts and that the association would be stronger for contacts who were ART experienced, of the same age, and who were named multiple times as sources of social support (strong ties).


Search Study

The overall objective of the SEARCH study was to evaluate the impact of “universal test and treat” on HIV incidence and health using a multidisease approach. The parent study enrolled 32 rural communities in western Kenya (12 communities, baseline HIV prevalence 19%), southwestern Uganda (10 communities, baseline prevalence 7%), and eastern Uganda (10 communities, baseline prevalence 4%).16 Each community was composed of geopolitical units just above village level (a “parish” in Uganda and a “sublocation” in Kenya) with a combined population of approximately 10,000 people; of whom, approximately 50% were ≥15 years of age. A door-to-door household census conducted at study baseline (June 2013–June 2014) collected demographic information from all residents, including age, sex, marital status, education level, income, and occupation. Census enumeration was followed by HIV and multidisease testing using a hybrid model that combined multidisease community health campaigns with home-based testing for nonattendees, and which reached 90% of the population.16,17 Residents found to be HIV infected were referred to their local health facility where they received ART based on country guidelines (16 control communities) or immediately (16 intervention communities).

Social Networks

A name generator adopted from earlier work by Perkins et al in rural Uganda18 was administered to adults aged ≥15 years during the hybrid testing. Residents were asked the name up to 6 contacts in each of 5 different social domains: with whom, over the past 12 months, they (1) shared food with outside of their household; (2) spent free time; (3) discussed money matters; (4) discussed health issues; and (5) went to for emotional support (see Appendix 1, Supplemental Digital Content, Residents could name contacts in more than one domain. An example of the question used to elicit contacts in the health domain is “Over the last 12 months, with whom have you discussed any kind of health issue?” Contact names, village, and age were collected by field staff on tablet computers. Using these raw data, sociocentric networks for each community were constructed by matching named contacts to census enumerated community residents; the matching algorithm, which weighted the different attributes (names, village, age) and determined a threshold for a match, is described in detail in the study by Chen et al.19 Among named contacts reported to live within the community, 85% were successfully matched in Southwest Uganda, 74.8% were matched in East Uganda, and 35.4% were matched in Kenya.19 The average degree of the networks were 5.8 in Southwest Uganda, 5.7 in East Uganda, and 1.6 in Kenya.19

Study Population

Youth who were (1) between the ages of 15–24 years, HIV infected, ART naive, and stable residents of a study community (in the community ≥6 of the preceding 12 months) during baseline testing and (2) still alive and resident in the community 3 years later were included in this analysis. Individuals were considered to be ART naive at baseline if there was no Ministry of Health HIV medical record indicating prior or current ART and they did not have an undetectable plasma HIV RNA level.


Patient demographics were obtained during the baseline census. Patients who self-reported no previous HIV diagnosis and did not have a Ministry of Health HIV medical record at the time of baseline testing were considered new HIV diagnoses. Point of care CD4+ cell count measurements (Pima; Alere, Waltham, MA) were performed at the time of baseline testing. Plasma HIV-1 RNA viral load was measured at the time of baseline testing from finger-prick capillary or venous blood collection by commercial real-time polymerase chain reaction assays at multiple reference laboratories. ART initiation between hybrid testing at baseline and 3 years later was determined through the Ministry of Health HIV medical records and HIV RNA level at year 3.

Contact demographics, HIV status, and ART status were determined identically (network construction allowed for linkage to study records for matched contacts). Participants were considered to have at least one HIV-infected contact if ≥1 matched contact was documented to be HIV infected through baseline testing. For this analysis, a contact was considered a strong tie if named in ≥1 social network domain by the participant.


Demographic variables at baseline were described overall and stratified on sex. Among participants with at least 1 HIV-infected contact, we evaluated the proportion who named at least 1 HIV-infected contact in the same age group (15–24 years) as the participant, the proportion who were ART experienced at baseline, and the proportion who initiated ART between baseline and follow-up year 3. We further evaluated tie strength (number of domains in which an HIV-infected contact was nominated) and sex concordance (same sex vs. opposite sex of participant) of HIV-infected contacts.

