Project Engage: Snowball Sampling and Direct Recruitment to Identify and Link Hard-to-Reach HIV-Infected Persons Who Are Out of Care : JAIDS Journal of Acquired Immune Deficiency Syndromes

Secondary Logo

Journal Logo

Implementation Science

Project Engage: Snowball Sampling and Direct Recruitment to Identify and Link Hard-to-Reach HIV-Infected Persons Who Are Out of Care

Wohl, Amy Rock MPH, PhD*; Ludwig-Barron, Natasha MPH*; Dierst-Davies, Rhodri MPH, PhD*; Kulkarni, Sonali MD, MPH*; Bendetson, Jesse BA*; Jordan, Wilbert MD; Bolan, Robert MD; Smith, Terry MPA§; Cunningham, William MD; Pérez, Mario J. MPH*

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: June 1, 2017 - Volume 75 - Issue 2 - p 190-197
doi: 10.1097/QAI.0000000000001312
  • Free



Timely linkage to HIV care and ongoing engagement with HIV medical providers is a key tenet of the “Treatment as Prevention” (TasP) strategy to promote widespread antiretroviral treatment (ART) use, viral suppression, and reduced forward HIV transmission.1–4 Beginning in 2012, the HIV prevention and care community began the widespread adoption of the TasP approach, an innovative strategy needed to address the gaps highlighted in national and local HIV treatment cascades.5,6 In Los Angeles County (LAC) in 2012, as in other large urban jurisdictions in the United States, only 50% of the total persons estimated to be living with HIV were retained in care (defined as at least 2 HIV-related laboratory tests 90 days apart) and 48% had a suppressed viral load.6 Project Engage was developed as an innovative method to try to locate and link the “hidden” population of HIV-infected persons who had limited contact with both the HIV care delivery system and the nonmedical agencies that provide ancillary support services.

Numerous researchers have demonstrated success using snowball sampling or chain-referral methods to identify hard-to-reach populations including men who have sex with men (MSM) in Brazil, injection drug users (IDUs), and persons with undiagnosed HIV infection.7–10 In addition, the Centers for Disease Control and Prevention–funded National HIV Behavioral Surveillance project uses a similar method, respondent-driven sampling, to recruit and enroll high-risk heterosexuals and IDUs for HIV serologic testing and a risk behavior survey.11 Both snowball sampling and respondent-driven sampling identify target populations through referrals or “alters” from “seeds” or participants with social networks that are likely to include the target population who are not easily accessible to outsiders.

In this study, snowball sampling methods were adapted in combination with a direct recruitment (DR) approach to determine whether hard-to-find, marginalized HIV-positive persons who were not in consistent HIV care could be identified and subsequently linked to HIV care. In the DR approach, trained study staff distributed study promotional materials in public areas and agencies, described the project to individuals and groups, and screened interested participants.

Interventions that can successfully identify and link out of care (OOC) HIV-positive persons have particular public health importance, as it is estimated that 61% of new HIV infections in the United States can be attributed to transmission from persons who are not adequately retained in care.12 This 2-pronged recruitment strategy was combined using LAC HIV surveillance data to determine whether potential Project Engage participants were truly OOC and eligible for the linkage intervention. The methods and results presented here may be useful to other health departments and agencies as an approach to finding and linking “hidden” HIV-infected persons who are OOC, as there are limited published and evaluated programs that have successfully located and linked this population to critical HIV care.13,14


Snowball Sampling and Eligibility Criteria

As part of traditional snowball sampling methods or what is termed here as social network recruitment (SNR), Project Engage staff met with clinic and nonclinic agency (eg, substance abuse treatment facilities and homeless shelters) staff to identify “seeds” or patients/clients who were likely to know and recruit HIV-positive persons who were OOC. The HIV-positive persons who were OOC who were referred to the program by the seeds are called “alters.”15 A screening form was developed to help agency staff identify persons who had large social networks and were community gatekeepers or opinion leaders within their respective communities. Potential seeds were asked the following questions: (1) if they knew anyone in their social network who was HIV positive and OOC, (2) the number of people who they knew who were OOC, and (3) whether they felt comfortable talking to their friends or potential alters about their HIV status and their engagement in ongoing HIV care. An OOC alter was defined as an HIV-positive person who (1) had no HIV laboratory tests reported in the LAC HIV surveillance system for more than 12 months; (2) had no HIV laboratory tests reported in the HIV surveillance system for 6–12 months and the last viral load test was unsuppressed (>200 copies/mL); (3) was recently released from a jail, residential treatment facility, or other institution and had no regular HIV care provider; (4) had fewer than 2 medical visits in a 12-month period at the same HIV care provider; or (5) was recently diagnosed with HIV with no HIV medical visits within 3 months of diagnosis.

