Skip to main content
NCBI home page
VA Author Manuscripts logoLink to VA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jun 1.
Published in final edited form as: Psychiatr Rehabil J. 2020 Oct 12;44(2):176–185. doi: 10.1037/prj0000457

Do cognition and other person-level characteristics determine housing outcomes among homeless-experienced adults with serious mental illness?

Sonya Gabrielian 1, Gerhard Hellemann 1, Ella R Koosis 1, Michael F Green 1, Alexander S Young 1
PMCID: PMC8435461  NIHMSID: NIHMS1732561  PMID: 33048564

Abstract

Objective

Many persons with serious mental illness (SMI) who have experienced homelessness struggle to sustain stable and independent housing. We know little about determinants of this population’s housing status. This study aimed to identify person-level determinants of housing status among homeless-experienced Veterans with SMI, focused primarily on cognition.

Methods

We administered cross-sectional surveys and detailed cognitive assessments on a convenience sample of homeless-experienced Veterans with SMI (n=90); we also reviewed these participants’ medical records. We captured person-level potential predictors of housing status (demographics, cognition, diagnoses, symptoms, and service utilization) and two years of retrospective housing history. Participants’ housing status was conceptualized as the setting (stable housing, other sheltered settings, streets) they lived in for >50% of the past two years. We used the chi-square test and analysis of variance to determine how potential predictors differed by housing status. We used recursive partitioning to identify the combination of potential predictors and corresponding scores that best-differentiated participants by housing status.

Results

No between-group differences (p<.05) in cognition, symptoms, or other person-level factors were found among participants grouped by housing status. Recursive partitioning did not yield a stable model to predict housing status from the potential predictor variables.

Conclusions and Implications for Practice

These data suggest that clinical interventions addressing studied person-level factors (e.g., cognitive rehabilitation) may not affect housing status for homeless-experienced Veterans. As housing is highly influenced by social determinants of health, policies and practices that affect contextual factors (e.g., affordable housing supply) may be more likely to improve housing status.

Keywords: homelessness, serious mental illness, cognition, Veterans

Impact and Implications

Many persons with serious mental illness who have experienced homelessness struggle to attain and retain housing. Though cognition and other person-level characteristics predict work and social outcomes for this group, these factors were not associated with housing status in this study of homeless-experienced Veterans with serious mental illness. Our data suggest that clinical interventions that improve cognition and functioning may not enhance housing outcomes for this vulnerable population.

In the United States, deinstitutionalization contributed to a surge of homelessness among persons with serious mental illness (SMI). (Kuno, Rothbard, Averyt, & Culhane, 2000; Lamb & Bachrach, 2001) Unfortunately, adults with SMI are 10–20 times more likely to become homeless than the general population. (Kuno et al., 2000) A breadth of services effectively improves health and housing for persons who have experienced homelessness (hereafter, “homeless-experienced persons”) and have SMI diagnoses, including assertive community treatment, substance use disorder services, and supportive housing. (Fitzpatrick-Lewis et al., 2011; Hwang & Burns, 2014) Yet, even with these services, many homeless-experienced persons with SMI struggle to sustain stable and independent housing. (Padgett, Smith, Henwood, & Tiderington, 2012; Patterson, Somers, & Moniruzzaman, 2012) Persons in this vulnerable group often vacillate between residing in stable housing, other sheltered settings, or on the streets. (Gabrielian et al., 2015) We know strikingly little about determinants of housing status among homeless-experienced persons with SMI.

Though few have studied person-level factors associated with housing status in this vulnerable population, there is robust literature on determinants of other functional outcomes among adults with SMI. In particular, cognition–including nonsocial (attention, memory, executive functioning, processing speed) and social cognition (mental processes that underlie social interactions)–is highly predictive of work and social functioning among adults with SMI. (Barbato et al., 2013; Green, 1996; Green, Kern, Braff, & Mintz, 2000) Among persons with SMI who have experienced homelessness, cognition seems highly relevant to one’s ability to navigate systems of housing and health services. The literature suggests that cognitive dysfunction is common among homeless-experienced persons; a review article estimates the frequency of cognitive impairment among homeless-experienced persons at 25% (Depp, Vella, Orff, & Twamley, 2015) and a study of 1500 homeless adults with mental illness found that 72% of these adults had cognitive impairment. (Stergiopoulos et al., 2015) Yet, despite these figures, the role of cognition as a predictor of housing status is relatively unexplored in this vulnerable population.

Here, we conceptualize SMI broadly, including schizophrenia spectrum and other psychotic disorders, bipolar illness, chronic post-traumatic stress disorder (PTSD), and major depressive disorder. (SAMHSA’s Co-Occurring Center for Excellence, 2006) Though the relationships between cognition and functional outcomes are best understood among persons with psychotic disorders (Fett et al., 2011) and bipolar illness, (Goldberg & Chengappa, 2009) several studies show that cognition is strongly correlated with disability across psychiatric diagnoses, including depression (Bora, Harrison, Yücel, & Pantelis, 2013; McIntyre et al., 2013; Roca et al., 2015) and PTSD. (Kaye et al., 2014; Vasterling et al., 2002) Among persons with SMI, cognitive deficits are evident at illness onset and stable over time. (Meyer & Kurtz, 2009; Pinkham, 2014)

If cognition is a determinant of housing status among homeless-experienced persons with SMI, there are effective interventions that improve cognition and functioning. (Paquin, Wilson, Cellard, Lecomte, & Potvin, 2014; Wykes, Huddy, Cellard, McGurk, & Czobor, 2011) Interventions for nonsocial cognition have two modalities: a “restorative approach” that use repetitive training (often on the computer) to improve functioning; and “compensatory” rehabilitation, which compensates for impairments with aids (e.g., notebooks or calendars) or by recruiting intact areas of cognition. (Kern et al., 2005; Leshner, Tom, & Kern, 2013) Compensatory approaches effectively teach entry-level job tasks and social-problem solving skills. (Kern et al., 2005) Similarly, social cognitive training interventions often break down social cognitive processes into component skills, training persons to be “social detectives” who, for example, can detect misalignment between vocal tone and the literal meaning of what a peer says (i.e., sarcasm). (Horan et al., 2009)

Should cognition prove relevant for housing status, effective cognitive training paradigms could be tailored for homeless-experienced individuals and embedded in the context of evidence-based housing services. Conversely, finding that cognition does not determine long-term housing status would be informative, suggesting that housing has different determinants than other functional outcomes in this population.

The Department of Veterans Affairs (VA) is an ideal setting for this line of research. The VA is the nation’s largest provider of health and housing services for homeless-experienced persons with SMI and has broadly implemented evidence-based practices for this vulnerable group. (“FY 2013 Budget: Housing and Communities Built to Last,” 2012) Homeless-experienced Veterans benefit from enriched services, with less fragmentation of social and health services than other safety net systems of care. (N. J. Wilson & Kizer, 1997) This paper describes a preliminary study that sought to understand the degree to which nonsocial cognition, social cognition, and other person-level attributes are associated with concurrent housing status among homeless-experienced Veterans with SMI. We focus on Veterans with a history of homelessness; the relationships between cognitive functioning and risk factors for new-onset homelessness are beyond the scope of this study.

