A computer-assisted motivational social network intervention to reduce alcohol, drug and HIV risk behaviors among Housing First residents

The study is being conducted in collaboration with Skid Row Housing Trust (SRHT) and Single Room Occupancy Housing Corporation (SRO Housing), two of the largest HF providers in Los Angeles County. Both organizations combine the HF framework for providing rapid housing with minimal preconditions with permanent supportive housing (PSH) to many of its residents [70]. PSH is long-term community-based housing linked with voluntary supportive services for individuals with disabilities who have experienced homelessness. SRHT and SRO Housing manage properties in the Skid Row area of downtown Los Angeles, which has the largest concentrations of homeless adults in Los Angeles County.

Participants will include recently housed residents aged 18 or older who are (1) able to speak and understand English, (2) are not cognitively impaired (based on administration of the Short Blessed Scale Exam during screening for eligibility [71]), and (3) screen positive for past-year harmful AOD use (either drugs or alcohol) using the Alcohol Use Disorders Identification Test (AUDIT-C) (a score >4 for men and >3 for women) [72] and the Drug Abuse Screen Test (DAST) (a score >2) [73‐75]. Recruitment, screening for eligibility, and assessment procedures are depicted in Fig. 1.

Leasing staff will be instructed to introduce the study to all housing applicants that have recently received assignment to a housing unit. Those who express interest in being contacted for the study will be given a consent-to-be-contacted form to complete. Leasing office staff will share the completed consent-to-be-contacted form with the research team. Next, the research team will attempt to contact the potential participant by phone to complete a study screener. If a resident is interested in the study but does not have access to their own phone, the leasing staff will provide access to a phone. If the resident does not contact research staff by phone at leasing and does not provide a working phone number on the consent-to-be-contacted form, a member of the research team will attempt to locate the resident at their residence and provide access to a phone to complete the study screener. Once an interested resident is connected by phone for screening, the research team will describe the study, obtain consent to conduct the screener, and conduct the screener interview to determine eligibility. Next, eligible participants will be asked to schedule an in-person study consent and baseline assessment. Participants will also be contacted for a follow-up assessment 3 months after baseline. Residents will be provided with study appointment cards to assist them in recalling their appointments. Participants will be paid $30 for the baseline (i.e., pre-intervention) assessment and $40 for the follow-up (i.e., post-intervention, approximately 90 days after baseline) assessment.

After participants complete the baseline assessment, they will be assigned to one of two experimental conditions (i.e., intervention or control). We aim to randomly assign 60 participants using a stratified permuted block randomization strategy. Given the relatively small sample size, the randomization might produce an unbalanced assignment with respect to gender. For this reason the randomization will be stratified. About 26 % of SRHT clients are women and stratifying on this characteristic will ensure an approximately equal number of women assigned to both conditions. If the intervention and control groups turn out to be unbalanced with respect to other characteristics, we will use covariate adjustment at the analysis phase. Next, participants assigned to the intervention arm will be contacted by project staff by phone to schedule the four MNI sessions, one every 2 weeks. If the participant is difficult to reach by phone, a project staff member will attempt to contact the participant by visiting their residence to schedule their first session. Participants will be provided with appointment cards as reminders. Participants will be given a $5 food gift card for each intervention session they attend.

The MNI intervention consists of four biweekly sessions that last approximately 45–60 min and consist of two parts: a network interview with closed-ended network questions and a discussion of resulting network visualizations conducted using a MI approach. Structured network interview questions will be similar to previous studies with homeless populations [6‐8, 41, 43, 76, 77]. Participants will be asked to identify people they have had contact with in the past 2 weeks, with follow-up questions about each of these contacts (e.g., likelihood that they will drink alcohol or use drugs in the next 2 weeks, if the participant drank more alcohol or used more drugs than they wanted to when with the contact; if the participant has had unprotected sex with any of the contacts in the past 2 weeks; whether any of the contacts provide social support to the participant; whether the contacts interacted in the past 2 weeks). As the participant talks, the facilitator will enter the participant’s interview responses into the computer-assisted interface, which will subsequently generate several social network visualizations based on the participant’s responses. See Fig. 2 for examples of network visualizations for a hypothetical MNI participant. These visualizations are intended to provide content for discussions during the MNI. The facilitator would show the diagram and then use MI techniques (e.g., asking evocative open-ended questions, stating reflections, elaborating on change talk) to discuss what patterns the participant notices in the diagram (e.g., how certain contacts may trigger their AOD use and how they feel about those contacts). For the example shown in Fig. 2, in Session 1, the facilitator would gather information by stating “Tell me about the different groups of people you see” and “The red circles are people you said you tend to drink more with, tell me about those people.” The facilitator would use MI techniques to explore and resolve any ambivalence about changing their AOD and risky sexual behaviors.

