The Stonewall Project Harm Reduction Treatment Model
The Stonewall Project model translates evidence-based interventions such as the Matrix Model 13,14 into a clinical setting with a harm reduction focus. 19 Harm reduction is a client-centered philosophy which does not assume that all individuals are ready, willing, and able to pursue abstinence as a treatment goal. It also acknowledges that clients often seek out substance abuse treatment services to address risky injection practices and sexual risk-taking behavior, even when they are not interested in changing substance use patterns. Consistent with the harm reduction philosophy, the Stonewall Project model assists clients with pursuing self-identified treatment goals as a means of engaging individuals who might not otherwise initiate or remain in abstinence-based substance abuse treatment. Regarding substance use, clients can choose to pursue abstinence as a treatment goal, but strategies for managing substance use are often a primary focus of treatment. Selected substance use management strategies in the Stonewall Project model include: (1) transitioning to less potent modes of methamphetamine administration (e.g., injecting to smoking, smoking to snorting); (2) promoting self-care strategies while using methamphetamine (e.g., hydration, nutrition); and (3) delivering education about safer injection practices with linkage to needle exchange and access to sterile syringes. The Stonewall Project model also delivers sexual risk-reduction interventions to promote condom use during anal sex as well as seroadaptive behaviors (e.g., serosorting, strategic positioning) for when clients choose not to use condoms.20 Clients are encouraged to develop plans for addressing barriers to sexual risk reduction, even in the context of methamphetamine use. As part of the Stonewall Project model, clients receive outpatient treatment that consists of weekly individual counseling, group counseling twice a week, and psychotropic medications where appropriate. Unlike the Matrix Model,13,14 clients are not asked to provide weekly urine samples to test for recent stimulant use as part of their ongoing treatment. This 1-year outpatient drug treatment program has been implemented for over 15 years by the community-based substance abuse treatment programs that were the focus of the present outcome studies.
Treatment Outcome Study 1
Study 1 was conducted with two co-located outpatient programs that were implementing the Stonewall Project model. One of the programs specialized in the treatment of methamphetamine-using MSM. The other served methamphetamine users of any gender or sexual orientation, but only MSM clients were included in the present study. All consecutive admissions to these programs from March 2005 to May 2008 were eligible. At the end of their program intake session, interested clients gave permission for research staff to contact them (i.e., self-selection).
At the baseline assessment, research staff uninvolved with treatment services completed the informed consent process and administered the ASI at the substance abuse treatment site. At 6 and 12 months, the follow-up ASI was administered at the substance abuse treatment site. In total, 132 methamphetamine-using MSM enrolled in study 1, but nine were removed because they were also enrolled in study 2. Of the 123 unduplicated participants, 107 (87 %) and 98 (80 %) completed follow-up assessments at 6 and 12 months, respectively. In total, 112 participants (91 %) completed at least one follow-up assessment over the 12-month period. For their time and travel expenses, participants received US$10 at the baseline and 6-month assessments as well as US$20 at the 12-month assessment. All participants signed a separate authorization allowing research staff to extract data regarding treatment utilization from clinical billing records. Study 1 procedures were approved by the University of California, San Francisco Committee on Human Research.
Demographic information including items assessing sexual orientation and gender identity was obtained from a modified demographics form for the ASI. Among HIV-positive persons, self-reported ART utilization and undetectable viral load were assessed via questionnaire. The ASI was administered at each study visit. Responses were entered into ASI Drug Evaluation Network System (DENS) software, which calculated composite scores.21 As part of the ASI, participants reported the number of days that they had used specific substances in the past 30 days. Substances examined in this study included: methamphetamine, cocaine/crack, marijuana, and drinking to intoxication.
Treatment Outcome Study 2
Methamphetamine-using MSM were recruited from the outpatient substance abuse treatment program in study 1 that is implementing the Stonewall Project model exclusively with MSM from July 2010 to June 2012.22 Clients receiving treatment at the Stonewall Project were eligible to enroll in study 2 up to: (1) 60 days after treatment initiation; or (2) 60 days following re-initiation after more than 30 days out of treatment (NCT 01129401). Clients who were interested in participating completed a consent to contact form with their Stonewall Project counselor (i.e., self-selection) that included permission to verify eligibility using treatment records.
After providing informed consent at the baseline assessment visit, participants completed an assessment administered by a research assistant at the substance abuse treatment site. This assessment included measures of self-reported substance use and sexual risk taking that were administered using audio computer-assisted self-interviewing (ACASI). Prior research has demonstrated that ACASI enhances the reliability of self-report measures for substance use and sex risk.23,24 This assessment was re-administered at 3 and 6 months follow-up at the substance abuse treatment site. Participants also provided a urine sample for on-site toxicology screening for methamphetamine and cocaine metabolites at each study assessment (Redicup®; Redwood Toxicology Laboratory; Santa Rosa, CA). All study assessments were administered by a research assistant who did not have a clinical role in the substance abuse treatment program.