Conceptual Framework

This research was guided by the Behavioral Model for Vulnerable Populations, (Gelberg, Andersen, & Leake, 2000) an adaptation of the Andersen model (Andersen, 1968; 1995; Andersen & Davidson, 2007) that includes domains aligned to the distinct vulnerabilities of homeless-experienced persons. This framework models the interplay of contextual and person-level factors that align with the health and psychosocial circumstances of vulnerable persons. As this inquiry was focused on person-level attributes, we assessed factors that predispose individuals to access services (demographics, legal history), which interact with enabling factors (nonsocial and social cognition, income, case management, social supports) and needs (perceived by patients and providers) to influence behaviors (service use) and housing status.

Methods

Participants

This study was conducted at the VA Greater Los Angeles (GLA), which serves more homeless-experienced Veterans than any VA in the nation, including 7,449 unique individuals in fiscal year 2018. First, we queried VA administrative data (the Corporate Data Warehouse (CDW)) (Price, Shea, & Gephart, 2015) to identify persons served at GLA who: 1) were 18–70 years of age (to reduce the influence of age-related cognitive deficits); 2) received an ICD9 or ICD10 code for homelessness from 2014–2016 (indicating that they had experienced homelessness over this time frame); and 3) received an ICD9 or ICD10 code for an SMI diagnosis in a mental health setting from 2014–2016. These criteria generated a roster of 6,616 persons. For this preliminary study examining cognition and other person-level attributes, we aimed to recruit 90 persons (divided equally into three groups defined by predominant housing status over the past two years) for detailed research assessments.

First, we sent letters and telephone calls (to 335 persons randomly selected from the roster) to recruit individuals from this list for data collection, anticipating that these methods would primarily recruit previously homeless persons who had achieved stable, independent housing. However, we found that addresses and telephone numbers in CDW were frequently inaccurate; these methods recruited only 7 persons.

Next, we identified subjects from GLA’s homeless services, including residential treatment programs, medical clinics tailored for homeless-experienced individuals, (Gabrielian et al., 2014) and a social service walk-in center. (O’Toole & Pape, 2015) Program staff referred interested persons (n=92) who met age and diagnostic eligibility criteria for research assessments.

Persons interested in study participation (n=99) were asked to self-report a history of neurologic illness (including seizures unrelated to substance use, stroke, coma, or neurodegenerative disorders) that could influence cognition. This self-report was confirmed by review of VA’s electronic health record (EHR). We also administered the Brief Traumatic Injury Screen (Schwab et al., 2007) to screen out Veterans with traumatic brain injury, given this diagnosis’ effects on cognitive functioning. Nine persons were deemed ineligible due to neurologic diagnoses or the traumatic brain injury screen. One-time, cross-sectional research assessments were administered to a sample of 90 persons. The GLA Institutional Review Board approved all study procedures; we obtained verbal informed consent using a rigorous procedure studied in persons with SMI, including a questionnaire relating to the research protocol (D. A. Wirshing, Wirshing, Marder, Liberman, & Mintz, 1998) and an opportunity to withdraw from the study if desired.

Measures

Predisposing factors.

Demographics (age, gender, race/ethnicity, and marital status) were abstracted from the VA’s EHR. Participants were asked to self-report their highest level of education. To capture legal history, participants were queried as to their lifetime felony history (yes/no); lifetime history of any incarceration (yes/no); and asked to estimate their lifetime total duration of incarceration.

Enabling factors.

Though we are unaware of cognitive measures that are specifically validated for homeless persons, we used measures that were employed in several other studies of homeless persons with SMI. (Llerena, Gabrielian, & Green, 2018; Wynn et al., 2020) Specifically, nonsocial cognition was measured with the MATRICS Consensus Cognitive Battery (MCCB), (Green, Kern, & Heaton, 2004; Nuechterlein & Green, 2006) a detailed (~60 minutes/participant), consensus-derived set of measures that is widely used to measure cognition among persons with SMI. (Bo et al., 2017; Kumar et al., 2016; Nitzburg et al., 2014; Rodriguez-Jimenez et al., 2019) The MCCB was used to assess participants’ overall cognition; subscores were also generated in six distinct domains: speed of processing; attention/vigilance; working memory; verbal learning; visual learning; and reasoning and problem solving. T-scores for each domain and the composite score were derived from established age and gender norms for this instrument. (Kern et al., 2008) Though the MCCB includes a test of social cognition, it was omitted as it is limited to the social cognitive domain of emotion management.

Instead, social cognition was measured with the Empathic Accuracy Task, (Lee, Zaki, Harvey, Ochsner, & Green, 2011; Zaki & Ochsner, 2009; Zaki, Bolger, & Ochsner, 2008) a validated instrument that captures a person’s ability to integrate the mental processes needed for empathy. This task is a representative instrument of social cognition across a range of social processing domains; similar measures of empathic accuracy are comparable across genders, (Klein & Hodges, 2001) ethnicities, (Roberts & Levenson, 2006) and positively associated with social adjustment and relationship satisfaction. (Ripoll et al., 2013)

In addition, participants were asked if they have a case manager (yes/no, from a housing or mental health program); persons who identified a case manager were queried as to the number of case manager visits/month they had on average over the last two years. Monthly income was self-reported by participants. Social support was assessed using the Medical Outcomes Study-Social Support Survey (MOS-SSS), 4-item version. (Sherbourne & Stewart, 1991)

Needs.

To capture participants’ mental health diagnoses, we reviewed mental health notes in the EHR from 2014–2016, noting psychiatric diagnoses associated with these notes. Mental health symptoms were measured with the 24-item Behavior and Symptom Identification scale (BASIS-24), a measure of self-reported difficulty in six domains (depression/functioning, interpersonal relations, psychotic symptoms, alcohol/drug use, emotional lability, and self-harm) that also provides a global symptom assessment. (Idiculla, Speredelozzi, & Miller, 2005) The Substance Use Brief Screen (SUBS), (McNeely et al., 2015) a validated screening tool for unhealthy alcohol, illicit drug, and prescription drug use, was used to assess for the presence vs. absence of problematic substance use.

Behaviors.

Use of health services was assessed from the EHR, covering two years prior to the study assessment date. Specifically, we captured participants’ number of primary care visits, outpatient mental health visits, and Emergency Department visits. We also identified whether or not participants who screened positive on the SUBS were engaged in substance use disorder treatment (residential or outpatient). For all participants, we abstracted the presence vs. absence of at least one medical/surgical and/or mental health hospitalization over the past two years.

With regards to housing services used in the two years prior to study assessment, we captured whether or not participants were engaged in the Department of Housing and Urban Affairs-VA’s Supportive Housing program (HUD-VASH, the VA’s largest service for homeless Veterans that offers independent, permanent housing, supportive services, and non-mandated linkages to health services). For persons enrolled in HUD-VASH, we captured the duration of engagement; as some HUD-VASH participants exit the program prematurely, before attaining housing, we also captured whether or not participants attained housing through the program.

