Investigating the Potential of Respondent-Driven Sampling to Reach Undiagnosed, Out-of-Care HIV-infected People Who Inject Drugs

Johns Hopkins University, Baltimore MD

Investigators

Linked publications & trials

Abstract

7. PROJECT SUMMARY/ABSTRACT Significance: A large proportion of HIV-infected people who inject drugs (PWID) are unaware of their positive status and many are not successfully linked to medical care. For the purposes of surveillance, respondent-driven sampling (RDS), a widely used type of chain-referral sampling that leverages peer network connections to recruit members of hidden populations, has been an efficient strategy for recruiting PWID. RDS may have the potential to become part of a strategy to effectively and efficiently reach those PWID that are not successfully engaged by other means. However, its potential in this regard has not been formally evaluated or compared to more traditional strategies. Specific Aims: Research on strategies to improve HIV testing and retention in care in order to reduce the incidence of HIV is a key priority for NIH-funded HIV research. This study will inform the utility and implementation of RDS as an intervention strategy to improve levels of diagnosis and linkage to medical care among HIV-infected PWID. Specifically, the study will (1) assess the ability of RDS to reach key sub-populations of PWID as compared to a venue-based strategy (integrated care centers [ICCs]) in India; (2) identify individual recruiter characteristics associated with identification of PWID via RDS who are HIV-infected and undiagnosed and diagnosed but out-of-care; and (3) evaluate whether the efficiency of identification of undiagnosed and out-of- care HIV-infected PWID can be enhanced through alterations to the RDS coupon system. Approach: Aim 1 will compare the number needed to recruit (NNR) for undiagnosed HIV-infected PWID between RDS and the ICCs. The NNR is defined as the average number of PWID who are recruited or reached in order to find one undiagnosed HIV-infected PWID. Multi-level regression models will also be used to identify socio- demographic and risk characteristics that are associated with being captured only within the RDS versus being an ICC client. Aim 2 will use prediction models and areas under the receiver operating curve (AUC) to explore individual recruiter characteristics that best predict identification of at least one HIV-infected PWID via RDS that is undiagnosed and diagnosed but out-of-care. Aim 3 will then evaluate alterations to the RDS coupon system in one PWID community in India by comparing the NNR for undiagnosed and out-of-care HIV-infected PWID between an altered recruitment chain and an unaltered recruitment chain in one community. Fellowship Information: This study is the dissertation for Ms. Allison McFall, a PhD student in the Department of Epidemiology at Johns Hopkins University. Ms. McFall has chosen one primary sponsor and one co-sponsor with complementary experience and expertise in the field of HIV research among substance-using populations. Her training includes mentored research, coursework, and opportunities to interact with other scientists.