Eligible seeds were given 10 vouchers that included an identification number that linked “alters” (or persons referred by the seed) back to the original seed. Seeds and alters were reimbursed $40 when they completed the project questionnaire. In addition, seeds were reimbursed an additional $40 when a referred alter completed their first medical visit and alters were reimbursed $40 when they completed a first medical care visit. Enrolled alters could in turn become recruiters (ie, “seeds”) and bring in eligible alters from their own social networks who were then reimbursed similarly to the method described above. Alters could only become recruiters after they had completed their first HIV medical care visit.

Direct Recruitment

The DR arm included deployment of trained study staff to HIV clinics, social service agencies, substance use treatment facilities, needle exchange programs, and homeless shelters to recruit both OOC alters and seeds into the program. Study staff also directly recruited seeds and alters in public areas including parks, street corners, freeway underpasses, and other public areas where OOC HIV-positive persons were known to congregate. Passive recruitment included posting of project fliers and project pocket cards at agencies, clinics, and public areas. The project pocket cards were also used for DR of participants in the field at parks and other outdoor areas. The reimbursement structure for DR participants was the same as that described above for the SNR participants.


Once an eligible “alter” was identified using SNR or DR methods, a trained staff member met with the patient to assess his/her needs. The intervention was primarily designed to link an OOC HIV-positive client to HIV care at one of the approximately 35 LAC Ryan White–funded HIV clinic sites with Medical Care Coordination (MCC) teams that were being established during the study period. The clinics with MCC teams had the necessary resources and support to retain high acuity patients in consistent HIV care and were staffed by a nurse, a case manager, and a social worker.16 The MCC team coordinates with the HIV clinic medical providers and provides patient referrals for substance abuse treatment, mental health services, housing, transportation, and other nonmedical service needs.

If an eligible participant was not ready or not interested in linking to HIV care at the time of the initial contact, the Project Engage staff person would provide a standard educational intervention that included a discussion of the importance of HIV care, obtain contact information, and ask permission to continue to follow-up in case a person became ready to engage in HIV care at a later date. For participants who were interested in enrolling in the Project Engage linkage to HIV care program, a staff person provided any of the following linkage to care activities: shared a map and list of HIV care services for a client to select; scheduled an HIV medical care appointment; made reminder calls or sent reminder text messages about the clinic and/or intervention visit; provided transportation vouchers to get to the HIV clinic; accompanied the client to a medical visit; and helped the client navigate the HIV clinic system including the financial screening process. Linkage to care was defined as a medical visit with an HIV provider that included a viral load test and the potential for an ART prescription.

Data Collection

A baseline survey was administered to all seeds and alters to collect information on demographics, housing, insurance, incarceration history, HIV testing history, health status, stigma, stress, sexual risk behaviors, HIV disclosure, substance use, HIV medical and nonmedical service needs, barriers to services, attitudes toward medical care, ART adherence, readiness to engage in care, and transportation costs. All study materials were administered in both English and Spanish, and the project received institutional review board approval from all the participating institutions. The baseline descriptive analyses presented are for the 112 enrolled OOC alter participants. The analyses that involve comparisons between viral load at the time of linkage versus retention are based on 71 alters who had completed at least 6 months of follow-up. Persons who died (n = 1) and/or were incarcerated (n = 10) were considered lost to follow-up and were not included in the retention analyses.

Data Analysis

Univariate analyses are presented to describe and compare the sociodemographic characteristics of the seeds and alters using chi-square analyses and Fisher exact tests. Descriptive information is also presented on baseline HIV testing and care history, clinical characteristics, self-reported service needs, and the most important barrier to HIV care for the OOC alters. Demographics, baseline testing/care history, and process indicators for the alters who were recruited through DR vs. SNR are compared using chi-square analyses and Fisher exact tests. The percentage of alters who were linked within 3, 4–6, 7–12, and more than 12 months are described to assess the timeliness of linkage to care. The percentage who were retained in care, defined as a second viral load between 6 and 12 months after an initial HIV medical visit, is described to determine whether the OOC alters remained in HIV care between 6 and 12 months after linkage to care by study staff. In addition, the mean and median viral load at the time of linkage to care was compared with that at 6–12 months of follow-up among alters who were retained in care.