Outcomes.

Housing history was captured with the Residential Time-Line Follow Back (TLFB) Inventory, (Tsemberis, McHugo, Williams, Hanrahan, & Stefancic, 2006) which gathers a retrospective event history of an individual’s residences. The TLFB was administered prior to participants’ selection for this study, as we aimed to recruit approximately equal groups by predominant housing status.

Participants reported two-years of retrospective residential history; study eligibility criteria assured that each participant had experienced homelessness at some point during this time frame. Characterizing housing history over two-year periods has precedent in several other studies of persons who have experienced homelessness. (Lim, Singh, Hall, Walters, & Gould, 2018; Rosenthal et al., 2007)

Each housing type reported was classified as “stable,” “temporary,” “institutional,” or “street homelessness.” Stable settings include individuals’ own apartments (obtained with or without the help of HUD-VASH or another housing program) or apartments of family or friends. Temporary and institutional settings, e.g., transitional housing and residential rehabilitation, were merged under the umbrella of “sheltered housing,” as many VA housing services combine transitional housing with rehabilitation. Street homelessness included settings like cars, abandoned buildings, or other places not intended for human habitation.

We classified each participant by the housing status (stable, sheltered, or street) in which he/she spent >50% of time over the past two years; participant recruitment aimed to identify a similar number of participants in each of these groups. We viewed this outcome as categorical, not continuous, given substantive differences in types of housing that are more clinically relevant than the percent of days in stable housing. (Gabrielian et al., 2015) Four participants (4.4%) did not spend >50% of their time in one setting; these participants were removed from our analyses, resulting in an analytic sample of n=86, relatively evenly divided between the stable housing, sheltered housing, and street homeless groups (n=29/29/28).

Analyses

First, we used the chi-square test and analysis of variance (ANOVA) to determine how predisposing, enabling, need, and behavioral factors differed by each of the three categorical housing outcomes. These analyses were performed using State/SE software version 12.1. (Stata Statistical Software: Release 12StataCorp, n.d.)

Second, we used recursive partitioning (Zhang & Singer, n.d.) to identify the combination of potential predictors and corresponding scores that best differentiated participants by the three housing status categories. Recursive partitioning is a data mining technique that uses “decision trees” to predict outcomes from a group of predictor variables. (Hellemann, Conner, Anglin, & Longshore, 2008; Strobl, Malley, & Tutz, 2009) This analytic technique is better suited than regression for studies with large numbers of predictors in comparison to sample size. Moreover, it accounts for potential interactions among predictor variables, including multivariable relationships and moderating effects, and non-linear effects of included predictors. (Hellemann et al., 2008; Strobl et al., 2009)

For this analysis with 44 potential predictors (across predisposing, enabling, need and behavioral variables, Supplemental Table 1) on a sample of n=86, recursive partitioning independently evaluates each potential predictor on the categorical outcome variable. The variable and its corresponding cutpoint (or value) that best split the data into subsamples by housing outcome is selected as the first predictor, or the first two “branches” of a classification and regression tree (CART). (Hellemann et al., 2008; Strobl et al., 2009) The process is subsequently repeated on each of the created subsamples, again identifying the variable and its value that best predicts the most homogenous subsamples within each previously formed branch. Branching continues until there is no further improvement in correct differentiation of participants by housing outcome.

The recursive partitioning approach aimed to simplify the complex set of 44 potential predictor variables into a few simple “if-then” rules that predicted outcomes. (Hellemann et al., 2008; Strobl et al., 2009) A 10-fold cross validation approach (Kohavi, 1995) was used to identify if predictors identified were likely generalizable to the underlying population of persons with SMI who had experienced homelessness (as opposed to overfitting). Analyses were performed using the recursive partitioning algorithm in the rpart package version 3.1–33 for the R language and environment. (Therneau & Atkinson, n.d.)

Results

Participants were stratified by the housing status in which they resided for >50% of days in the two years prior to assessment. Over these two years, the stable housing group (n=29) resided a mean/standard deviation (SD) of 83.0%/15.1% days in stable housing; the sheltered housing (n=29) and street homeless (n=28) groups resided a mean/SD of 75.8%/18.3% and 77.5%/13.2% of days, respectively, in their predominant residential setting.

Table 1 describes the three groups across predisposing, enabling, need, and behavioral variables. Across groups, no statistically significant differences were found among predisposing factors (all p>.05). Most participants (87.2% of sample) were male, with a mean age of 51.8 years. The largest percentage of participants was African American (50.0%), followed by White (29.1%) and Hispanic (17.4%). Across all groups, few participants were married (4.7% of sample). Educational attainment was similar across housing outcomes, ranging from 12.2–12.8 years. Most participants (79.1%) had a history of criminal justice involvement, with nearly half (48.8%) identifying a history of at least one felony charge. Among participants who had experienced incarceration, mean lifetime incarceration was substantive (5.3 years).

Table 1.

Predisposing, Enabling, Need, and Behavioral Variables, by Predominant Housing Status over the Past 2 Years

