Peer influence of injection drug use cessation among dyads in rural eastern Kentucky

Abstract

Background

This analysis aims to assess whether injection drug use cessation among peers predicts injection drug use cessation among individuals and explores whether this association varies by relationship type and strength.

Methods

Data were collected through baseline and 6-month assessments for the Social Networks among Appalachian People study (2008–2011). Interviewer-administered surveys collected sociodemographic and drug use behaviors (past 6 months and lifetime). Participants also listed sex, drug use, and social support partners (past 6 months). Listed names were cross-referenced with survey participants to identify relationships between study participants. The analytic sample was further restricted to include only those relationship pairs where both individuals reported a history of injection drug use at baseline (n = 244 unique individuals and 746 dyads). We fit a generalized estimating equations logistic regression model to (1) assess the relationship between peer injection cessation and individual injection cessation and (2) determine whether the strength of this association differs by relationship-level variables (i.e., relationship role, relationship type, relationship duration, frequency of interaction, residential proximity).

Results

Overall, those with a network member who ceased injection drug use were more likely to stop injecting over the following 6-month period (AOR=1.65). The magnitude of this association was greater for social support partners (AOR=2.95), family members (AOR=3.56), those with whom the participant interacted at least daily (AOR=2.17), and those who the participant knew longer (AOR=2.09). Further, among family members, the effect size was greater when relationships were further restricted to immediate family members (AOR=5.35).

Conclusion

Our findings suggest that in this rural community, closer, more supportive relationships, may be more influential for modeling injection cessation; however, relationship-types were not mutually exclusive so differences in effect size across strata may not be statistically significant. In this setting, social support through the recovery process (including cessation attempts with peers) may increase likelihood of injection cessation.

Introduction

Rural Appalachia has been hit hard by the intersecting opioid use disorder (OUD), injection drug use, and HCV epidemics. (Zibbell et al., 2015) Between 2006 and 2012, the annual HCV incidence rates in rural areas were nearly double those in urban areas, and were particularly high in the central Appalachian region of the United States. (Zibbell et al., 2015) Following the HIV outbreak in rural Indiana, the CDC conducted a vulnerability assessment, which identified 220 counties with high rates of HCV and increased vulnerability to an HIV outbreak (if HIV were introduced) among persons who inject drugs (PWID). (Van Handel et al., 2016) Of note, 54 of these 220 counties are in Kentucky. Kentucky is also an epicenter for overdose. For example, between 2000 and 2010, Kentucky's overdose fatality rate increased by 282%. The mortality rate continued to increase between 2010 and 2016 (23.6 to 33.5 deaths per 100,000 population). (Centers for Disease Control & Prevention, 2017) In the context of the risk environment framework and the modified social ecological model, disparities in the prevalence of HCV, injection drug use, overdose, and other drug-related harms are the result of upstream social and structural factors (i.e., social, economic, political, organizational). (Baral, Logie, Grosso, Wirtz & Beyrer, 2013; Galea, Ahern & Vlahov, 2003; Rhodes, 2002, 2009) Although these harms are experienced at the individual-level, their relationship with upstream factors may be mediated or moderated by social networks and social support. Drug use is ultimately a “social activity, involving social interactions within particular social environments” (Rhodes et al., 2003).

At the individual-level, injection cessation has been associated with the type(s) of drug(s) used, less frequent recent injection drug use, and sociodemographic characteristics (i.e., younger age, employed, male gender) (Deren, Kang, Colón & Robles, 2007; Evans, Hahn, Lum, Stein & Page, 2009; Mehta et al., 2012; Nambiar, Agius, Stoové, Hickman & Dietze, 2015; Steensma, Boivin, Blais & Roy, 2005). Those using heroin, heroin with other drugs, alcohol, or benzodiazepines; and those who were homeless were less likely to cease injection drug use (Evans et al., 2009; Nambiar et al., 2015; Steensma et al., 2005). Several studies noted that those who recently participated in drug detoxification, residential treatment, methadone maintenance, and 12-step programs (Evans et al., 2009), and specifically methadone maintenance (Horyniak et al., 2018; Nambiar et al., 2015), were more likely to cease injection drug use. However, one study reported that participation in methadone maintenance, outpatient treatment, residential treatment, Narcotics’ Anonymous, or detoxification was not associated with cessation of injection drug use – likely due to limited access (Hadland, Wood, Nosova, Kerr & DeBeck, 2017) and another reported that abstinence-based treatment methods like 12-step programs impeded injection cessation efforts (Boyd, Fast, Hobbins, McNeil & Small, 2017). There have also been mixed findings with respect to the direction of the association between injection drug use cessation and recent arrest (Hadland et al., 2017; Horyniak et al., 2018; Nambiar et al., 2015), and incarceration (Deren et al., 2007; Mehta et al., 2012).