As shown in Figure 1, a total of 112 OOC alters were recruited, including 74 (67%) by the SNR seeds and 38 (33%) by DR. Most of the SNR seeds recruited 1–4 OOC alters, and 1 SNR seed recruited 59 eligible OOC alters. Among the total 120 seeds who were screened, 62 (52%) were deemed eligible, and among the 335 alters that were screened, 112 (33%) were eligible.

HIV-positive OOC alter participants (n = 112) recruited by Social Network Recruitment (SNR) and DR methods.

As shown in Table 1, 61% of the alters were aged 30–49, 80% were male, 14% were male to female transgender persons, 38% were African American, 22% were Latino, 60% were gay, 23% were bisexual, 85% were US-born, 60% were publicly insured, 89% had annual incomes less than $12,000, 63% were disabled, on Supplemental Security Income, or on public assistance, and 69% were high school graduates. In addition, 78% reported being homeless and 32% had engaged in sex work in the previous 6 months; 50% had been incarcerated in the last 12 months, 74% had used illicit drugs in their lifetimes, 24% had injected drugs in the last 3 months, and 34% were MSM and who also used injection drugs.

Baseline Characteristics for Project Engage Seeds and OOC Alters

Also shown in Table 1, compared with the seed participants, the alters were more likely to be transgender, less likely to self-report as gay, more likely to self-report as bisexual, less likely to have private insurance, more likely to have an annual income less than $12,000, less likely to be employed, more likely to be on disability, SSI, or public assistance, less likely to have any education after high school, more likely to have been homeless or engaged in sex work in the last 6 months, more likely to have been incarcerated in the previous 12 months, more likely to have injected drugs both in the last 3 months and over a lifetime, more likely to have ever used noninjection drugs, and more likely to be an MSM who injects drugs.

As shown in Table 2 that includes baseline participant clinical characteristics for the 112 alters, 38% had no HIV medical visits in more than 12 months before enrollment in the project, 11% had no HIV medical visit for 7–12 months and their most recent viral load was unsuppressed, 14% were newly diagnosed with no medical visit within 3 months of their diagnosis, and 47% had been recently released from jail with no regular HIV care provider. In addition, 9% had received HIV care only in a jail or prison setting and 24% had only received HIV care in a jail or prison setting in the 5 years before enrollment. Based on HIV surveillance data reported before alter enrollment, 35% were virally suppressed at their last recorded viral load measure with a mean and median viral load of 61,075 and 3155 copies/mL, respectively. In addition, 68% had been on ART at some point in the past. On average, almost 12 years had elapsed since the alters were first diagnosed with HIV.

Baseline HIV Care History and Clinical Characteristics for OOC Alters (n = 112)

Although not shown, HIV-related medical care was the service need reported by the largest proportion of alters (96%), followed by HIV case management (91%), general non-HIV medical services (89%), and mental health counseling (81%). The most commonly reported primary barriers to obtaining HIV medical care among the alters were challenges navigating the medical system (26%), substance use (14%), and incarceration (10%).

Demographics, testing and care history, and linkage activities are presented and compared in Table 3 between alters recruited using the DR versus the SNR approach. Compared with alters recruited using DR, SNR alters were more likely to be African American, uninsured, unemployed, homeless, a sex worker in the previous 6 months, engage in non-injection drug use (lifetime), incarcerated in the last 12 months, injected drugs in the last 3 months and over a lifetime, and more likely to be both an MSM and an IDU. Alters enrolled using the SNR approach also reported a greater number of unmet service needs compared with DR alters (P ≤ 0.05). In addition, although more staff hours were required to link the DR versus the SNR participants to care, there was no statistical difference in the number of days elapsed for DR and SNR participants to link to care.

Demographics, Social Determinants, and Care History for Directly Recruited Versus Social Network Recruited OOC Alters (n = 112)
Demographics, Social Determinants, and Care History for Directly Recruited Versus Social Network Recruited OOC Alters (n = 112)

Table 4 includes the main linkage and retention outcome measures for the 112 alters. As shown, after participation in the intervention, approximately 69% of alters linked to care within 3 months, 5% within 4–6 months, 8% between 7 and 12 months, and 4% in more than 12 months. Moreover, 15% (n = 17) of alters were lost to follow-up, including 10 participants who were incarcerated within 6 months of project enrollment and 1 participant who died. In addition, 72% of alters were retained in care defined as a second viral load measure between 6 and 12 months after project-facilitated linkage to care.

Linkage to Care, Retention in Care, and Viral Suppression for OOC Alters (n = 112)

The percent of alters who were virally suppressed based on their viral load measure at the time of linkage to care was 27% compared with 41% of alters who were virally suppressed at their second viral load measurement 6–12 months after study linkage (P = 0.04). There was no difference in the mean viral load for alters at the time of linkage versus retention (P = 0.6), although there was a significant decrease in the median viral load (P = 0.003).

