HSR&D Home » Research » RRP 12-259 – HSR&D Study
Improving the Continuum of Care for Homeless Veterans with Serious Mental Illness and Co-Occurring Disorders
Sonya Emi Gabrielian, MD MPH
VA Greater Los Angeles Healthcare System, West Los Angeles, CA
West Los Angeles, CA
Funding Period: August 2012 - July 2013
President Obama is committed to ending homelessness amongst Veterans. Though the VA offers many services for homeless Veterans with serious mental illness (SMI) and co-occurring substance use disorders (SUD), the majority of these services focus on basic needs for housing and healthcare. Little attention is given to deeper needs that grow important once people are housed, including social support and community involvement. As we build programs to facilitate exits from homelessness for Veterans with SMI and SUD, it is important to explore the relationships between their interpersonal/community connections and housing outcomes.
Some authors assert that vulnerability towards homelessness derives from deficits in accessing and/or mobilizing support systems. Yet, few have characterized factors that lie upstream of social support in homeless persons with SMI and SUD, i.e., the influence of psychiatric symptoms and cognition, on personal connections. Similarly, the relationships between social capital, community involvement, and housing outcomes are largely unexplored.
This study explored the community integration of Veterans with a history of homelessness, SMI, and SUD who were admitted to the West Los Angeles VA Domiciliary (Dom), a 296-bed residential rehabilitation and treatment program that aims to facilitate exits from homelessness. We conceptualized community integration in two domains: community participation -- or involvement in civic life, outside institutional settings -- and social capital, or interpersonal relationships. More specifically, for Veterans with a history of homelessness, SMI, and SUD who were admitted to the Dom over a three-year period, we aimed to describe the associations between community integration, patient characteristics (psychiatric symptoms, substance use, cognition, health service use) and housing outcomes.
We used administrative data to identify Veterans who were admitted to the Dom between December 2008 and November 2011. Using ICD-9 codes, we narrowed this list to Veterans (n=710) who received VA mental health care for at least one SMI (depression, bipolar illness, psychotic disorder, PTSD) and SUD (abuse or dependence of any substance, except nicotine) diagnosis within a year of their Dom admission date. We used letters and phone calls to approach 196 Veterans randomly selected from this list, using medical record review to confirm that each subject was homeless at Dom admission. Of these Veterans, n=36 agreed to an in-person assessment that included an event history of housing arrangements since exiting the Dom and surveys to assess their community integration, psychiatric symptoms, substance use, cognition, and health service use. We used the medical record to gather diagnostic information and data about the chronicity of each subject's homelessness. Of these subjects, a convenience sample of n=11 also completed qualitative interviews to explore the role of interpersonal relationships in transitions between housing arrangements.
We used the chi-square test and ANOVA to determine how community integration and patient characteristics varied among Veterans in three groups: 1) Veterans who achieved stable housing without VA housing programs; 2) Veterans who lacked stable housing outside VA housing programs; and 3) Veterans who remained continuously enrolled in VA housing programs. Age was included as a covariate when comparing cognitive outcomes. We used a recursive partitioning based analysis to identify the best predictive subset of community integration and patient characteristic variables on these housing outcomes. To complement analyses of cognitive outcomes, we identified examples of problem-solving described by Veterans in qualitative interviews.
The event histories revealed that some Veterans (n=14) gained stable housing on their own and others (n=15) spent >70% of their days outside a VA housing program without stable housing. The smallest group (n=7) remained continuously enrolled in VA housing programs. On average, the Veterans continuously enrolled in VA housing programs were older (60 years) than Veterans who achieved stable housing (50 years) or their peers who remained unstable (52 years) (p=0.04). Other demographic and community integration variables were not significantly different among the three groups. Incorporating age as a covariant, the most salient differences were seen in cognition. The symbol digit modalities test (SDMT), a measure of neurocognition, was significantly different (p=0.02) among the groups, with mean z-scores of 0.49 in the stable group, 0.05 in the unstable group, and -1.09 in the group that remained in VA programs. The Trail Making Test B, which measures processing speed and executive functioning, was suggestive of differences between the three groups (p=0.06), with mean z-scores of 0.04 in the stable group, 0.26 in the unstable group, and -0.40 in the group that remained in VA programs. The qualitative data provided rich examples of problem-solving skills to complement these findings.
In recursive partitioning, using 26 measures of community integration and patient characteristics collected for each Veteran as potential predictors, two measures were needed to characterize the three subsets of Veterans by housing outcomes: SDMT and the BASIS-interpersonal relationships subscale. The group predicted to remain permanently in VA housing programs had a raw SDMT score of <32.5 (z-score < -0.63). The group predicted to attain stable housing had a raw SDMT score of 32.5 and a raw BASIS-interpersonal relationships score of <0.81 (z score < -0.29). The group that was unable to achieve stable housing outside of VA housing programs had a SDMT score of 32.5 and a raw BASIS-interpersonal relationships score of 0.81. That is, Veterans with worse cognition tended to remain continuously in VA housing programs. Among Veterans with better cognition, interpersonal difficulties predicted unstable housing. The relative error was 0.43, i.e., this model explains 57% of the variance in this sample. The cross-validated standard error is 0.86, i.e., this model is estimated to explain at least 14% of the variability in the population.
Among homeless Veterans with SMI and SUD, these data highlight the role of cognition and interpersonal skills in housing outcomes. Future studies will explore using the SDMT as a screening tool at admission to VA housing programs. Veterans with lower scores may experience difficulties achieving stable housing without programmatic support and thus require intensive services. Veterans with higher scores may benefit from social skills training to improve their interpersonal relationships, and in turn, their housing outcomes.
External Links for this Project
Dimensions for VADimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.
If you have VA-Intranet access, click here for more information vaww.hsrd.research.va.gov/dimensions/
VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address. Search Dimensions for this project
DRA: Mental, Cognitive and Behavioral Disorders
DRE: Treatment - Observational
MeSH Terms: none