Logistic regression with robust standard errors (treating communities as independent) was used to evaluate the association between contact type and ART initiation before follow-up year 3. Sex, prior diagnosis, study arm, total number of contacts, and region were included as covariates in multivariate models a priori, based on their known relationship to ART initiation.


The Makerere University School of Medicine Research and Ethics Committee (Uganda), the Ugandan National Council on Science and Technology (Uganda), the Kenya Medical Research Institute Ethical Review Committee (Kenya), and the University of California San Francisco Committee on Human Research (USA) approved the study protocol including the consent procedures. All participants provided verbal informed consent in their preferred language with fingerprint biometric confirmation of agreement.


Characteristics of ART-Naïve Youth

Among 1120 HIV-infected youth who were ART naive at baseline, 805 remained alive and resident in the community after 3 years of follow-up (Fig. 1). These 805 HIV-infected ART-naive youth were approximately 80% women (646 of 805); 75% (600/805) were aged 20–24 years, 67% were married (535 of 805), and 67% (538 of 805) resided in 1 of the 12 communities in Kenya. The majority of youth in these rural communities were employed in farming (41%, 331 of 805), and 13% (107 of 805) were students at the time of the baseline survey. Baseline viral load was >10,000 copies per milliliter in 75% (604 of 805), and most (59%, 471 of 805) had a CD4+ count of >500 cells per cubic millimeter; 71% (569 of 805) were new diagnoses. Approximately 70% (569 of 805) initiated ART before follow-up year 3 (Table 1).

Study population of HIV+ youth resident in SEARCH communities who were ART naive at baseline and still alive and in the community at follow-up year 3.
Demographics of HIV-Infected Antiretroviral–Naive Youth in SEARCH Communities at Baseline Overall and by Sex (n = 805)

Overall, 81% (651 of 805) of youth named at least 1 contact; 75% (602 of 805) named at least 1 contact who was successfully matched to another SEARCH participant; 63% (507 of 805) named at least 1 contact who was successfully matched to another SEARCH participant outside the same household. One-third (270 of 805, 33.5%) matched at least 1 HIV-infected contact, and 18% (144 of 805) matched at least 1 HIV-infected nonhousehold contact (Table 1 and Fig. 1). Compared with HIV-infected youth who matched at least 1 contact, youth who did not match any contacts were slightly more likely to be 15–19 years old, single, and reside in Kenya (see Table 1, Supplemental Digital Content,

Characteristics of HIV+ First-Degree Network Contacts

Among the 270 ART-naive youth with at least 1 HIV-infected contact, 147 (54.4%) had at least 1 HIV-infected contact of the same sex and 154 (57%) had at least 1 HIV-infected contact of the opposite sex. ART-naive youth with at least 1 HIV-infected contact were more likely to have at least 1 older (≥25 years) HIV-infected contact than to have at least 1 same-age HIV-infected contact (80.0% vs. 27.8%). Among both men and women, most same age contacts were women; 66% (35 of 53) of women with a same-age HIV-infected contact had a contact who was a same-age HIV-infected woman, and 82% (18 of 22) of men with a same age HIV-infected contact had a contact who was a same-age HIV-infected woman. Less than half (119/270, 44%) had at least 1 HIV-infected contact who was on ART at baseline; 53.7% (145 of 270) had at least 1 HIV-infected contact who initiated ART between baseline and follow-up year 3 (Table 2).

Number and Proportion With First-Degree Network Contacts by Sex, Age, ART Status, and Strength of Tie Among ART-Naive Youth With ≥1 Matched HIV+ Contact (n = 270)

Among those with at least 1 HIV-infected contact, 43% (117) had at least 1 HIV-infected contact who was a strong tie. Of these, most had an opposite sex HIV-infected strong tie: 76% (79/104) of HIV-infected ART-naive women and 69% (9/13) of HIV-infected ART-naive men with an HIV-infected strong tie had an opposite sex HIV-infected strong tie (Table 2).

