The ethnographic evaluation is guided by the Consolidated Framework for Implementation Research. The CFIR assesses five domains of interventions, outer settings, inner settings, provider characteristics, and participant characteristics. The evaluation includes pre- and post intervention in-depth interviews with key informants who participate in the study and ethnographic observation with participants in their daily activities at the clinics and in the community settings.Participants are recruited from the methadone clinics in Hanoi and Ho Chi Minh City —the two largest urban settings in Vietnam. As of March 2021, there are 23 methadone clinics in HCMC and 18 in Hanoi, treating 5047 and 4655 patients, respectively. Criteria for selecting clinics include number of patients, estimated prevalence of methamphetamine use, availability of human resources to implement study interventions, and space for intervention activities. Ten clinics in each city will be randomly selected from those that meet the criteria.An investigator will stratify participants by HIV status and randomize them into two front line interventions using REDCap, as shown in Fig. 1. She will send the allocation results to the site research assistants who will then inform providers at the study clinics. This is an open-label study so unblinding does not occur. Participants and their counselors, research assistants, and data managers are aware of participants’ assignment of intervention. Participants will receive two individual sessions of motivational interviewing with their counselors before they get into the frontline interventions and, in week 13, before the adaptive strategies start. These sessions would boost participants’ motivation for intervention and provide them with greater details of the upcoming intervention activities.

In each selected clinic,cannabis grow equipment a physician, two counselors, and one nurse will participate in the study as intervention providers. The physician will ensure referral to HIV and psychiatric services when necessary; two counselors will run motivational interviewing, group education sessions, and Matrix meetings; the nurse will collect urine twice a week and conduct contingency management based on the UDS results. Before the start of the intervention, to ensure the accuracy,integrity, and fidelity to the EBIs, all intervention staff at methadone clinics will receive didactic training on the theory behind the approach, evaluate their comprehension of the concepts within and behind the approach, watch a video of a Master Behavioral Counselor conducting intervention sessions and discuss the details of the session, and conduct at least two pilot intervention instances. All intervention sessions, except contingency management, will be audio-recorded, transcribed, and coded to ensure intervention fidelity. Intervention staff who have lower levels of intervention integrity or who have significant drift will be provided detailed feedback and supervision until there is parity with other staff.Sample sizes were chosen to compare primary outcomes based on first-stage randomization into one of two groups: high intensity or low intensity front line interventions. Sample size calculations are conducted in PASS 2008 for a two-group comparison of binary outcomes, a power of 80%, a 5% alpha level, and a conservative attrition rate of 20%. Using estimates from our prior work, we anticipate base rates of 80 to 90% for substance use and 60 to 70% for viral suppression. Based on these assumptions and a proposed sample of 200 HIV-positive participants , we can detect randomization group differences of 20% or more for binary outcomes, such as substance use and viralload suppression.

We can detect even smaller group differences for substance use outcomes in the proposed sample of 400 HIV-negative participants and the combined sample of HIV-positive and HIV-negative participants. If estimated outcome probabilities are similar between first-stage randomization groups at 12 weeks, we will pool 12-week results for even greater power in evaluating second-stage randomization differences.Different datasets collected from different sources will be linked through a unique identification code using RED Cap for quantitative data. Data will be uploaded in real-time from the 20 study clinics onto our database. The study data manager will assess transferred data for completeness, query sites regarding any inconsistencies, and code merged data files for analysis. For qualitative data, field notes written on site are expanded and recorded electronically within 24 h. After removing all personal identifiable information, the research team will upload password-protected transcripts on a secured database. The transcripts will be uploaded into Atlas.ti software to organize data and facilitate analysis.We will use a time-varying mixed-effects model that will be fitted to the participants’ common outcome measures over time. The unadjusted model will include indicators of first-stage and second-stage intervention conditions, time of the assessment , and intervention indicators-by-time interaction terms. An additional interaction term of the two intervention indicators will be included to account for any interaction effect between the first and the second stage interventions. The adjusted model will include patients’ socio-demographic characteristics, drug use history, HIV-serostatus, and location as fixed effects. The mixed effects models will include a participant-level random effect to account for repeated observations of each participant, as well as a clinic-level random effect to account for the nested nature within the clinics. We will conduct subgroup analyses among HIV positive and HIV-negative participants.

