The influence of poly-drug use patterns on the association between opioid agonist treatment engagement and injecting initiation assistance

The PReventing Injecting by Modifying Existing Responses (PRIMER) study seeks to assess the impact of socio-structural factors on the risk that PWID provide injection initiation assistance [3]. The PRIMER study protocol and rationale have been previously described [3]. For the present analysis, data were drawn from three linked open prospective cohort studies of people who use drugs (PWUD) in Vancouver, Canada: The At-Risk Youth Study (ARYS; street-involved youth aged 14–26 who use drugs), the AIDS Care Cohort to evaluate Exposure to Survival Services study (ACCESS; HIV-seropositive adult [≥18 years of age] PWUD), and the Vancouver Injection Drug Users Study (VIDUS; HIV-seronegative adult PWID). To be eligible, participants had to report unregulated use of (i.e., illegal/street-based) drugs other than cannabis ≤30 days prior to baseline. The PRIMER baseline is defined as the visit when specific survey items soliciting reports of assisting others in their first injection were introduced into the cohort questionnaires (December, 2014-May, 2017). The current study is cross-sectional and was restricted to participants who reported past six-month IDU at PRIMER baseline. The inclusion criterion of recent IDU was selected given that the provision of IDU initiation assistance is a relatively rare event [23, 24], and that focusing on the population most likely to provide this assistance would allow for the most accurate description of the phenomena of interest [1, 3]. This study was approved by Institutional Review Boards at UC San Diego and the University of British Columbia-Providence Health Care. All participants provided written informed consent.

Participants completed an interviewer-administered questionnaire assessing sociodemographics, drug use behaviors, health services enrollment, and other domains. The primary outcome, recently assisting others in their first injection, was assessed via the following survey item: “In the past six months, have you helped someone inject who had never injected before?”). The independent variable, recent OAT engagement, was defined via endorsement of the statement: “In the past six months have you been in (methadone/methadose program or Suboxone) treatment?” The covariates of age and sex (i.e., male/female) were selected based on previous studies of IDU initiation assistance [22‐24], and all variables of interest were captured at participants’ PRIMER baseline visit. Analyses were stratified based on drug use practices identified within a previously conducted latent profile analysis (LPA) ([25]; Rammohan I, Jain S, Sun X, Marks C, Milloy M-J, Hayashi K, et al: Identifying latent polydrug use patterns and assessing their association with the provision of injection initiation assistance in three north American settings: a latent profile analysis, In Preparation). The LPA methods have previously described in full; briefly, six classes of poly-drug use were identified based on participants’ recent IDU and non-injection (NIDU) drug type (i.e., methamphetamine, heroin, cocaine, and prescription opioids [PO]) and frequency of use (i.e., low, moderate, and high) ([25]; Rammohan I, Jain S, Sun X, Marks C, Milloy M-J, Hayashi K, et al: Identifying latent polydrug use patterns and assessing their association with the provision of injection initiation assistance in three north American settings: a latent profile analysis, In Preparation).

Cross-sectional bivariate associations between OAT engagement and providing IDU initiation assistance were assessed among each LPA class using cross-tabulation and Fisher’s exact tests. Multivariable logistic regression models were performed for each LPA class separately to study the independent association of reporting OAT use on the provision of injection initiation assistance while controlling for age and sex. Analyses were performed using R software (version 3.5.1). Two-sided P-values < 0.05 were considered statistically significant.

This study has limitations typical of observational cross-sectional research. Non-probability sampling was used for participant recruitment due to the mobile nature of the population and the corresponding difficulty of developing a sampling frame. Consequently, we cannot assume generalizability for the population of PWID in this study setting. Secondly, we relied on self-report, and underreporting of experiences of initiating others into injecting is likely given it is highly stigmatized, particularly within the context of an overdose crisis [32]; however, research suggests that PWID accurately self-report OAT engagement [33] and substance use-related behaviors [34], indicating these measures are valid. Additionally, to protect against potential response bias, participants were notified of the confidential nature of the study. Third, given the cross-sectional nature of the data collected and the fact that we did not construct causal inference models, we cannot determine causality among OAT engagement and reductions in providing IDU initiation assistance. Further, given that we conducted separate multivariable analyses for each LPA class, we cannot directly compare findings between classes. Finally, we were limited in the number of covariates we could include in our multivariable models due to small subsample sizes and because our outcome was a rare event. Small subsample sizes could have also increased the likelihood of committing Type II error and potentially limited the statistical significance of the study findings [35]. Nevertheless, the findings of this study are in-line with similar studies that found associations between substance use, OAT engagement, and injection initiation events [11, 22].