Stable (n=29, 33.7%) Sheltered (n=29, 33.7%) Street (n=28, 32.6%) Total (N=86) p-value
Predisposing factors
Age in years (mean, SD) 54.4, 9.2 51.1, 11.8 49.9, 12.5 51.8, 11.3 0.29
Gender (n, % male) 25, 86.2% 25, 86.2% 25, 89.3% 75, 87.2% 0.92
Race/Ethnicity (n, %) 0.63
 Hispanic, any race 4, 13.8% 5, 17.2% 6, 21.4% 15, 17.4%
 White 6, 20.7% 10, 34.5% 9, 32.1% 25, 29.1%
 Black 18, 62.1% 14, 48.3% 11, 39.3% 43, 50.0%
 Asian 1, 3.5% 0, 0.0% 1, 3.6% 2, 2.3%
 Decline to state 0, 0.0% 0, 0.0% 1, 3.6% 1, 1.2%
Marital status (n, %) 0.07
 Never married 5, 17.2% 10, 34.5% 14, 50.0% 29, 33.7%
 Married 3, 10.3% 0, 0.0% 1, 3.6% 4, 4.7%
 Separated 3, 10.3% 6, 20.7% 2, 7.1% 11, 12.8%
 Divorced 14, 48.3% 10, 34.5% 11, 39.3% 35, 40.7%
 Widowed 4, 13.8% 3, 10.3% 0, 0.0% 7, 8.1%
Highest level of education in years (mean, SD) 12.3, 1.5 12.2, 1.2 12.8, 2.9 12.4, 2.0 0.49
Legal history
 Lifetime felony history (n, %) 17, 58.6% 12, 41.4% 13, 46.6% 42, 48.8% 0.40
 Lifetime history of incarceration (n, %) 24, 82.8% 25, 86.2% 19, 67.9% 68, 79.1% 0.20
 Lifetime duration incarcerationa, in years (mean, SD) 5.8, 7.3 4.6, 7.4 5.7, 6.7 5.3, 7.1 0.82
Enabling factors
Nonsocial cognition (MCCB, mean, SD)b
 Speed of processing 47.0, 12.0 46.7, 9.9 48.1, 10.9 47.3, 10.9 0.88
 Attention/vigilance 45.1, 10.1 42.9, 10.8 46.1, 10.0 44.7, 10.3 0.49
 Working memory 42.0, 11.6 43.5, 10.6 40.1, 12.4 41.9, 11.5 0.54
 Verbal learning 42.1, 10.0 44.1, 6.8 40.3, 8.9 42.2, 8.7 0.26
 Visual learning 44.1, 12.3 47.7,12.9 45.6, 13.0 45.8, 12.7 0.56
 Reasoning and problem solving 49.7, 9.7 49.4, 9.1 51.8, 11.6 50.25, 10.1 0.63
 Composite 45.0, 8.1 45.7, 7.2 45.3, 8.2 45.3, 7.8 0.94
Social cognition (Empathic Accuracy Task) 0.55, 0.20 0.53, 0.23 0.46, 0.18 0.51, 0.20 0.31
Case manager assignment identified (n, %) 26, 89.7% 28, 96.6% 23, 82.1% 77, 89.5% 0.21
Case manager visits/monthc* 1.1, 1.1 4.3, 7.44 2.1, 2.2 2.6, 4.8 0.04
Social support (MOS-SSS) 11.2, 4.7 9.7, 3.8 10.8, 4.5 10.6, 4.4 0.37
Needs
Psychiatric diagnoses (n, %)
 Depression 19, 65.5% 17, 58.6% 16, 57.1% 52, 60.5% 0.79
 Bipolar disorder 4, 13.8% 9, 31.0% 4, 14.3% 17, 19.8% 0.17
 Psychotic disorder 8, 27.6% 8, 27.6% 6, 21.4% 22, 25.6% 0.83
 PTSD 13, 44.8% 12, 41.4% 13, 46.6% 38, 44.2% 0.93
 Other anxiety disorder 2, 6.9% 7, 24.1% 4, 14.3% 13, 15.1% 0.18
Mental health symptoms (BASIS-24, mean, SD)
 Depression and functioning 1.7, 0.9 1.8, 0.7 1.9, 0.8 1.8, 0.8 0.61
 Symptoms affecting relationships 1.9, 0.8 1.6, 0.7 1.7, 0.9 1.7, 0.8 0.55
 Self-harm 0.1, 0.3 0.2, 0.6 0.4, 0.5 0.2, 0.5 0.12
 Emotional lability 1.7, 0.8 1.7, 0.9 1.8, 0.7 1.7, 0.8 0.85
 Psychosis 1.4, 1.1 1.2, 0.9 1.3, 0.9 1.3, 0.9 0.77
 Substance abuse 0.9, 0.7 1.0, 0.7 1.1, 0.8 1.0, 0.7 0.76
 Composite 1.5, 0.7 1.5, 0.5 1.6, 0.6 1.5, 0.6 0.64
Problematic substance use (positive SUBS screen, n, %)
 Alcohol 18, 62.1% 18, 62.1% 20, 72.4% 56, 65.1% 0.70
 Illicit drugs 24, 82.8% 17, 58.6% 21, 75.0% 62, 72.1% 0.11
 Prescription drugs 6, 20.7% 2, 6.9% 9, 32.1% 17, 19.8% 0.06
Behaviors
Health service use (two years prior to assessment)
 # primary care visits (mean, SD) 8.3, 6.8 9.2, 5.7 6.8, 4.4 8.1, 5.8 0.26
 # mental health visits (mean, SD)* 8.7, 9.4 15.1, 11.6 7.5, 8.5 10.5, 10.4 0.01
 # Emergency Department visits (mean, SD) 1.2, 2.5 2.7, 4.1 2.9, 3.4 2.2, 3.4 0.13
 Outpatient substance use disorder treatment used (n, %) 1, 3.9% 2, 9.5% 1, 3.9% 4, 5.5% 0.63
 Residential substance use disorder treatment used (n, %) 22, 84.6% 19, 90.5% 22, 84.6% 63, 86.3% 0.81
 ≥1 medical/surgical hospitalization (n, %) 3, 10.3% 4, 13.8% 4, 14.3% 11, 12.8% 0.89
 ≥1 mental health hospitalization (n, %) 3, 10.3% 4, 13.8% 6, 21.4% 13, 15.1% 0.49
Housing service use (two years prior to assessment)
 HUD-VASH engagement (n, %) 12, 41.4% 18, 62.1% 13, 46.4% 43, 50.0% 0.27
 Duration of HUD-VASH engagement in monthse* (mean, SD) 20.6, 5.3 8.6, 7.1 11.1, 7.3 12.7, 8.3 0.01
 Achieved housing via HUD-VASHe* (n, %) 11, 91.7% 6, 33.3% 2, 15.4% 19, 44.2% 0.00
Open in a new tab

SD, standard deviation; MCCB, MATRICS Consensus Cognitive Battery; MOS-SSS, Medical Outcomes Study-Social Support Survey; PTSD, post-traumatic stress disorder; BASIS-24, 24-item Behavior and Symptom Identification Scale; SUBS, Substance Use Brief Screen; HUD-VASH, Department of Housing and Urban Development-VA Supportive Housing

*

p<.05

a

Lifetime duration of incarceration presented only for a subset of participants who had a history of incarceration, including 24 stable participants; 25 sheltered participants; 19 street participants; and 68 participants across all groups

b

T-Scores, corrected for age/gender norms, are presented for the MCCB

c

Case manager visits/month presented only for participants who identified case manager assignment, including 26 stable participants; 28 sheltered participants; and 23 street participants, and 77 participants across all groups

d

Substance use disorder treatment rates presented only among persons with positive SUBS screen, including 26 stable participants; 28 sheltered participants; 26 street participants, and 73 participants across all groups

e

Duration of HUD-VASH engagement and housing achievement in HUD-VASH presented only among persons engaged in HUD-VASH, including 12 stable participants; 18 sheltered participants; 13 street participants; and 43 participants across all groups

With regards to enabling factors, nonsocial (across six domains and the composite MCCB score) and social cognition were very similar across groups. Most participants (89.5%) identified a case manager; though more persons in the sheltered group (96.6%) had case management than their stably housed (89.7%) and street homeless (82.1%) counterparts, these differences were not statistically significant (p>.05). However, between-group differences in the frequency of case manager visits were significant (p<.05), with the highest mean visits/month in the sheltered group (4.3), followed by the street homeless (2.1), then the stably housed (1.1). Social support was similar across groups (p>.05).

In the domain of needs, no significant (p<.05) between-group differences were seen in psychiatric diagnoses, mental health symptoms, or screening results for problematic substance abuse. Most participants (60.5%) had depressive disorders; many (44.2%) had PTSD or psychotic disorders (25.6%). As expected with EHR abstraction on persons with SMI, many participants were labeled with multiple psychiatric diagnoses. Screening for problematic substance use resulted in positive screens for most participants, particularly for illicit drugs (72.1% of sample) and alcohol (65.1% of sample).