Although not presented in tabular form, the alters were also asked about the acceptability and satisfaction with the Project Engage program. Among the 75 alters who completed the survey, 100% said they would recommend the program to others, 81% said that they would not have entered HIV care without the efforts of the program, and 100% said they were satisfied with the services that they received from Project Engage staff.


The goal of Project Engage was to implement and compare two novel methods to identify hard-to-find marginalized HIV-infected persons who were not in regular HIV care and link them to care. The results demonstrate that the mixed methodology of SNR and DR approaches resulted in the successful location of a “hidden” HIV-positive population with limited access to the HIV medical care system. In fact, the alters were more socially disadvantaged than the seeds who recruited them, affirming the notion that the identified seeds had unique access to a population that is likely not reachable using standard recruitment techniques. In addition, based on the acceptability survey, a large proportion (81%) of the enrolled alters indicated that they would not have accessed HIV care without Project Engage. The SNR method in particular achieved its a priori objective which was to locate the most marginalized individuals by incentivizing members of a social network to refer OOC HIV-positive persons to our HIV care linkage program.9 To our knowledge, this is the first published data on the use of snowball sampling for identifying OOC HIV-infected persons, although a similar method has been used to identify HIV-positive persons who are unaware of their HIV infection.7 Project Engage also demonstrated the feasibility of DR for hidden populations using agency and gatekeeper contacts to identify successful recruitment locations. One of the lessons learned for this type of project is that it is critical to hire program staff who are trusted members of the community with connections to shelters, substance use treatment centers, and other agencies that provide non-medical services to the target population.

Although the 69% of Project Engage participants who linked to HIV medical care within 3 months was lower than the original National HIV/AIDS Strategy (NHAS) linkage goal of 85%, the severely marginalized characteristics of our study population suggest linkage and retention challenges that are far more serious than those of the general HIV-positive population targeted for the first NHAS.1 These data suggest that NHAS goals may be challenging to achieve for the most severely affected subgroups even with the dedication of linkage and retention staffing resources.

There was a statistically significant increase in the proportion of alters who were virally suppressed at 6–12 months after linkage, a significant decrease in the median viral load, and a nonsignificant although favorable decrease in the mean viral load, suggesting improvement with adherence to care and treatment for this marginalized population due in large part to the dedicated MCC staff retention efforts. The 0.2 log10 decrease in mean viral load between linkage and retention in this primarily gay and bisexual study group is most similar to a 20% reduction in HIV transmission that was associated with a 0.3 log10 decrease in HIV RNA among a heterosexual study cohort.17 These data suggest decreases in HIV transmission risk after linkage and retention in care for the Project Engage participants. In addition, because 68% of the study group had previously been on ART, it is likely that reductions in viral suppression after a new ART prescription will be slower because of the increased risk of drug resistance.

These findings highlight the utility of dedicating sufficient agency resources to manage both the medical and non-medical needs of HIV-positive patients with complex psychosocial issues. Follow-up at 18–24 months for the OOC alters will be important to determine, if sufficient clinic-based support was effective at long-term care retention and viral load suppression for the Project Engage alters. It should be noted that 10 of the alters were incarcerated during program participation and that 24% of the study group had received HIV care only when in a jail setting in the last 5 years, underscoring the importance of transitional case management for recently incarcerated persons to support ongoing HIV care retention.

Other important findings from this research include the value of the use of HIV surveillance data to monitor viral loads, linkage and retention in HIV care, and confirm OOC program eligibility criteria. Historically, HIV surveillance data have not been used in this manner, and this project demonstrates the value in using HIV surveillance data to promote the identification, linkage, and tracking of hard-to-reach HIV-positive persons who are OOC.18

It should be noted that although the SNR method was effective at locating HIV-positive persons who were OOC, the networks of the seeds were fairly shallow with most persons referring only 1–4 members of the target population. These data underscore that this type of recruitment is time intensive and that any similar activities should plan for a sufficient investment of resources along with the anticipated slow enrollment. Project Engage was staffed with 1 part-time project coordinator and 2 full-time outreach staff over the study period and the addition of more staff would likely result in recruitment of additional marginalized HIV-positive OOC persons.

There are several limitations or lessons learned from the program and analysis presented here. First, there was 1 “super seed” recruited in the SNR arm who recruited (n = 59) most of the alters, whereas the remainder of the seeds had shallow recruitment networks. Thus, without the 1 “super seed,” the program would likely not have been as effective. In addition, previous knowledge and shared knowledge by the “super seed” and project staff regarding the location of outdoor and public areas where OOC HIV-positive persons congregate was critical to successful project recruitment. Also, because of the recruitment methods, the study sample is a convenience sample recruited from social networks and specific agency and public locations and may not be representative of all OOC persons in LAC.