Characteristics of HIV+ Nonhousehold First-Degree Network Contacts

When household contacts were excluded, it was more common overall for ART-naive youth with at least 1 nonhousehold HIV+ contact (N = 144) to have an HIV+ contact of the same sex (117/144, 81%) than to have an HIV+ contact of the opposite sex (34/144, 24%). This pattern was consistent for all types of contacts across age groups, ART status, and strong ties (Table 3).

Number and Proportion With First-Degree Nonhousehold Network Contacts by Sex, Age, ART Status, and Strength of Tie Among ART-Naive Youth With ≥1 Matched Nonhousehold HIV+ Contact (n = 144)

Association Between Contact Type and ART Initiation

After adjusting for sex, region, new diagnosis, total number of contacts named, and study arm, ART-naive youth with at least one same-age HIV-infected contact were almost 3 times as likely to initiate ART within 3 years than those without a same-age HIV-infected contact (aOR, 2.95; 95% confidence interval (CI): 1.49 to 5.86); the association was stronger if the same age HIV-infected contact was a strong tie (aOR, 5.33; 95% CI: 3.34 to 8.52). ART-naive youth with a contact who initiated ART between baseline and follow-up year 3 were also twice as likely to themselves initiate ART, compared with youth without such a contact (aOR, 2.01; 95% CI: 1.17 to 3.45). Having an older age (≥25 years) HIV-infected contact or an HIV-infected contact who was on ART at the time of baseline testing was not associated with ART initiation before 3 years among ART-naive youth (Table 4).

Association Between Contact Type and ART Initiation Before FUY 3 Among HIV+ ART-Naive Youth (N = 805)

When household contacts were excluded, the magnitude of associations between contact characteristics and ART initiation was reduced. However, ART-naive youth with at least 1 same-age nonhousehold HIV-infected contact who was a strong tie remained significantly more likely to themselves initiate ART by 3 years (aOR, 2.81; 95% CI: 1.76 to 4.49) (Table 5).

Association Between Nonhousehold Contact Type and ART Initiation Before FUY 3 Among HIV+ ART-Naive Youth (N = 805)


HIV-infected youth aged 15–24 years in SSA have not benefited from the expansion of life-saving ART to the same degree as older adults, in part because they are less likely to start treatment after diagnosis than older adults.4–6 We examined the role that the immediate social network of HIV-infected ART-naive youth plays in predicting whether these youth initiate ART. We found that ART initiation was more likely among ART-naive youth who had at least 1 contact in their first-degree social network who was also HIV infected and in their same age group (15–24 years) and that this association was stronger if the HIV-infected contact was named in more than 1 domain of a youth's social network (ie, was a strong tie). Furthermore, ART initiation was also more likely among those who had an HIV-infected first-degree social network contact who also initiated ART, most of whom were same-sex contacts. These social relationships between HIV-infected persons may have important effects on clinical outcomes that can be leveraged for interventions.

The population of ART-naive youth in our study communities was mostly female, aged 20–24 years, and married, reflecting the demographics of the HIV epidemic in eastern Africa.20 Overall, most HIV-infected social connections of both men and women were with members of the opposite sex, and the majority of these connections were with members of the same household, many of them spouses. Disclosure to spouses and spousal support is associated with improved initiation, adherence, and retention to ART among pregnant and postpartum women,21 and men experience increased social support following disclosure that facilitates ART initiation and retention in care.22 Some of the protective association we observed between having an HIV-infected contact and ART initiation may have been mediated by spousal support, and efforts to promote such disclosure and support should continue.

Importantly, however, our results are also consistent with an even more important role played by peer support, including support of same age and same-sex HIV-infected peers. Among young women in particular, same age HIV-infected social contacts (the type of contact most strongly associated with ART initiation) were more likely to be of the same sex. Moreover, having a strong tie with a same-age nonhousehold HIV+ peer more than doubled the probability of initiating ART among youths. Efforts to promote support (and possibly disclosure) should extend to peers outside of the household.

