Effects of a Three-Group Social Network Intervention on Sexual Health Conversations Among Young Adults Experiencing Homelessness

Background and Purpose: Young adults experiencing homelessness are a marginalized population who are at heightened risk for contracting HIV and other sexually transmitted infections. One promising approach for reducing the incidence of HIV involves enlisting peer change agents (PCAs) to have targeted conversations about HIV prevention behaviors with their peers, with the goal of disseminating risk reduction messages throughout social networks of youth experiencing homelessness. The current study uses data from a recent three-group social network intervention to examine changes in the number of network members youth discussed sexual health topics with over a three-month period. We also examine the role of sexual health conversations as a mechanism of change leading to decreased sexual risk behaviors.

Methods: Data come from Have You Heard?, a three-group social network intervention to increase HIV testing and reduce sexual risk behaviors among youth experiencing homelessness (Rice et al., 2021). In two intervention groups, PCAs were selected to disseminate HIV prevention messages using an Artificial Intelligence algorithm (AI group) or Degree Centrality, a network measure of popularity (DC group); a third group served as an observation only comparison (OBS group). Participants in all groups reported on the number of people they talked to about HIV, HIV testing, condoms, and Pre-exposure Prophylaxis (PrEP). Surveys were completed at baseline, 1-month, and 3-months follow-up. (The PCA intervention occurred between baseline and 1-month.) A second-order latent growth curve model was used to model change in sexual health conversations over time; changes in the AI, DC, and OBS group were compared using multiple-group models. Additionally, mediation models examined whether changes in sexual health conversations mediated relationships between intervention group and behavioral outcomes (condomless sex and HIV knowledge).

Results: Significant increases in sexual health conversations were observed in the AI (M_slope = 0.22, p < 0.05) and DC (M_slope = 0.40, p < 0.05) groups, but not in the OBS group (M_slope ≈ 0, p > 0.05). Average increases did not significantly differ between the AI and DC groups. Mediation models showed that increases in sexual health conversations significantly mediated the association between intervention group (relative to OBS) and HIV knowledge (AI group: b = 0.01, p = 0.02; DC group: b = 0.01, p = 0.05). However, sexual health conversation did not significantly mediate the relationship between intervention group and condomless sex.

Conclusions and Implications: The current study suggests that using PCAs to disseminate HIV prevention messages results in increased conversations about sexual health topics throughout social networks of youth experiencing homelessness. Over a three-month period, these conversations may result in increased knowledge about HIV, but not increased condom use during intercourse. More research is needed to understand how PCA interventions may lead to behavior change, and how to implement PCA interventions more broadly.