There is a growing interest in understanding the influence of social networks on substance use risk and health-promoting behaviors. Analyses of network data can provide insights not possible from an examination of individual behaviors alone. For example, network analyses can identify individuals, who based on their network position and connections to others, are at increased risk for acquiring disease and/or who might benefit most from behavioral risk reduction or treatment as prevention (Bell et al., 1999, Friedman et al., 1997, Helleringer et al., 2009, Neaigus, 1998, Potterat et al., 1999, Rb et al., 1998, Rolls et al., 2013, Rothenberg et al., 1995, Wang et al., 2011).

Prior research suggests that networks may influence both high risk substance use behaviors and health-seeking or risk-reducing behaviors (e.g., injection cessation, medication for opioid use disorder enrollment, syringe sharing practices). For example, Latkin et al. (1995) reported positive associations between frequency of injecting drugs and network density/size among PWUD in Baltimore, Maryland. Previous work in the Appalachian context found that (1) an individual's injection status was positively and significantly associated with the number of direct and indirect relationships to others who injected drugs (Rudolph et al., 2017), (2) individuals with a network member who ceased injecting were more likely to stop injecting drugs over the next 6-month period (Rudolph et al., 2020) and (3) network norms better predicted both recent and sustained injection cessation than did individual-level characteristics (Rudolph et al., 2021). These studies suggest the potential for network-based interventions to impact drug use behaviors among PWUD in rural Appalachian Kentucky; however, the specific mechanisms responsible for this influence are not completely understood.

Findings from multiple studies suggest that network-based interventions can effectively promote positive behavior change (Hunter et al., 2019; Latkin and Knowlton, 2015, Valente, 2012). However, it is important that the intervention considers network structure and aligns with the mechanisms driving (1) the formation, maintenance, and dissolution of relationships and (2) behavior change (Valente, 2017). A better understanding of the extent to which drug use behaviors are influenced by selection (i.e., the tendency for people to form or maintain social connections with those with similar behaviors) and social influence (the propensity for people to adjust their behaviors to match those of their peers), can be used to tailor an intervention to a specific population. Evidence supporting both selection and influence theories has been well-documented for a variety of behaviors, but understanding the role each plays with respect to the evolution of social network connections and behavior changes is challenging (Jackson et al., 2023, McPherson et al., 2001).

Bohnert et al. (2009) utilized structural equation models and found evidence of selection and influence for heroin and cocaine use among PWUD in Baltimore, Maryland. However, in that study, participants were asked to report their network members’ behaviors (i.e., data were not sociometric and network member behaviors were not self-reported). As the study was conducted in an urban setting between 1997 and 2003, the findings may not reflect the recent time period nor be generalizable to rural settings, where the types of drugs used, drug use behaviors, and network stability and structure may differ.

To address the gap in the literature, we employed SAOMs to study the effects driving network and behavior change with respect to self-reported drug use within a network of adults in rural Eastern Kentucky. The analysis aimed to (1) identify factors that contribute to relationship turnover and maintenance within the network, (2) determine whether influence and/or selection shape participants use of injection drugs, heroin, and stimulants (methamphetamine and cocaine), and (3) assess the extent that these mechanisms impact network ties (i.e. relationships) and/or behaviors and whether these effects vary across time.