In conclusion, Project Engage demonstrated the feasibility and efficacy of using a mixed methodology of SNR and DR to identify a hidden and marginalized population of HIV-positive persons not in regular HIV care. In particular, the program structure demonstrated effective linkage to HIV medical care clinics. It is critical however that robust HIV clinic support services be in place to sustain long-term patient retention in care, treatment adherence, and sustained viral load suppression for HIV-positive marginalized populations to ensure ongoing management of multiple comorbidities. Each of these are necessary steps toward support of the health and well-being of hard-to-reach OOC HIV-positive persons and will likely facilitate reductions in new infections.


The authors thank all the study participants and agency staff for their participation in the project. In addition, the authors thank Alla Kalenich, Christopher Moore and Susanna Moreno for their contributions to the project.


1. White House. National HIV/AIDS strategy for the United States;2010:1–60. Available at: Accessed July 15, 2013.
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. Das M, Chu PL, Santos GM, et al. Decreases in community viral load are accompanied by reductions in new HIV infections in San Francisco. PLoS One. 2010;5:e11068.
4. Hoffmann CJ, Gallant JE. Rationale and evidence for human immunodeficiency virus treatment as prevention at the individual and population levels. Infect Dis Clin North Am. 2014;28:549–561.
5. Maulsby C; The Positive Charge Intervention Team; Charles V, Kinsky S, et al. Positive charge: filling the gaps in the US HIV continuum of care. AIDS Behav. 2015;19:2097–2107.
6. Division of HIV and STD Programs (DHSP), Los Angeles County Department of Public Health. 2014 Annual HIV/STD Surveillance Report. 2016:1–66. Available at: Accessed January 7, 2016.
7. Gwadz M, Cleland CN, Leonard NR, et al. Hybrid STTR intervention for heterosexuals using anonymous HIV testing and confidential linkage to care: a single arm exploratory trial using respondent-driven sampling. BMC Public Health. 2015;15:1133.
8. Heckathorn DD. Respondent driven sampling: a new approach to the sampling of hidden populations. Social Probl. 1997;44:174–199.
9. Heckathorn DD. Extensions of respondent-driven sampling: analyzing continuous variables and controlling for differential recruitment. Sociological Methodol. 2007;37:151–208.
10. Kendall C, RFS Kerr L, Gondim RC, et al. An empirical comparison of respondent-driven sampling, time location sampling and snowball sampling for behavioral surveillance in men who have sex with men, Fortaleza, Brazil. AIDS Behav. 2008;12:S97–S104.
11. Lansky A, Abdul-Quader AS, Cribbin M, et al. Developing an HIV behavioral surveillance system for injecting drug users: the national HIV behavioral surveillance system. Public Health Rep. 2007;122:48–55.
12. Skarbinski J, Rosenberg E, Paz-Bailey G, et al. Human immunodeficiency virus transmission at each step of the care continuum in the United States. JAMA. 2015;175:588–596.
13. Brennan A, Browne JP, Horgam M. A systematic review of health service interventions to improve linkage with or retention in HIV care. AIDS Care. 2014;26:804–812.
14. Okeke NK, Ostermann J, Thielman NM. Enhancing linkage and retention in HIV Care: a review of interventions for highly resourced and resource-poor settings. Curr HIV/AIDS Rep. 2014;11:376–392.
15. Biernacki P, Waldorf D. Snowball sampling: problems and techniques of chain referral sampling. Sociol Methods Res. 1981;10:141–163.
16. Centers for Disease Control and Prevention, Program Evaluation Branch, Division of HIV/AIDS Prevention. Program Monitoring Update: Progress towards National HIV/AIDS Strategy Goals: Success Stories from the Enhanced Comprehensive HIV Prevention Planning Project. Atlanta, GA: CDC, 2015
17. Modjarrad K, Chamot E, Vermund SH. Impact of small reductions in plasma HIV RNA levels on the risk of heterosexual transmission and Disease progression. AIDS. 2008;22:2179–2109.
18. Wohl AR, Dierst-Davies R, Victoroff A, et al. The navigation program: an intervention to Re-engage lost patients at 7 HIV clinics in Los Angeles county, 2012–2014. J Acquir Immune Defic Syndr. 2016;71:e44–e50.

linkage; re-engagement; retention; marginalized populations; viral suppression

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.