At the neighborhood level, Nandi et al. (2010) reported a strong association between neighborhood poverty and injection drug use cessation, with those living in neighborhoods with a greater proportion of residents living in poverty being more likely to continue injecting.(Nandi et al., 2010) Genberg et al. (2011) similarly found that those living in neighborhoods with greater deprivation were less likely to cease injection drug use. They also reported that those who moved neighborhoods were more likely to stop injecting drugs, but that this effect varied depending on the level of deprivation of the neighborhood that the individual was moving to; moving from a highly deprived to less deprived neighborhood had the strongest positive effect, while continuing to reside in a deprived neighborhood was most detrimental. (Genberg et al., 2011)

According to the risk environment framework and the modified social ecological model, the relationship between aspects of the physical and social environment (i.e., neighborhood disadvantage, availability of treatment for substance use and mental health disorders, access to drugs and sterile injecting equipment) may be modified by social networks and social support. Similarly, in Berkman's conceptual model for how social networks impact health, social networks (measured via network characteristics, network relationships, and structure) act at the mezzo level to provide opportunities for social support, social influence/selection, social engagement, access to resources, and person-to-person contact. These in turn influence risk and health-seeking behaviors. (Berkman & Glass, 2000; Berkman, Glass, Brissette & Seeman, 2000) Prior studies have shown that those with larger and more dense drug use networks (i.e., structure) reported more frequent injection drug use (Latkin et al., 1995) and riskier injection practices. (De, Cox, Boivin, Platt & Jolly, 2007; Latkin, Mandell, Vlahov, Oziemkowska & Celentano, 1996) Another cross-sectional study reported that the odds of injection increased as the number of socially proximal peers (≤ 2° of separation) who injected increased (i.e., network characteristics). (Rudolph, Young & Havens, 2017). There have been a few longitudinal studies which have examined injection risk behaviors and changes in drug use over time. For example, one longitudinal study characterized the influence of network turnover (i.e., network characteristics) for both network members who used drugs and those who did not on injection risk behaviors. They reported an increase in riskier injection practices when individuals who injected were added to the network, but a reduction in risk behaviors when those added to the network did not inject. They also reported a reduction in risky injection practices when individuals who injected drugs left the network. (Costenbader, Astone & Latkin, 2006) Buchanan and colleagues similarly used longitudinal data to examine changes in one's network following drug use cessation (but not injection drug use cessation) and found that drug use cessation was positively associated with a reduction of PWUD in one's network at a later time point. (Buchanan & Latkin, 2008)

Another longitudinal study specifically aimed to examine the role of both social selection and social influence in the longitudinal association between heroin and/or cocaine use among individuals and their networks. Their findings provide support for both social influence and social selection processes among adults with persistent drug use over time. However, most longitudinal changes in drug use were due to changes in network composition rather than to changes in friends’ behaviors. (Bohnert, Bradshaw & Latkin, 2009) Compared with prior findings from studies among adolescents, this study among adults reported that social selection played a greater role. (Bohnert et al., 2009) They hypothesized that this is due to adults having greater control over who they spend their time with and having less malleable attitudes than adolescents. (Bohnert et al., 2009) The authors also explained that compared with network members with whom individuals report drinking alcohol, drug use network members may play a greater role in helping individuals to acquire drugs. (Bohnert et al., 2009)

Another study which examined the relationship between network characteristics and cessation of heroin, crack and cocaine among an urban sample of persons who injected drugs at baseline found that having a smaller proportion of PWUD in their network at baseline predicted drug use cessation at follow-up, suggesting a role for social influence on cessation of drug use. (Latkin, Knowlton, Hoover & Mandell, 1999) These findings remained significant even after adjusting for participation in a drug treatment program, which suggests that changes in network composition are not only the result of drug treatment enrollment. (Latkin et al., 1999)

A smaller number of studies have focused on network correlates of injection drug use cessation. For example, a few prospective studies reported that injection drug use cessation was less likely for those who maintained relationships with network members who use drugs (Bouhnik et al., 2004; Buchanan & Latkin, 2008) or inject drugs. (Deren et al., 2007) Findings from a qualitative study suggest that injection cessation is more likely among those with friends, family, or care providers who could provide housing or social support. (Boyd et al., 2017)

An area that has not been extensively studied is whether injection drug use cessation among peers predicts injection drug use cessation among individuals, particularly in a rural setting and further, whether this association is modified by relationship-level factors. Because injection drug use is an observable behavior that is not necessarily transmitted through risk behaviors, it is unknown whether the behaviors of some peers are more influential than others. For example, is the relationship between peer injection cessation and individual injection cessation different for network members with whom the individual uses drugs or has sex, or for relationships where social support is exchanged? Similarly, the extent to which the association between peer injection cessation and individual injection cessation varies by relationship duration, geographic proximity or shared environment, frequency of interaction, and closeness of the relationship has not been studied. The purpose of this analysis is to (1) assess whether having a peer who had a history of injection but who stopped injecting (i.e., injection drug use cessation) predicts injection drug use cessation within the past 6 months among individuals (all relationships considered equal) and (2) explore whether the strength of this association varies by relationship-level factors (i.e., relationship role, relationship type, duration of relationship, frequency of interaction, residential proximity) among an adult sample of PWUD in rural Eastern Kentucky.