Prior qualitative studies have found peer support to be a facilitator to adherence,23 and adolescent support groups during clinic hours have been associated with improved clinic attendance.24 Our results provide longitudinal quantitative evidence that peer support also plays an important role in initial linkage to care and ART initiation. Peers who provide support in more than one domain may play a particularly important role. Interestingly, network contacts who were ART-experienced or were HIV-infected older adults were not associated with ART initiation among youth. The age, sex, and ART status of peers should be considered when designing peer-based interventions for HIV-infected youth.

Our results also shed new light on the design of youth-specific peer-based interventions to support HIV care. Although some data suggest that trained peer navigators can support high rates of linkage to care following community-based testing in SSA,25 and WHO recommends that peer counseling and navigation be offered to all HIV-infected individuals following diagnosis to promote timely linkage to care,26 and current evidence supporting the effectiveness of such peer-based interventions on engagement in HIV care, particularly among youth, is mixed.27 When this approach was implemented in East Africa, overall linkage to HIV care and ART initiation among youth aged 15–24 years continued to lag behind adults.28 More effective peer navigation systems for youth are thus needed. Current peer navigation systems are imposed upon social networks; that is, peer supporters are assigned to patients. This model should be reexamined in light of our findings, which indicate the importance of endogenous support via existing relationships in facilitating care among youth populations. Following linkage to care and ART initiation, continued engagement in care is essential to realize the benefits of ART. Social support groups are associated with higher retention in care and adherence to ART.4,29 Our results also suggest that support groups composed of other, same-age, ART initiators may be the most effective to promote ART initiation and prevent early dropout of ART programs.

Several plausible mechanisms may explain our findings. The presence of other HIV-infected youth, in particular those who are also initiating ART, may promote self-efficacy for starting ART through vicarious efficacy, in line with precepts for the adoption of new behavior in Social Cognitive Theory.30 Through seeing others attending clinic and staying healthy on ART, these ART youth may expect that they too can successfully take ART. In addition, both social instrumental and emotional support may play a role in influencing the decision for ART-naive youth to start ART.31,32 Other HIV-infected youth may provide instrumental support through facilitating transportation to clinic, picking up medications, or sharing food that enables taking medication. They may provide emotional support that motivates youth to participate in ART programs. Future work will address these mechanisms through qualitative evaluation and measuring the resources youth have in their social network, including the emotional, cognitive, tangible, and physical support that they perceive themselves to receive from others. The association between social network attributes and ART initiation may also be to the result of other network mechanisms. Social networks have been associated with social-multiplier effects that reinforce HIV prevention messages in Kenya and Malawi,33 and similar amplifying effects may be operating in the networks of these HIV-infected youth. Understanding additional network effects beyond first-degree contacts, such as network density and diffusion, will also be important to understanding the mechanism of network influence, which will inform intervention design. For example, in Kenya, social learning was the dominant mechanism influencing contraception choice in highly dense social networks, whereas social influence was more important in more moderately dense networks.34 Future work will investigate these potential influences on ART initiation and engagement in ART care.

Our study had several limitations. The study population is limited to those individuals who are still alive and living in a study community after 3 years of follow-up, and thus, our results cannot be generalized to youth who migrate out to other communities. Compared with those who remained alive and in the study community after 3 years of follow-up, the 281 youth who outmigrated before follow-up year 3 were slightly less likely to name a matched contact (71% vs. 75%) and name at least 1 HIV+ contact (23.8% vs. 33.5%), suggesting that those who remain in the community may have stronger social ties. Both HIV status and network relations likely contribute to migration patterns,35 and network strength may play an important role among this selected population of youth than those who outmigrated. We were only able to incorporate information about matched contacts and may be missing contacts who reside outside the community, who were not matched, or who were not named by youth. Although links to HIV+ contacts are missing to the same extent among ART initiators and noninitiators, it is likely to bias our findings toward the null; however, if persons who initiated ART were less likely to have missed HIV+ contacts, then the magnitude of association may be overestimated. Although we have data on marital status, we do not have data on whether named contacts were spouses; however, nonhousehold same-age contacts were important determinants of ART initiation. The observed association between network characteristics and ART initiation may reflect homophily (ie, HIV-infected individuals who are likely to start ART are more likely to associate with each other). We controlled for some of these factors (sex, prior diagnosis, total number of contacts), but latent homophily may still be present. Network selection may also have played a role among youth previously diagnosed with HIV but not yet on ART (30% of the study population) if those who were better able to access care and had fewer barriers to ART initiation were also those who reached out and formed relationships with other HIV-infected individuals after their diagnosis. Finally, we do not know about the quality of the relationships or the content of either perceived or received support. However, identifying potentially important relationships through social network analysis is a first step toward understanding mechanisms of support.