Among health-seeking behaviors, statistically significant differences (p<.05) were only found in the number of mental health visits in the two years prior to assessment. Persons in the sheltered housing group had the highest mean visits (15.1); persons in the stable and street housing groups had similar mean visits (8.7 and 7.5, respectively). No differences at this significance threshold were seen in the number of primary care visits (sample mean 8.1 visits/two years) or Emergency Department visits (2.2 visits/two years). Though outpatient substance use disorder treatment was relatively low (5.5% of sample with positive SUBS screen), use of residential treatment was high (86.3% of sample with positive SUBS screen). Hospitalization rates were notable, with 12.8% of the sample having at least one medical/surgical admission and 15.1% of the sample having at least one mental health admission.

In considering housing-seeking behaviors, HUD-VASH engagement was high in all three groups. No statistically significant differences (p<.05) were found in rates of HUD-VASH engagement by housing outcome; 50.0% of the sample had engaged in program services over the past two years. However, among participants engaged in HUD-VASH, those who achieved stable housing had the longest (p<.05) mean duration of HUD-VASH engagement (20.6 months), compared to the sheltered (8.6 months) and street homeless (11.1 months) groups. Between-group differences (p<.05) were also seen in rates of participants who attained housing via HUD-VASH, including the majority of the stable group (91.7%), and far fewer of the sheltered (33.3%) and street homeless (15.4%) groups.

Recursive partitioning analyses did not yield a stable model to classify participants by housing status. The best-available model highlighted visual learning (a cognitive domain) and monthly income as relevant variables; however, cross-validation of this model, (Kohavi, 1995) which aims to identify if potential predictors are generalizable to the underlying population of homeless persons at SMI, showed no stable solution. As such, the role of these two variables can only be considered exploratory; these analyses did not generate a robust multivariate model.

Discussion

This study aimed to identify whether cognition and other person-level factors are associated with concurrent housing status among homeless-experienced Veterans with SMI. Though cognition predicts other functional outcomes among adults with SMI, cognition and other person-level factors did not predict the type of housing in which homeless-experienced Veterans with SMI spent the majority of their past two years. In fact, few significant between-group differences were revealed in these analyses.

Our findings parallel the scant literature that examines cognition among homeless-experienced persons. In a large, well-characterized Canadian cohort of homeless adults with mental illness, a prolonged course of homelessness was not significantly associated with nonsocial cognition. (Stergiopoulos et al., 2015) In this same cohort, housing stability was not associated with changes in cognition seen over time. (Stergiopoulos et al., 2019) Similarly, in a sample of Veterans with traumatic brain injury, between-group differences between homeless-experienced and non-homeless Veterans were only seen on two of fifteen neurocognitive tests. (Twamley et al., 2019) Among Veterans with SMI who have experienced homelessness, cognition may simply have a different role in affecting housing versus other functional outcomes.

We note that persons in the sheltered housing group had more mental health visits than their peers who were stably housed or street homeless. Rather than reflecting the role of mental health care on housing outcomes, this difference presumably stems from frequent embedding of mental health services in residential treatment and transitional housing programs; it highlights the success of these programs in increasing access to mental health care. Similarly, the sheltered group had a greater frequency of case manager visits; this difference likely reflects on-site case management in many sheltered settings. Though all Veterans engaged in HUD-VASH have access to a full range of financial subsidies and supportive services, it is notable that the stably housed group was the most likely to have attained housing through the HUD-VASH program; this between-group difference speaks to the value of HUD-VASH and other supportive housing initiatives in meeting the needs of high-risk, homeless-experienced persons with SMI. Moreover, these findings suggest that, for this population, services aimed at addressing health and social service needs may be more effective in improving housing status than clinical interventions shown to be effective in other SMI populations in improving cognition and functioning.

For this high-risk group, cognition and other person-level deficits may be less relevant to concurrent housing status than contextual factors not considered in these analyses. The financial subsidies and clinical services offered by HUD-VASH are finite, and limited by program eligibility criteria (e.g., persons with chronic homelessness are prioritized). (HUD-VASH Eligibility Criteria, n.d.) Los Angeles and other urban communities suffer from limited supply and high costs of housing; (“Housing Needs Assessment,” 2013) there is a distinct lack of spaces that offer specific accommodations for the fragile support networks of this high-risk group, such as residential arrangements that meet the distinct needs of homeless-experienced Veterans with significant trauma histories. Compared to other functional outcomes (e.g., vocational pursuits), housing status is likely more strongly influenced by contextual factors.

However, it is overly simplistic to assume that housing status is determined in whole by social determinants of health. Though housing costs and stock are critical considerations, it is likely that person-level factors not considered here do influence housing outcomes. This inquiry into cognition as a potential determinant of housing status was important; cognition is a robust predictor of other functional outcomes for persons with SMI (Barbato et al., 2013; Green, 1996; Green et al., 2000) and there are effective interventions that improve nonsocial and social cognition that could be tailored for this homeless-experienced group. However, we studied a convenience sample of homeless-experienced Veterans with SMI; though some participants may have desired to improve their housing status, not all participants were actively engaged with homeless services. Cognition may matter when it is called upon, i.e., when individuals desire to change their housing status and are motivated for such. We also note that cognitive interventions may affect economic and social factors (e.g., employment) that are proximal to housing stability (e.g., by increasing income). Person-level factors not assessed here, e.g., personal attitudes, preferences, or motivation, may be relevant, along with clinical characteristics that are challenging to assess with standardized instruments, e.g., difficulty being indoors (due to anxiety, depression, or persecutory delusions), social isolation, or disruptive behaviors.

There are limitations to this study. We conducted this study within VA, an enriched setting for housing and mental health services that serves primarily men, all of whom qualified for military service. As such, our findings are limited to Veterans and may not extrapolate to other homeless-experienced persons with SMI or to a population of women who have experienced homelessness. While a portion of non-Veterans who experience homelessness suffer from developmental deficits that impact cognition, (Depp et al., 2015) homeless-experienced Veterans are less likely to have cognitive impairments related to intellectual disabilities, developmental disorders, and/or autism, as they were accepted into the U.S. Armed Forces. As compared to homeless-experienced non-Veterans, they are better educated (more likely to have completed 12 years of education but less likely to have graduated from college). (M A Winkleby, 1993; Tessler, Rosenheck, & Gamache, 2002) By nature of their military service, a cohort of homeless-experienced Veterans with SMI may be more homogenous than the general population of adults who have experienced homelessness and who have mental illness.

In addition, these analyses reflect cross-sectional data collection on a small number of participants; at the time of assessment, participants’ cognitive abilities had influenced service use, social supports, and financial resources (disability monies, employment, or other monetary sources) for years on end. Moreover, examining paths to entering homelessness for persons with SMI, by level of cognitive deficits, may prove more informative but was beyond the scope of this study. Similarly, the role of cognition on housing retention was outside this study’s aims. We classified individuals by their predominant housing status over the past two years; an individual with longstanding street homelessness who achieved stable housing for the past two years might be grouped with someone who was homeless-experienced but had never lived on the streets. It could prove valuable to gather housing status for a period longer than two years. However, our methods of grouping participants parallel work done in several other studies of homeless persons with SMI. (Gabrielian et al., 2015; Llerena et al., 2018) We also note that nights spent in residences of participants’ family and friends were classified as stable housing; though this classification reflects guidance from the instrument (Tsemberis et al., 2006) used to collect housing history, some of these individuals may be vacillating between multiple residences (i.e., “couch-surfing”) and thus lack housing stability.