The second–fourth sessions are designed to build on previous discussions. Facilitators will start these sessions with a discussion of the goals identified in the previous session by asking, “Tell me how those goals have been going for you. Have things changed? Did anything get in the way?” After discussion of the goals, the session will include all of the same questions about the participant’s network they were asked in the previous sessions, including a discussion of the current and previous visualizations. For the example shown in Fig. 2, in Session 2, the facilitator might ask “What differences do you see between this week’s diagram and our first one?” The facilitator would highlight any change talk about the new people in their network who do not use AOD (i.e., changes in network structure) and the strategies they used to change their network composition (e.g., fewer who are using AOD). The facilitator would also explore whether the participant is willing to interact less with the two remaining people who do use AOD in their network.

This four session strategy will allow time for network change strategies to be put into place and for participants to discuss change experiences with the intervention facilitators. For example, participants may discuss wanting to spend more or less time with network contacts, wanting to connect network members who do not know each other, or ties to network contacts that they would like to break. Participants may not be able to put strategies to achieve these goals in place right away or they may need to try different strategies. Initial changes in network structure or composition may trigger the opportunity to develop strategies for secondary or tertiary changes. Also, participants may require encouragement from facilitators to maintain initial strategies. Four sessions will allow ongoing discussions of this process. Also, some of their strategies may produce unexpected negative results and they will be able to discuss these unexpected effects with facilitators who can help them develop alternative strategies. In summary, the visualizations resulting from the network interview are designed to provide both intervention facilitators and participants with a tangible object to focus questions and answers and make abstract discussions of social life, relationships, and social change more concrete.

Participants assigned to both the intervention and control conditions will receive usual case management support provided by SRHT and SRO Housing. After an initial case management intake assessment that is scheduled to occur within the first 30 days that a resident is housed, residents are expected to meet with case managers regularly for up to 60–90 days, based on need. Case management meetings last 15–60 min and are customized to the residents’ needs. Case managers facilitate residents’ access to on-site services, such as support groups, as well as make referrals to offsite services.

Intervention facilitators will be trained in MI techniques and a clinical psychologist affiliated with the motivational interviewing network of trainers (MINT) will review the recording of each session to provide regular feedback and re-training, as necessary. Fidelity to the intervention protocol will be assessed with an 8-item scale to determine if key components of the intervention were discussed, similar to previous studies [78‐80]. The fidelity checklist will consist of a five-point Likert scale with response options ranging from “completely”, “partially”, to “not at all” covered. For example, items will assess whether the facilitator discussed the client’s desires to make network changes, such as interacting with specific network ties more or less frequently.

Table 1 gives an overview of the pilot study measures and the phase in which they will be assessed. Study measures were selected based on their use in similar populations, including low income, low education AOD using populations [81, 82] and populations engaging in heavy AOD use and high-risk sexual behavior [83]. Many of the measures have also been recommended by the VA Mental Health QUERI taskforce [84] and NIDA [85]. Because of the pilot nature of this study, only one follow-up assessment will be conducted. We plan to assess participants within 2 weeks after the final MNI session in order to measure the immediate benefits of the intervention rather than longer-term effects. The assessments will be collected in a one-on-one in-person interview by an independent data collection organization using EgoWeb as a computer-assisted personal interviewing (CAPI) software. The data collectors will be blind to the study condition assigned to the participant.

We will measure both proximal and distal outcomes. Proximal outcomes are readiness and self-efficacy to change AOD use and safe-sex behavior, and social network composition and structure. Distal outcomes are AOD use behaviors and HIV risk behaviors. For the proximal outcomes, we will measure readiness to change in two ways. The first method is a one-item contemplation ladder adapted from the smoking literature [86] that displays a picture of a ladder with corresponding stages of change. For the substance that the participant considers their biggest current problem, they are asked to identify where they are in changing their use of the substance on a 0–10 scale (e.g., 0 = “No thought of quitting”, 5 = “Think I should quit but not quite ready”, 10 = “Taking action to quit”). A separate ladder will ask participants to give a similar rating of their willingness to change their unprotected sex behavior. The ratings that these ladders produce have adequate reliability and good validity [86] and have been used for smokers with schizophrenia [87], drinking adults with multiple sex partners [83], and welfare applicants [81]. We will also measure readiness to change using the Readiness to Change Questionnaire (RTCQ) [88‐90], which consists of 12 items that assess the readiness to change AOD use (precontemplation, contemplation, preparation, or action) and the Readiness to Change Risky Sexual Behavior (RTCQ-SB) scale, which consists of 11 items adapted from the RTCQ for safe-sex behavior [83]. We will measure self-efficacy using an abbreviated version of the Alcohol Abstinence Self-Efficacy Scale (AASE), [91] which consists of five items indicating how confident participants feel in their ability to abstain when depressed, relaxed, craving substances, and offered alcohol. We will measure self-efficacy to use condoms with an instrument that has shown good reliability and concurrent validity [92] and has predicted unprotected sex in homeless populations [41, 43, 93].

