About half of homeless Veterans have serious mental illness (SMI). Prior work from our group established a three-tier hierarchy of longitudinal housing outcomes for homeless Veterans with SMI: stable and independent housing; sheltered housing (e.g., long-term engagement in residential treatment programs); and street homelessness. To improve housing outcomes for Veterans with SMI who are experiencing homelessness, we must identify factors strongly associated with sustaining stable and independent housing and reorganize VA homeless services to include interventions that address these factors.
We used the Behavioral Model for Vulnerable Populations for these specific aims: 1) To identify predisposing variables, needs, and behaviors that predict group membership in three longitudinal housing outcomes among homeless Veterans with SMI: stable and independent housing, sheltered housing, and street homelessness. We hypothesized that mental health neurocognition and social cognition (predisposing variables), along with mental health symptoms (need variables) strongly predict group membership in these housing outcomes; and 2) To inform the reorganization of VA homeless services to include interventions that address factors that predict longitudinal housing outcomes among homeless Veterans with SMI.
We collected two-years' of retrospective housing history from homeless Veterans with SMI (N=86) at the VA Greater Los Angeles. Using an adapted version of the Residential Time-Line Follow Back inventory, we classified each Veteran into one of three longitudinal housing outcomes, identifying the setting he/she lived in for >50% of the time over the past two years: 1) stable, independent housing (n=28); 2) sheltered housing (n=29); and 3) street homelessness (n=29). We reviewed Veterans' medical records and conducted in-person assessments-including comprehensive measures of neurocognition and social cognition-to identify predictors of group membership in housing outcomes. Classification and regression trees (CART) were used to identify the best predictive subset of 46 variables (demographics, VA health and social service utilization, psychiatric and substance use disorder diagnoses, mental health symptoms, social support) that classified Veterans into the three housing groups. We also performed semi-structured interviews with 7 Veterans per housing group (n=21); interviews explored Veterans' housing-related problems, strategies used to solve problems, and money management skills. Thematic analyses were performed on qualitative data.
Three measures-number of mental health visits over the past two years, visuospatial memory, and monthly income-were sufficient to capture information provided by all 46 potential predictor variables to classify Veterans by housing outcomes. Veterans with 15 outpatient mental health (psychiatry or psychology) visits over the past two years were predicted to be in sheltered housing. Among Veterans with 14 outpatient mental health visits, those who scored highly on visuospatial memory (brief visuospatial memory test-revised score 19.5) were also predicted to be in sheltered housing. Those with lower scores visuospatial memory scores were predicted to be street homeless if their income was less than $221.50/month. Persons with higher incomes were predicted to remain in stable housing. This model correctly classifies 62.7% of participants in our sample; it explains 44% of variance in this sample (relative error=0.56). Based on cross-validation, it also explains 14% of variability in housing outcomes from any other sample selected from the at large population of homeless Veterans with SMI (cross-classified standard error=0.86). Qualitative data showed that housing-related problems for the street homelessness group often stemmed from substance use disorders; this group seemed to rely heavily on VA case managers to solve their housing-related problems. The stable, independent housing and sheltered housing groups demonstrated more planning and forethought in their problem-solving strategies than their peers who were street homelessness. The stable, independent housing group described the fewest problems with money management.
Though the VA has many services that address the housing, physical health, and mental health care needs of homeless Veterans with SMI, we lack information to identify if cognition influences housing outcomes for this vulnerable population. Though cognition is a strong predictor of community functioning among persons with SMI, these data suggest that, contrary to our hypotheses, most domains of cognition and mental health symptoms do not predict housing outcomes in this vulnerable population. Linkages to vocational and Veterans' benefits services (to address income needs) and engagement in mental health care may be critical to improve housing outcomes for homeless Veterans with SMI.
- Gabrielian S, Hellemann G, Koosis ER, Green MF, Young AS. Do cognition and other person-level characteristics determine housing outcomes among homeless-experienced adults with serious mental illness? Psychiatric Rehabilitation Journal. 2020 Oct 12.
- Riggs KR, Hoge AE, DeRussy AJ, Montgomery AE, Holmes SK, Austin EL, Pollio DE, Kim YI, Varley AL, Gelberg L, Gabrielian SE, Blosnich JR, Merlin J, Gundlapalli AV, Jones AL, Gordon AJ, Kertesz SG. Prevalence of and Risk Factors Associated With Nonfatal Overdose Among Veterans Who Have Experienced Homelessness. JAMA Network Open. 2020 Mar 2; 3(3):e201190.
- Montgomery AE, Cusack M, Blonigen DM, Gabrielian S, Marsh L, Fargo J. Factors Associated With Veterans' Access to Permanent Supportive Housing. Psychiatric services (Washington, D.C.). 2016 Aug 1; 67(8):870-7.
- Montgomery AE, Cusack M, Blonigen DM, Gabrielian SE, Marsh L, Fargo J. Factors associated with Veterans’ access to permanent supportive housing. Paper presented at: VA HSR&D Field-Based Meeting to Engage Diverse Stakeholders and Operational Partners in Advancing Health Equity in the VA Healthcare System; 2016 Sep 20; Philadelphia, PA.
Mental, Cognitive and Behavioral Disorders
Treatment - Observational