We only considered SMI diagnoses over a two-year period; however, these illnesses are chronic and participants were recruited from VA homeless programs that are closely linked to mental health services. Last, this preliminary study used retrospective housing data, juxtaposed with assessments of potential predictors of housing status at the end of this time frame. As cognitive deficits appear early in psychiatric illness and are stable throughout the illness course, (Meyer & Kurtz, 2009; Pinkham, 2014) this inquiry was reasonable, but inferior to collecting person-level potential predictors and prospective housing data.

Conclusions

This is one of the first studies to comprehensively study the role of cognition as one of a breadth of person-level factors relevant to housing status for homeless-experienced Veterans with SMI. Though limited by a small sample of Veterans of the U.S. Armed Forces, these data suggest that cognition may not determine housing status for this vulnerable population. Further research, using prospective data collection among homeless persons motivated to improve their housing status (e.g., permanent supported housing enrollees), considering diverse contextual and person-level characteristics, can help build better algorithms to predict housing outcomes for this vulnerable group. Regardless, increased access to high quality supportive housing (e.g., Housing First programs) – or other affordable housing options with wraparound clinical services – is likely the most viable option to facilitate the greatest numbers of exits from homelessness for persons with SMI.

Supplementary Material

Supplemental Material

Acknowledgments

This research was supported by VA HSR&D PPO 14-188 (PI: Gabrielian).

