Attributes Associated with HIV Risk Behaviors in a Network-Based Study of People Who Inject Drugs.

INTRODUCTION

People who inject drugs (PWID) are often part of sexual and drug use networks. Engaging in unprotected sex or sharing drug injection equipment, which could occur between connections (ties) in these networks, is known to increase HIV transmission risk. This study aimed to identify attributes associated with network connections between PWID and their contacts during an HIV outbreak in Athens, Greece (2013-2015).

METHODS

Data from the Transmission Reduction Intervention Project (TRIP) were used. TRIP was a network-based intervention using information on recent HIV acquisition to reduce onward viral transmission among PWID. Descriptive network and individual-level statistics were calcu-lated. Exponential random graph models (ERGMs) were used to assess associations between in-dividual-level attributes and the likelihood of people having a risk tie (for instance, using drugs together) among PWID.

RESULTS

The network consisted of 356 participants (i.e., known as nodes in network terminology) and 542 connections (i.e., known as edges). TRIP participants had a mean age of 36 (8) years, and most were males (79%). A substantial proportion of participants were homeless (23%). Each participant was connected on average with three others (i.e., known as degree). Degree assorta-tivity was positive, indicating that participants were more likely to connect with people of similar degree. ERGMs showed that TRIP participants were more likely to have a connection with others like them in possibly important ways (e.g., were of the same sex and nationality, and had similar living conditions, such as being homeless).

DISCUSSION

This network analysis, including the use of ERGMs, indicated that individuals in networks of PWID tend to form ties with similar others, beyond what would be expected by chance alone. Limitations of the analysis include potentially reduced generalizability and repre-sentativeness due to the local context of the study, and diminished statistical power. Future re-search should prioritize longitudinal studies among PWID to examine how network connections evolve over time.

CONCLUSION

The analysis identified factors, such as housing instability, that are important in de-termining the observed network ties among PWID in Athens, Greece. Future development of interventions should consider these factors.