Exploring Substance Use and Network Composition of Homeless Youth with and without a History of Foster Care

Background: Homeless youth suffer from a wide range of risk factors that impact the length and quality of their lives. As many as 40% of all homeless youth report a history of foster care and emerging research suggests behavioral health differences between youth with and without such histories. While the literature has established higher methamphetamine use among homeless former foster youth, factors impacting this outcome have remained unclear. This study aims to explore network composition and differences that may exist in homeless youth networks where foster care experience and high methamphetamine use are concerned.

Methods: The YouthNet data set was used for this analysis, comprised of a community-based sample of 652 homeless youth ages 13-25 from two drop-in centers in Hollywood and Santa Monica, CA. A subset of this data was used for the current analysis, focusing on youth aged 18-25. Both egocentric and sociometric data were collected. All responses from the egocentric data were based on self-reports and included questions regarding foster care experience and recent methamphetamine. Exponential Random Graph Models (ERGM) were run in R to explore network properties and to determine if foster youth and youth using methamphetamines were independently disproportionately connected to one another.

Results: Initial network analysis revealed that the network had very low density (.009), a low level of connectedness (.278), and transitivity of 11.01%. This network has a very low level of centralization (.0061), with 89 components, and a high proportion of fragmented pairs of nodes (.722). Results of the preliminary ERGMs indicate that fewer than half of the possible edges exist in this network (β=-9.27, S.E.=.046, P<.001), there was a tendency toward reciprocity, and a very slight tendency toward triangles (β=.027, p<.001). The test for homophily indicated that former foster youth significantly disproportionately congregate with other foster youth (β=.268, p<.001) while controlling for high methamphetamine use, while youth using high rates of methamphetamines do not significantly disproportionately congregate with one another when controlling for foster care experience.

Discussion: This study has important implications for understanding network composition among homeless youth engaging in substance use. While prior literature has found that meth using youth tend to have meth using network ties, this study indicates that among homeless youth engaging in high rates of methamphetamine use, former foster youth are likely to connect to one another, regardless of their methamphetamine use, while homeless youth using high rates of methamphetamine are not significantly likely to group together.  These networks only include youth who are a part of drop in center programs and not the larger world of social ties encountered by youth (e.g. case workers, using friends who are older). Foster care histories are shaping these service using social spaces, but not heavy meth use.  Perhaps heavy meth using youth are connecting primarily to meth using peers outside this network space. Interventions aimed at promoting long-term behavioral health should focus on the unique network composition of former foster youth.