For the HIV positive subgroup, the specific outcomes of interest include HIV viral load suppression and adherence to antiretroviral treatment, and specific outcomes for HIV-negative subgroup include frequency of HIV testing and HIV seroconversion. Substance use will be the common outcomes in models including participants of both HIV statuses.The qualitative analysis team will read and provide a narrative summary for each transcript. A code book will be developed based on these summaries. Memo-writing and code-refining will be conducted throughout the analysis. Iterative analyses assess convergence of patient, provider and organizational dimensions on study measures, and the context of the policy subsystems, cross system interactions, and resource allocation.We will first describe the extent and patterns of missingness within each variable and check for associations between missing and observed data to determine the mechanism of missingness, which could be missing completely at random, missing at random, or missing not at random. Missing data will then be handled using multiple imputation. Appropriate imputation techniques will be chosen for the type of missing data and the statistical tools employed . For sensitivity analysis, we will conduct analyses with and without multiple imputations. All participants will be analyzed on an intent-to-treat basis where the study outcomes are examined based on the random intervention assignment and not on the actual intervention received or adherence to the intervention.There is no planned interim analysis as the behavioral therapies used in this trial have no known serious adverse events and are consistently more efficacious than control conditions in treatment-seeking participants. The effect sizes of the behavioral therapies in this trial are in the moderate range. Furthermore, any interim analysis and decision to stop the trial would likely be based on under powered data and susceptible to error.Our data monitoring committee is composed of members of the Data and Safety Monitoring Board for Addiction Medicine of the University of California – Los Angeles. These members are not connected to the study in any way. The DSMBAM is independent from the National Institute on Drug Abuse —the sponsor of this study. The DSMBAM meets quarterly to monitor subjects’ progress in the trial and considers whether adverse social harms differentially accrue by condition. Although there are no prospective stopping rules for this trial,horticulture rack the DSMBAM is within its charge to review aggregate data, request statistical tests of differences in social or other harms, and then advise changes in intervention type or intensity if statistically significant differences emerge in adverse events by condition. Prior to each meeting, the study team will submit a performance report including all reports of SAEs for DSMBAM’s consideration. After each meeting, recommendations will be made in writing to the principal investigators.Hanoi Medical University and the staff in the STAR-OM study provide oversight of financial management. The Vietnam teams and US teams maintain frequent communication via emails and bi-weekly online meetings to report updates on the study progress, discuss scientific aspects of the study, and troubleshoot issues when they arise. The teams in Hanoi and HCMC meet online once weekly and in-person quarterly during monitoring visits to discuss the study conduct. We submit annual research progress reports to the Ethics Committee of Hanoi Medical University.

Any protocol amendments need to get ethical approval before implementation. The UCLA Addiction Medicine Data Safety Monitoring Board independently review our data and data management twice a year.Adverse events in this trial are defined as medical issues that do not require hospitalization. Serious adverse events are defined as life-threatening events such or other events that have a negative impact on participants’ life such as incarceration or compulsory drug rehabilitation. The clinic staff will communicate information about adverse events and serious adverse events to the study team right after they are informed by participants or participant families. The study coordinators in Hanoi and HCMC are responsible to report adverse events within 7 days and serious adverse events within 24 h on REDCap with the time of onset, seriousness, duration, and outcomes. The principal investigator will decide what serious adverse events need to be reported to the Ethics Committee.Prior to participation in the trial, the participant will be informed about the research. Participants will complete a short questionnaire about the study objectives and main activities to show how they understand the study. Research assistants will provide more explanation based on the results of the questionnaire. If participants agree to join the study, they will sign a consent form. Each participant will be assigned a unique identifier at the time of screening. Participant data will be linked to this identifier only. Participant personal identifiable information is stored in a separate locked cabinet to which only responsible study staff have access. All study staff sign a confidentiality agreement to non-disclosure of participant information. We make extra efforts to ensure nodisclosure of drug use information to anyone other than participants and the study staff.Between July and October 2020, we conducted 4 focus group discussions of a convenience sample of participants from four methadone clinics in the downtown and suburbs of Hanoi and HCMC to inform intervention content and refinement. Respondents reported information on local taxonomy and patterns of methamphetamine use, triggering situations, methampheta minerelated sexual risks, motivations for seeking treatment, and perceived acceptability of the adaptive interventions. The pilot implementation lasted 12 weeks from November 2020 through February 2021. It identified issues to be addressed before the full implementation. At the conclusion of the pilot, we conducted 2 FGD with patients and 1 FGD with providers participating in the pilot to gauge their feedback about the interventions.With the cut-off point of ASSIST ≥ 4 and methamphetamine-positive UDS as originally proposed, there were 26 and 52 eligible participants in two pilot clinics in Hanoi and HCMC, respectively . For the pilot implementation, we randomly recruited 42 participants with ASSIST score ≥ 4 or methamphetamine-positive UDS. After the front line intervention, 16 participants were non-responders and randomized into adaptive interventions. At least 50% of the original sample must transition to the adaptive phase for sufficient statistical power. Thus, we decided to recruit more participants with severe use of methamphetamine, as evidenced in both ASSIST score ≥ 10 and methamphetamine-positive UDS. Furthermore, to recruit enough participants for the front line intervention phase, given most other clinics are smaller than the two pilot ones, we decided to use ASSIST score “OR” UDS instead of “AND” to increase the pool of potential participants. We kept the criterion of methamphetamine-positive UDS to compensate for participants with lower ASSIST scores due to desirability bias.The STAR-OM study is among the first studies to evaluate different combinations of EBIs for methamphetamine use among methadone patients in low-and-middle-income countries. The study will provide effectiveness and cost-effectiveness evidence for scaling up these interventions. The SMART design assesses different treatment strategies for participants who respond differently to front line interventions. The combination of trial and ethnographical study will provide insights on factors at multiple levels that need to be considered in decision making.As the interventions will be delivered by methadone providers at methadone clinics, they can be readily implemented if the trial demonstrates they help.