Of the 88 participants enrolled in this treatment outcome study, 81 (92 %) and 79 (90 %) completed follow-up assessments at 3 and 6 months, respectively. In total, 85 participants (96 %) completed at least one follow-up assessment over the 6-month period. At each study visit, participants were reimbursed with a US$50 pre-loaded debit card for their time and travel expenses. All participants signed a separate authorization allowing research staff to extract data regarding treatment utilization from the clinical records. Study 2 procedures were approved by the University of California, San Francisco Committee on Human Research.
Demographic characteristics assessed by questionnaire included: age, ethnicity, education, income, sexual orientation, gender identity, and HIV status. Among HIV-positive persons, self-reported ART utilization and undetectable HIV viral load were assessed at each study visit. Consistent with study 1, participants reported the number of days they used specific substances in the past 30 days. Substances examined in this study included: methamphetamine, cocaine/crack, club drugs (i.e., ecstasy, ketamine, or GHB), marijuana, binge drinking (i.e., six or more drinks on one occasion), and erectile dysfunction (ED) medications in combination with other substances (i.e., while “partying”). Participants also reported the longest number of days in a row they used cocaine/crack or methamphetamine in the past 30 days. Informed by previous research with methamphetamine users, 25 participants who reported using any of these stimulants two or more days in a row were classified as engaging in binge use (1) and compared to those who reported using 1 day at a time or no use in the past 30 days (0). Participants who provided a urine sample that was reactive for cocaine or methamphetamine metabolites (1) were compared to those who provided a urine sample that was not positive for recent stimulant use (0).
Participants reported the number of anal sex partners in the past 3 months, stratified by whether or not they were feeling the effects of methamphetamine during sexual intercourse. For HIV-negative MSM, sexual risk-taking behavior was operationalized as engaging in unprotected anal intercourse, irrespective of the serostatus of the sexual partner(s). For HIV-positive MSM, sexual risk-taking behavior was operationalized as unprotected anal intercourse with HIV-negative or unknown serostatus partners. Estimates of any sexual risk-taking behavior were calculated separately for receptive and insertive anal sex as a function of whether participants were using methamphetamine to yield four separate indicators.
Inferential analyses examining unadjusted change over time for each dependent variable were performed with generalized estimating equations (GEE) in Stata using the binomial distribution and logit link for binary dependent variables (e.g., ART, self-reported undetectable HIV viral load), the multinomial distribution and cumulative logit link for ordinal categorical dependent variables (i.e., ASI Alcohol and Medical composite scores), the negative binomial distribution and log link for count dependent variables (e.g., number of methamphetamine days, number of anal sex partners), and the normal distribution and identity link for continuous dependent variables (e.g. ASI Drug and Employment composite scores). As recommended by Diggle, 26 the unstructured covariance structure was used for binary, count, and continuous outcomes. For the ordinal categorical dependent variables, the independent covariance structure was used. Multiple imputation was employed to handle missing data for any models with incomplete data due to both participant non-response (≤18 % in study 1, <2 % in study 2) and loss to-follow-up (≤20 % in study 1, ≤10 % in study 2). Thus, the total sample size is the same (N = 123 for study 1, N = 88 for study 2) in each reported analysis.
For binary outcomes, the odds ratio (OR) per unit change in the independent variable is reported. For count outcomes, we report the incidence rate ratio (IRR) as well as the percent change in the expected number of the outcome (Δ expected = 100 × (e
B − 1)). For continuous outcomes, the raw change in the outcome per unit change in the independent variable (B) is reported. Observed effect sizes were determined using Cohen’s d for continuous and count dependent variables, Cohen’s h for binary dependent variables, and the Uncertainty Coefficient (c|r) for ordinal dependent variables.27–29
Dose–response analyses were performed only for dependent variables that exhibited significant change over time. To improve imputations by allowing information contained in other non-missing variables to inform missing values, all dependent variables in the dose–response analyses were used together to create 50 imputed datasets. Three different dose variables were examined individually as correlates of change over time for each significant dependent variable: number of individual counseling sessions, number of group counseling sessions, and number of psychiatric sessions. The linearity assumption, assessed via the cumulative sums of residuals method, 30 was not violated in any of the models. To assess the associations of dose variables with outcomes over time, the interaction of dose and time was considered in a model including both main effects. In the two instances where the dose-time interaction was found significant, the associations for dose were calculated at the final follow-up visit for that study.