Despite the contribution of Housing and Urban Development-Veterans Affairs Supported Housing (HUD-VASH) to ending Veterans' homelessness, research has shown that about 25% of veterans housed by HUD-VASH terminate the HUD-VASH program each year; and 19% return to homelessness and lose their voucher over a five-year period. A key predictor of housing loss was a return to substance abuse. This project addresses housing loss among veterans in HUD-VASH who have co-occurring substance abuse and mental illness by implementing a structured peer support service called Administering MISSION peer support (AMPS). The AMPS intervention can be readily transferred and can thus be used to improve veteran housing outcomes across the nation.
The study aimed to assess whether the addition of a peer support intervention to HUD-VASH services improves outcomes among Veterans at risk for housing loss due to substance use and mental illness. Hypotheses:
Compared to veterans receiving HUD-VASH treatment as usual (TAU), over a one year period veterans who receive AMPS will:
H1.1. have a reduction in the number of days of substance use: H.1.2. have more stable or improved overall mental health; H.1.3. have more outpatient visits of mental health and substance abuse services including self-help meetings;
H.2.1. have more days living in a voucher housing placement; H.2.2 show significantly greater activities reflecting community participation and improved social functioning. Additionally the project conducted a process evaluation, including semi-structured interviews and a cost tally.
AMPS randomized HUD-VASH dwelling Veterans into a treatment and control condition at two VA hospitals (Bedford MA and Pittsburgh PA). Eligibility criteria included active or history of substance use or dependence and a mental health diagnosis identified through VA records and/or as clinically judged and reported by a case manager. Study participants (N=167) completed up to 3 interviews using the following self-report measures: the BASIS-24, residential stability and homelessness interview, the Addiction Severity Index, and Temple University Community Participation Measure. Service utilization analysis used VA CPRS encounter records from the common data warehouse. For the process evaluation we randomly selected 20 Veterans, stratified by site and frequency of engagement, and 8 case managers to participate in an interview. The AMPS peer intervention was intended to last for 9 months to one year with forty total weekly sessions delivered in Veterans' homes and communities. One half of the sessions were intended to use the structured recovery oriented exercises and worksheets adapted from the MISSION-Vet Consumer workbook. The other half were "unstructured" sessions intended to respond to Veterans identified needs.
For outcome measures, we performed an intent-to-treat (ITT) analysis on data collected at three time points for each veteran using a generalized additive mixed effects model (GAMM) to describe responses over time (N=81 control, 85 treated). We determined the best-fit GAMM model to be one with a veteran-level random intercept and different nonlinear trends by study group, with a forced zero intercept at baseline. We adjusted for background covariates age, gender, education level, minority status, marital status, amount of social support, and number of employment activities in the past 30 days. We also included a fixed effect for specific peer and a fixed effect for site. We estimated treatment effect profiles over time for each outcome measure, and used a bootstrapping technique to generate 90% and 95% confidence envelopes for the estimated treatment effects. This allowed us to report the statistical significance of an estimated treatment effect at any point during the study period. Seven peers were involved in delivering services over the course of the study (n=4 at Pittsburgh, 3 at Bedford). Four peers (n=2 from each site) left the study early and were determined by study supervisors to not be able to consistently deliver services as AMPS intends. Thus, we also conducted a separate analysis of treated veterans who were assigned to one of the three peers who were able to deliver services as AMPS intends. We called this the "effective peers" analysis. These Veterans (N=39) were compared to all control Veterans (N=81).
Neither the ITT analyses, nor the "effective peers" analysis, detected statistically significant effects of AMPS for mental health, substance use, social support or overall community participation for any length of time at any point during the study observation. However, some significant effects were found for residential stability. For the ITT analysis measures for being housed in HUD-VASH within the last 30 days, as well as for being housed either in HUD-VASH dwellings or in other homes (e.g., family, friends), there were significant positive treatment effects at 90% confidence between roughly 600-800 days post-randomization and conversely fewer days in temporary, institutional or "other" dwellings. Similarly, the "effective peers" analysis showed significant effects (at 95% confidence level) for the same two housing stability measures between 400-800 days post randomization. Additionally, treated veterans also evidenced an increase in their confidence in managing tasks of daily living at 95% confidence interval roughly between 350 - 800 days post randomization.
An ad hoc analysis using linear mixed models to evaluate differences in the ratio of emergency room encounters to primary care encounters found that emergency room use increased in the intervention group relative to the control group although this was not statistically different.
The process evaluation showed that peer services were generally well received by Veterans and Case Managers. The program allowed the essential qualities of Peer Specialists to co-exist with a structured curriculum of recovery activities and exercises. Peers were able to provide genuine assistance while being accountable for building recovery skills from an evidence-based program.
The cost tally of AMPS peer services showed a large variability both in the cost/unit of service delivered ("effective peers" had an average cost per visit of $259 and others an average cost of over $1000/per visit). "Effective peers" accrued approximately 20 visits per month whereas others averaged less than 10 visits per month.
The findings of the positive impact of this relatively low cost innovation on housing stability for at-risk Veterans in HUD-VASH can contribute to further implementation of this model and to improved Veteran housing stability and reduced homelessness.
- Schutt RK, Ellison ML, Chinman M, Mitchell-Miland C, McCarthy S, Shah M, Schultz MR. Health service preferences among veterans in supported housing in relation to needs expressed and services used. Journal of mental health (Abingdon, England). 2019 Mar 12; 1-9.
- Chinman M, McCarthy S, Mitchell-Miland C, Bachrach RL, Schutt RK, Ellison M. Predicting Engagement With Mental Health Peer Specialist Services. Psychiatric services (Washington, D.C.). 2019 Apr 1; 70(4):333-336.
- Chinman M, McCarthy S, Bachrach RL, Mitchell-Miland C, Schutt RK, Ellison M. Investigating the Degree of Reliable Change Among Persons Assigned to Receive Mental Health Peer Specialist Services. Psychiatric services (Washington, D.C.). 2018 Dec 1; 69(12):1238-1244.
- Ellison ML, Schutt RK, Glickman ME, Schultz MR, Chinman M, Jensen K, Mitchell-Miland C, Smelson D, Eisen S. Patterns and predictors of engagement in peer support among homeless veterans with mental health conditions and substance use histories. Psychiatric Rehabilitation Journal. 2016 Sep 1; 39(3):266-73.
- Chinman MJ, Henze K. Integrating Peer Support Specialists into Clinical Practice. Poster session presented at: VA Psychology Leadership Conference; 2016 Jun 2; San Antonio, TX.