We will collect social network data similar to previous homeless studies to measure change in network composition and structure [6‐8, 43, 94‐96]. Participants will be asked to name 20 network contacts (“alters”), who are at least 18 years of age, during the baseline and follow-up assessments because a minimum of 20 alters is necessary to produce unbiased estimates of network structure [97, 98] and to measure network composition [7]. Participants will be asked questions about the demographics of each of these network alters (gender, recent and lifetime homelessness), the drinking/drug use/safe-sex behavior of each alter, their relationships with each alter (emotional closeness, frequency of contact), and the relationships between each pair of alters [e.g., “Does (Alter 1) know (Alter 2)?”]. Based on aggregating questions about the social support provided by each alter, we will construct measures that are similar to measures of Perceived Social Support [36, 99] (i.e., the number and/or proportion of network members who provide emotional support, advice, tangible support) and measures that result from the Important People Drug and Alcohol interview, in which network members are rated on relationship characteristics (e.g., emotional closeness), drinking, drug use, safe-sex behavior, and their influence on the participants’ own high-risk behavior [100].

For the distal outcomes, we will use self-reported AOD use as our primary AOD use measure. At each assessment we will assess alcohol drinking frequency (never, once a month, 2–4 times a month, 2–3 times a week, 4–5 times a week, 6 or more times a week) and drinks per drinking day (typical number of drinks when drinking). For drug use, we will assess frequency of past 30 day drug use (none, once, twice, three times, once a week, twice a week, 3–4 times a week, 5–6 times a week, every day) for each of the following: marijuana, cocaine, crack, heroin, amphetamines/methamphetamines, injection drugs, drugs typically requiring a prescription “on their own”, and any other drug to get high or feel good. Frequency and intensity of use are thought to represent independent dimensions of AOD use behavior and tend to be relatively uncorrelated, making them a good choice as outcome measures [101]. We will derive a Quantity Frequency Index (QFI), which has a history of reliability and validity across various populations [102].

To measure HIV risk behaviors, we will ask about the respondent’s sexual behavior in the past 30-days overall and with particular partners named as alters. We will ask them how many times they had sex (vaginal or anal) and how often they used condoms when they had sex. We will also ask if they were engaging in sex with other partners concurrently. We have used these measures in studies of HIV risk among homeless women [43], youth [41], and men [93, 103].

To measure network changes, we will construct measures of how central various network members are using standard social network centrality measures [104, 105] similar to other studies of the social networks of homeless populations [6‐8, 41, 43, 76, 77]. For example, we will measure degree centrality for each network member which is a count of the number of people connected to that network member in the network. We will also measure centrality with other measures that have been found to be associated with unprotected sex in homeless populations, such as betweenness centrality (a continuous measure of how often an alter lies on the shortest path between pairs of other alters in the network) [40] and closeness centrality (a continuous measure of how directly or indirectly connected a particular alter is to all the other members of a network) [77]. We will also calculate measures of network composition and size to determine if the intervention is associated with changes in the types and quantities of people in the network. For example, we will calculate proportion of alters who are drug users, who are supportive, who are perceived to engage in risky sex, etc. We will also calculate measures of overall structural characteristics to determine if the intervention affected overall interaction among network members. For example, we will calculate density at both time points. Density is an index that represents the proportion of ties that exist in a network relative to the total number of possible ties, and varies from 0 to 1. We will also calculate centralization, which is another index between 0 and 1 that measures the degree to which one or a few individuals in the network maintain the majority of ties.

During the post-intervention assessment, MNI participants will answer 23 questions about the quality of and their satisfaction with the MNI sessions they attended on a five-point Likert scale, with a higher score representing higher quality and satisfaction (e.g., I feel that the things I did in the sessions will help me to make the changes that I want; The different activities we did in the sessions were helpful; The facilitator valued my opinion; The different activities we did in the sessions were helpful). Similar satisfaction questions have been used in prior research [106‐108].

Given the planned sample size (i.e., n = 60), sophisticated modeling and adjustments for non-response might not be possible, and the bulk of the analyses will be primarily descriptive. We will estimate the outcomes’ variability in this population and more generally assess the hypothesized trends to determine the intervention promise and identify potential mediators for future studies. Analyses will use the standard intent-to-treat (ITT) approach to examine the effect of offering the MNI intervention to eligible and consented participants. We will attempt to follow up with all participants, regardless of their MNI participation. Our ITT approach will analyze participants as belonging to the group they were randomized to, regardless of their compliance, because excluding those who do not complete the MNI would bias results in favor of MNI, increasing type I errors [109].

We will be able to obtain intervention effect estimates on both the proximal (i.e., efficacy and readiness to change AOD, safe-sex behavior, and networks) and distal outcomes (i.e., AOD use and HIV risk behavior) using a difference in differences (DID) approach [110]. The DID approach is well-suited to the data generated by the adopted study design: for every subject we will have a pre- and post-intervention observation and a randomized group indicator (intervention or control). We will implement the DID approach by fitting a mixed-effects model in which participants are treated as random effects while time, the intervention group indicator, and their interaction are treated as fixed effects. This modeling approach accounts properly for the two repeated measures’ correlation on each participant and tends to produce efficient intervention effect estimates [111]. The intervention effect estimate is given by the coefficient of the interaction between the time and intervention indicators.