References

  1. Andersen RM (1968). Behavioral Model of Families’ Use of Health Services. Research Series No. 25. [Google Scholar]
  2. Andersen RM (1995). Revisiting the behavioral model and access to medical care: does it matter? Journal of Health and Social Behavior, 36(1), 1–10. [PubMed] [Google Scholar]
  3. Andersen RM, & Davidson PL (2007). Improving Access to Care in America. (R. M. Andersen, T. H. Rice, & G. F. Kominski, Eds.). San Francisco: Jossey-Bass. [Google Scholar]
  4. Barbato M, Liu L, Penn DL, Keefe RSE, Perkins DO, Woods SW, & Addington J (2013). Social cognition as a mediator between neurocognition and functional outcome in individuals at clinical high risk for psychosis. Schizophrenia Research, 150(2–3), 542–546. 10.1016/j.schres.2013.08.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bo Q, Mao Z, Li X, Wang Z, Wang C, & Ma X (2017). Use of the MATRICS consensus cognitive battery (MCCB) to evaluate cognitive deficits in bipolar disorder: A systematic review and meta-analysis. Plos One, 12(4), e0176212. 10.1371/journal.pone.0176212 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bora E, Harrison BJ, Yücel M, & Pantelis C (2013). Cognitive impairment in euthymic major depressive disorder: a meta-analysis. Psychological Medicine, 43(10), 2017–2026. 10.1017/S0033291712002085 [DOI] [PubMed] [Google Scholar]
  7. Depp CA, Vella L, Orff HJ, & Twamley EW (2015). A quantitative review of cognitive functioning in homeless adults. The Journal of Nervous and Mental Disease, 203(2), 126–131. 10.1097/NMD.0000000000000248 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Fett A-KJ, Viechtbauer W, Dominguez M-D-G, Penn DL, van Os J, & Krabbendam L (2011). The relationship between neurocognition and social cognition with functional outcomes in schizophrenia: a meta-analysis. Neuroscience and Biobehavioral Reviews, 35(3), 573–588. 10.1016/j.neubiorev.2010.07.001 [DOI] [PubMed] [Google Scholar]
  9. Fitzpatrick-Lewis D, Ganann R, Krishnaratne S, Ciliska D, Kouyoumdjian F, & Hwang SW (2011). Effectiveness of interventions to improve the health and housing status of homeless people: a rapid systematic review. BMC Public Health, 11(1), 638. 10.1186/1471-2458-11-638 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. FY 2013 Budget: Housing and Communities Built to Last. (2012). FY 2013 Budget: Housing and Communities Built to Last, 1–79. [Google Scholar]
  11. Gabrielian S, Bromley E, Hellemann GS, Kern RS, Goldenson NI, Danley ME, & Young AS (2015). Factors Affecting Exits From Homelessness Among Persons With Serious Mental Illness and Substance Use Disorders. The Journal of Clinical Psychiatry, e469–e476. 10.4088/JCP.14m09229 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gabrielian S, Gordon AJ, Gelberg L, Patel B, Manchanda R, Altman L, et al. (2014). Primary Care Medical Services for Homeless Veterans. Federal Practitioner, 31(10), 10–19.30369784 [Google Scholar]
  13. Gelberg L, Andersen RM, & Leake BD (2000). The Behavioral Model for Vulnerable Populations: application to medical care use and outcomes for homeless people. Health Services Research, 34(6), 1273–1302. [PMC free article] [PubMed] [Google Scholar]
  14. Goldberg JF, & Chengappa KNR (2009). Identifying and treating cognitive impairment in bipolar disorder. Bipolar Disorders, 11 Suppl 2, 123–137. 10.1111/j.1399-5618.2009.00716.x [DOI] [PubMed] [Google Scholar]
  15. Green MF (1996). What are the functional consequences of neurocognitive deficits in schizophrenia? The American Journal of Psychiatry, 153(3), 321–330. 10.1176/ajp.153.3.321 [DOI] [PubMed] [Google Scholar]
  16. Green MF, Kern RS, & Heaton RK (2004). Longitudinal studies of cognition and functional outcome in schizophrenia: implications for MATRICS. Schizophrenia Research, 72(1), 41–51. 10.1016/j.schres.2004.09.009 [DOI] [PubMed] [Google Scholar]
  17. Green MF, Kern RS, Braff DL, & Mintz J (2000). Neurocognitive deficits and functional outcome in schizophrenia: are we measuring the “right stuff”? Schizophrenia Bulletin, 26(1), 119–136. [DOI] [PubMed] [Google Scholar]
  18. Hellemann G, Conner BT, Anglin MD, & Longshore D (2008). Seeing the trees despite the forest: Applying recursive partitioning to the evaluation of drug treatment retention. Journal of Substance Abuse Treatment, 59–64. 10.1016/j.jsat.2008.03.005 [DOI] [PubMed] [Google Scholar]
  19. Horan WP, Kern RS, Shokat-Fadai K, Sergi MJ, Wynn JK, & Green MF (2009). Social cognitive skills training in schizophrenia: an initial efficacy study of stabilized outpatients. Schizophrenia Research, 107(1), 47–54. 10.1016/j.schres.2008.09.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Housing Needs Assessment. (2013). Housing Needs Assessment. In The Los Angeles Housing Element (pp. 1–79). [Google Scholar]
  21. HUD-VASH Eligibility Criteria. (n.d.). HUD-VASH Eligibility Criteria. RetrievedJune 4, 2012, from http://www.va.gov/homeless/hud-vash.asp
  22. Hwang SW, & Burns T (2014). Health interventions for people who are homeless. Lancet, 384(9953), 1541–1547. 10.1016/S0140-6736(14)61133-8 [DOI] [PubMed] [Google Scholar]
  23. Idiculla T, Speredelozzi A, & Miller A (2005). BASIS-24 Comparison Group Report. Department of Mental Health Services Evaluation, McLean Hospital, 1–14. [Google Scholar]
  24. Kaye JL, Dunlop BW, Iosifescu DV, Mathew SJ, Kelley ME, & Harvey PD (2014). Cognition, functional capacity, and self-reported disability in women with posttraumatic stress disorder: Examining the convergence of performance-based measures and self-reports. Journal of Psychiatric Research, 57(C), 51–57. 10.1016/j.jpsychires.2014.06.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kern RS, Green MF, Mitchell S, Kopelowicz A, Mintz J, & Liberman RP (2005). Extensions of errorless learning for social problem-solving deficits in schizophrenia. The American Journal of Psychiatry, 162(3), 513–519. 10.1176/appi.ajp.162.3.513 [DOI] [PubMed] [Google Scholar]
  26. Kern RS, Nuechterlein KH, Green MF, Baade LE, Fenton WS, Gold JM, et al. (2008). The MATRICS Consensus Cognitive Battery, part 2: co-norming and standardization. The American Journal of Psychiatry, 165(2), 214–220. 10.1176/appi.ajp.2007.07010043 [DOI] [PubMed] [Google Scholar]
  27. Klein KJK, & Hodges SD (2001). Gender Differences, Motivation, and Empathic Accuracy: When it Pays to Understand. Personality and Social Psychology Bulletin, 27(6), 720–730. 10.1177/0146167201276007 [DOI] [Google Scholar]
  28. Kohavi R (1995). A study of cross-validation and bootstrap for accuracy estimation and model selection. International Joint Conference on Artificial Intelligence (IJCAI), 14(2), 1137–1145. [Google Scholar]
  29. Kumar S, Mulsant BH, Tsoutsoulas C, Ghazala Z, Voineskos AN, Bowie CR, & Rajji TK (2016). An optimal combination of MCCB and CANTAB to assess functional capacity in older individuals with schizophrenia. International Journal of Geriatric Psychiatry, 31(10), 1116–1123. 10.1002/gps.4547 [DOI] [PubMed] [Google Scholar]
  30. Kuno E, Rothbard AB, Averyt J, & Culhane D (2000). Homelessness among persons with serious mental illness in an enhanced community-based mental health system. Psychiatric Services (Washington, D.C.), 51(8), 1012–1016. [DOI] [PubMed] [Google Scholar]
  31. Lamb HR, & Bachrach LL (2001). Some perspectives on deinstitutionalization. Psychiatric Services (Washington, D.C.), 52(8), 1039–1045. [DOI] [PubMed] [Google Scholar]
  32. Lee J, Zaki J, Harvey PO, Ochsner K, & Green MF (2011). Schizophrenia patients are impaired in empathic accuracy. Psychological Medicine, 41(11), 2297–2304. 10.1017/S0033291711000614 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Leshner AF, Tom SR, & Kern RS (2013). Errorless learning and social problem solving ability in schizophrenia: an examination of the compensatory effects of training. Psychiatry Research, 206(1), 1–7. 10.1016/j.psychres.2012.10.007 [DOI] [PubMed] [Google Scholar]
  34. Lim S, Singh TP, Hall G, Walters S, & Gould LH (2018). Impact of a New York City Supportive Housing Program on Housing Stability and Preventable Health Care among Homeless Families. Health Services Research, 53(5), 3437–3454. 10.1111/1475-6773.12849 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Llerena K, Gabrielian S, & Green MF (2018). Clinical and cognitive correlates of unsheltered status in homeless persons with psychotic disorders. Schizophrenia Research, 197, 421–427. 10.1016/j.schres.2018.02.023 [DOI] [PubMed] [Google Scholar]
  36. M A Winkleby DF (1993). Physical, addictive, and psychiatric disorders among homeless veterans and nonveterans. Public Health Reports, 108(1), 30–36. [PMC free article] [PubMed] [Google Scholar]
