The brief counseling intervention will utilize MI and skills-building techniques to modify personal overdose risk behaviors and develop skills as a peer responder for witnessed overdose. The counselor will draw upon themes of safer substance use to address HIV risk behaviors and determine readiness for change in substance use.

The control group will have access to brochures and be offered referral to services requested. SFDPH Community Behavioral Health Services (CBHS) provides immediate access to substance abuse treatment in San Francisco, including office- and clinic-based methadone and buprenorphine treatment. Given the pilot nature of this study, the control group will not be a full attention control; we will account for an attention effect due to assessment alone.

The brief counseling intervention will utilize MI and skills-building techniques to modify personal overdose risk behaviors and develop skills as a peer responder for witnessed overdose. The counselor will draw upon themes of safer substance use to address HIV risk behaviors and determine readiness for change in substance use.

To determine feasibility of a randomized trial with REBOOT, we will calculate screening and visit completion rates from the study database, with exact 95% confidence intervals (CIs), overall and by arm. Between-group differences will be assessed using Fisher's exact and Wilcoxon ranksum tests. We will calculate Kaplan-Meier curves for time to dropout, by group, and test for differences using the log-rank test.

To determine acceptability of REBOOT, we will calculate counseling completion rates from the study database by visit, and tabulate the proportions of active arm participants attending 0-4 counseling sessions. Via ACASI, we will inquire about participant satisfaction with the intervention and belief that it affected their drug use behaviors; responses will be presented as means, medians, or proportions, as appropriate, with 95% CIs.

To evaluate the influence of egocentric social network characteristics on overdose events and naloxone use, we will use GEE Poisson models with robust standard errors to evaluate the association between network size, evaluated at baseline and each return visit, and numbers of experienced and witnessed overdose events in the same period; zero-inflated models will be used if needed. In addition, we will explore the influence of homophily and assortativity on experienced and witnessed overdose events using similar methods. Newman's method will be used to calculate assortativity coefficients, a measure of the degree of demographic and risk behavior similarity within participants' egocentric networks.