HIV-infected youth are at risk of being left behind in the Global AIDS response. Leveraging peer relationships is an opportunity to mitigate the negative effects of stigma and poor social support that are cited as important barriers to effective engagement in HIV care in among youth.7–9 Understanding the influence of social networks on youth engagement in HIV care, including network type, network structure, and provision of support within the network, will suggest innovative network-based interventions to improve clinical outcomes in this vulnerable population.


1. UNAIDS. Global Report: UNAIDS Report on the Global AIDS Epidemic 2015. Geneva, Switzerland: Joint United Nations Programme on HIV/AIDS; 2016.
2. 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.
3. Wandeler G, Johnson LF, Egger M. Trends in life expectancy of HIV-positive adults on antiretroviral therapy across the globe: comparisons with general population. Curr Opin HIV AIDS. 2016;11:492–500.
4. Lamb MR, Fayorsey R, Nuwagaba-Biribonwoha H, et al. High attrition before and after ART initiation among youth (15–24 years of age) enrolled in HIV care. AIDS. 2014;28:559–568.
5. Nachega JB, Hislop M, Nguyen H, et al. Antiretroviral therapy adherence, virologic and immunologic outcomes in adolescents compared with adults in southern Africa. J Acquir Immune Defic Syndr. 2009;51:65–71.
6. Evans D, Menezes C, Mahomed K, et al. Treatment outcomes of HIV-infected adolescents attending public-sector HIV clinics across Gauteng and Mpumalanga, South Africa. AIDS Res Hum Retroviruses. 2013;29:892–900.
7. Cluver LD, Orkin FM, Boyes ME, et al. Cash plus care: social protection cumulatively mitigates HIV-risk behaviour among adolescents in South Africa. AIDS. 2014;28(suppl 3):S389–S397.
8. Cluver LD, Toska E, Orkin FM, et al. Achieving equity in HIV-treatment outcomes: can social protection improve adolescent ART-adherence in South Africa? AIDS Care. 2016;28(suppl 2):73–82.
9. Wolf HT, Halpern-Felsher BL, Bukusi EA, et al. “It is all about the fear of being discriminated [against]...the person suffering from HIV will not be accepted”: a qualitative study exploring the reasons for loss to follow-up among HIV-positive youth in Kisumu, Kenya. BMC Public Health. 2014;14:1154.
10. Christakis NA, Fowler JH. Social network visualization in epidemiology. Nor Epidemiol. 2009;19:5–16.
11. Yamanis TJ, Dervisevic E, Mulawa M, et al. Social network influence on HIV testing among urban men in Tanzania. AIDS Behav. 2017;21:1171–1182.
12. Yamanis TJ, Fisher JC, Moody JW, et al. Young men's social network characteristics and associations with sexual partnership concurrency in Tanzania. AIDS Behav. 2016;20:1244–1255.
13. Mulawa M, Yamanis TJ, Hill LM, et al. Evidence of social network influence on multiple HIV risk behaviors and normative beliefs among young Tanzanian men. Soc Sci Med. 2016;153:35–43.
14. Neary J, Wagner AD, Mugo C, et al. Influence and involvement of support people in adolescent and young adult HIV testing. AIDS Care. 2019;31:105–112.
15. Ruria EC, Masaba R, Kose J, et al. Optimizing linkage to care and initiation and retention on treatment of adolescents with newly diagnosed HIV infection. AIDS 2017;31(suppl 3):S253–s260.
16. Havlir DV, Balzer LB, Charlebois ED, et al. HIV testing and treatment with the use of a community health approach in rural Africa. N Engl J Med. 2019;381:219–229.
17. Chamie G, Clark TD, Kabami J, et al. A hybrid mobile approach for population-wide HIV testing in rural east Africa: an observational study. Lancet HIV. 2016;3:e111–e119.
18. Perkins JM, Nyakato VN, Kakuhikire B, et al. Actual versus perceived HIV testing norms, and personal HIV testing uptake: a cross-sectional, population-based study in rural Uganda. AIDS Behav. 2018;22:616–628.
19. Chen Y, Zheng W, Brown LB, et al. Semi-Supervised Record Linkage for Construction of Large-Scale Sociocentric Networks in Resource-Limited Settings: An Application to the SEARCH Study in Rural Uganda and Kenya. 2019. Available at: Accessed August 29, 2019.
20. UNAIDS. Global AIDS Update 2018: Miles to Go. Geneva, Switzerland: Joint United Nations Programme on HIV/AIDS; 2018.
21. Hodgson I, Plummer ML, Konopka SN, et al. A Systematic Review of Individual and Contextual Factors Affecting ART Initiation, Adherence, and Retention for HIV-Infected Pregnant and Postpartum Women. PLoS One. 2014;9:e111421.
22. Brown LB, Getahun M, Ayieko J, et al. Factors predictive of successful retention in care among HIV-infected men in a universal test-and-treat setting in Uganda and Kenya: a mixed methods analysis. PLoS One. 2019;14:e0210126.
23. Ammon N, Mason S, Corkery JM. Factors impacting antiretroviral therapy adherence among human immunodeficiency virus-positive adolescents in Sub-Saharan Africa: a systematic review. Public Health. 2018;157:20–31.
24. Barker D, Enimil A, Galarraga O, et al. In-clinic adolescent peer group support for engagement in sub-saharan Africa: a feasibility and acceptability trial. J Int Assoc Provid AIDS Care. 2019;18:2325958219835786.
25. Hatcher AM, Turan JM, Leslie HH, et al. Predictors of linkage to care following community-based HIV counseling and testing in rural Kenya. AIDS Behav. 2012;16:1295–1307.
26. WHO. Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV infection. Recommendations for a Public Health Approach. Geneva, Switzerland: World Health Organization; 2016.
27. Genberg BL, Shangani S, Sabatino K, et al. Improving engagement in the HIV care cascade: a systematic review of interventions involving people living with HIV/AIDS as peers. AIDS Behav. 2016;20:2452–2463.
28. MacKellar D, Maruyama H, Rwabiyago OE, et al. Implementing the package of CDC and WHO recommended linkage services: methods, outcomes, and costs of the Bukoba Tanzania Combination Prevention Evaluation peer-delivered, linkage case management program, 2014-2017. PLoS One. 2018;13:e0208919.
29. Bateganya MH, Amanyeiwe U, Roxo U, et al. Impact of support groups for people living with HIV on clinical outcomes: a systematic review of the literature. J Acquir Immune Defic Syndr. 2015;68(suppl 3):S368–S374.
30. Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev. 1977;84:191–215.
31. Barrera M Jr. Distinctions between social support concepts, measures, and models. Am J Community Psychol. 1986;14:413–445.
32. Holt-Lunstad JU, BN. Social support and health. In: Karen Glanz BKR, Viswanath V, eds. Health Behavior: Theory, Research, and Practice. San Francisco, CA: Jossey-Bass; 2015:183–204.
33. Kohler HP, Behrman JR, Watkins SC. Social networks and HIV/AIDs risk perceptions. Demography. 2007;44:1–33.
34. Kohler HP, Behrman JR, Watkins SC. The density of social networks and fertility decisions: evidence from South Nyanza district, Kenya. Demography. 2001;38:43–58.
35. Anglewicz P, VanLandingham M, Manda-Taylor L, et al. Migration and HIV infection in Malawi. AIDS. 2016;30:2099–2105.

HIV/AIDS; sub-Saharan Africa; social networks; youth; antiretroviral therapy

Supplemental Digital Content

Copyright © 2019 The Author(s). Published by Wolters Kluwer Health, Inc.