  37. McIntyre RS, Cha DS, Soczynska JK, Woldeyohannes HO, Gallaugher LA, Kudlow P, et al. (2013). Cognitive deficits and functional outcomes in major depressive disorder: determinants, substrates, and treatment interventions. Depression and Anxiety, 30(6), 515–527. 10.1002/da.22063 [DOI] [PubMed] [Google Scholar]
  38. McNeely J, Strauss SM, Saitz R, Cleland CM, Palamar JJ, Rotrosen J, & Gourevitch MN (2015). A Brief Patient Self-administered Substance Use Screening Tool for Primary Care: Two-site Validation Study of the Substance Use Brief Screen (SUBS). The American Journal of Medicine, 128(7), 784.e9–784.e19. 10.1016/j.amjmed.2015.02.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Meyer MB, & Kurtz MM (2009). Elementary neurocognitive function, facial affect recognition and social-skills in schizophrenia. Schizophrenia Research, 110(1–3), 173–179. 10.1016/j.schres.2009.03.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Nitzburg GC, DeRosse P, Burdick KE, Peters BD, Gopin CB, & Malhotra AK (2014). MATRICS cognitive consensus battery (MCCB) performance in children, adolescents, and young adults. Schizophrenia Research, 152(1), 223–228. 10.1016/j.schres.2013.11.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Nuechterlein KH, & Green MF (2006). MATRICS (Measurement and Treatment Research to Improve Cognition in Schizophrenia) Consensus Cognitive Battery (MCCB) Manual. Los Angeles, CA: MATRICS Assessment, Inc. [Google Scholar]
  42. O’Toole TP, & Pape L (2015). Innovative efforts to address homelessness among veterans. North Carolina Medical Journal. 10.2307/3070692?ref=search-gateway:35c663507716b18fc23ae793349ecbd7 [DOI] [PubMed] [Google Scholar]
  43. Padgett DK, Smith BT, Henwood BF, & Tiderington E (2012). Life course adversity in the lives of formerly homeless persons with serious mental illness: Context and meaning. American Journal of Orthopsychiatry, 82(3), 421–430. 10.1111/j.1939-0025.2012.01159.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Paquin K, Wilson AL, Cellard C, Lecomte T, & Potvin S (2014). A systematic review on improving cognition in schizophrenia: which is the more commonly used type of training, practice or strategy learning? BMC Psychiatry, 14, 139. 10.1186/1471-244X-14-139 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Patterson M, Somers J, & Moniruzzaman A (2012). Sealing the cracks: Preliminary findings from an inter-ministry initiative to address chronic homelessness in British Columbia. Journal of Interprofessional Care, 26(5), 426–428. [DOI] [PubMed] [Google Scholar]
  46. Pinkham AE (2014). Social Cognition in Schizophrenia. The Journal of Clinical Psychiatry, 75(suppl 2), 14–19. 10.4088/JCP.13065su1.04 [DOI] [PubMed] [Google Scholar]
  47. Price LE, Shea K, & Gephart S (2015). The Veterans Affairs’s Corporate Data Warehouse: Uses and Implications for Nursing Research and Practice. Nursing Administration Quarterly, 39(4), 311–318. 10.1097/NAQ.0000000000000118 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Ripoll LH, Zaki J, Perez-Rodriguez MM, Snyder R, Strike KS, Boussi A, et al. (2013). Empathic accuracy and cognition in schizotypal personality disorder. Psychiatry Research, 210(1), 232–241. 10.1016/j.psychres.2013.05.025 [DOI] [PubMed] [Google Scholar]
  49. Roberts NA, & Levenson RW (2006). Subjective, behavioral, and physiological reactivity to ethnically matched and ethnically mismatched film clips. Emotion, 6(4), 635–646. 10.1037/1528-3542.6.4.635 [DOI] [PubMed] [Google Scholar]
  50. Roca M, Monzón S, Vives M, López-Navarro E, Garcia-Toro M, Vicens C, et al. (2015). Journal of Affective Disorders. Journal of Affective Disorders, 171(C), 85–92. 10.1016/j.jad.2014.09.018 [DOI] [PubMed] [Google Scholar]
  51. Rodriguez-Jimenez R, Santos JL, Dompablo M, Santabárbara J, Aparicio AI, Olmos R, et al. (2019). MCCB cognitive profile in Spanish first episode schizophrenia patients. Schizophrenia Research, 211, 88–92. 10.1016/j.schres.2019.07.011 [DOI] [PubMed] [Google Scholar]
  52. Rosenthal D, Rotheram-Borus MJ, Batterham P, Mallett S, Rice E, & Milburn NG (2007). Housing stability over two years and HIV risk among newly homeless youth. AIDS and Behavior, 11(6), 831–841. 10.1007/s10461-007-9235-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. SAMHSA’s Co-Occurring Center for Excellence. (2006). Definitions and Terms Relating to Co-Occurring Disorders: COCE Overview Paper 1. Rockville, MD. Retrieved from coce.samhsa.gov/cod_resources/papers.htm [Google Scholar]
  54. Schwab KA, Ivins B, Cramer G, Johnson W, Sluss-Tiller M, Kiley K, et al. (2007). Screening for traumatic brain injury in troops returning from deployment in Afghanistan and Iraq: initial investigation of the usefulness of a short screening tool for traumatic brain injury. The Journal of Head Trauma Rehabilitation, 22(6), 377–389. 10.1097/01.HTR.0000300233.98242.87 [DOI] [PubMed] [Google Scholar]
  55. Sherbourne CD, & Stewart AL (1991). The MOS social support survey. Social Science & Medicine (1982), 32(6), 705–714. [DOI] [PubMed] [Google Scholar]
  56. Stata Statistical Software: Release 12StataCorp. (n.d.). StataCorp (2011. ed.). College Station, TX: Stata Statistical Software: Release 12. [Google Scholar]
  57. Stergiopoulos V, Cusi A, Bekele T, Skosireva A, Latimer E, Schütz C, et al. (2015). Neurocognitive impairment in a large sample of homeless adults with mental illness. Acta Psychiatrica Scandinavica, 131(4), 256–268. 10.1111/acps.12391 [DOI] [PubMed] [Google Scholar]
  58. Stergiopoulos V, Naidu A, Schuler A, Bekele T, Nisenbaum R, Jbilou J, et al. (2019). Housing Stability and Neurocognitive Functioning in Homeless Adults With Mental Illness: A Subgroup Analysis of the At Home/Chez Soi Study. Frontiers in Psychiatry, 10, 865. 10.3389/fpsyt.2019.00865 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Strobl C, Malley J, & Tutz G (2009). An introduction to recursive partitioning: Rationale, application, and characteristics of classification and regression trees, bagging, and random forests. Psychological Methods, 14(4), 323–348. 10.1037/a0016973 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Tessler R, Rosenheck R, & Gamache G (2002). Comparison of homeless veterans with other homeless men in a large clinical outreach program. The Psychiatric Quarterly, 73(2), 109–119. [DOI] [PubMed] [Google Scholar]
  61. Therneau TM, & Atkinson B (n.d.). The rpart package. University of California, Los Angeles Statistics Department Comprehensive R Archive Network. Retrieved from http://cran.stat.ucla.edu [Google Scholar]
  62. Tsemberis S, McHugo G, Williams V, Hanrahan P, & Stefancic A (2006). Measuring homelessness and residential stability: The residential time-line follow-back inventory. Journal of Community Psychology, 35(1), 29–42. 10.1002/jcop.20132 [DOI] [Google Scholar]
  63. Twamley EW, Hays CC, Van Patten R, Seewald PM, Orff HJ, Depp CA, et al. (2019). Neurocognition, psychiatric symptoms, and lifetime homelessness among veterans with a history of traumatic brain injury. Psychiatry Research, 271, 167–170. 10.1016/j.psychres.2018.11.049 [DOI] [PubMed] [Google Scholar]
  64. Vasterling JJ, Duke LM, Brailey K, Constans JI, Allain AN, & Sutker PB (2002). Attention, learning, and memory performances and intellectual resources in Vietnam veterans: PTSD and no disorder comparisons. Neuropsychology, 16(1), 5. 10.1037//0894-4105.16.1.5 [DOI] [PubMed] [Google Scholar]
  65. Wilson NJ, & Kizer KW (1997). The VA health care system: an unrecognized national safety net. Health Affairs, 16(4), 200–204. 10.1377/hlthaff.16.4.200 [DOI] [PubMed] [Google Scholar]
  66. Wirshing DA, Wirshing WC, Marder SR, Liberman RP, & Mintz J (1998). Informed consent: assessment of comprehension. The American Journal of Psychiatry, 155(11), 1508–1511. [DOI] [PubMed] [Google Scholar]
  67. Wykes T, Huddy V, Cellard C, McGurk SR, & Czobor P (2011). A meta-analysis of cognitive remediation for schizophrenia: methodology and effect sizes. The American Journal of Psychiatry, 168(5), 472–485. 10.1176/appi.ajp.2010.10060855 [DOI] [PubMed] [Google Scholar]
  68. Wynn JK, Gabrielian S, Hellemann G, Horan WP, Kern RS, Lee J, et al. (2020). Motivational and cognitive factors linked to community integration in homeless veterans: Study 2 - clinically diverse sample. Psychological Medicine, 3, 1–8. 10.1017/S0033291720001609 [DOI] [PubMed] [Google Scholar]
  69. Zaki J, & Ochsner K (2009). The Need for a Cognitive Neuroscience of Naturalistic Social Cognition. Annals of the New York Academy of Sciences, 1167(1), 16–30. 10.1111/j.1749-6632.2009.04601.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Zaki J, Bolger N, & Ochsner K (2008). It Takes Two: The Interpersonal Nature of Empathic Accuracy. Psychological Science, 19(4), 399–404. 10.1111/j.1467-9280.2008.02099.x [DOI] [PubMed] [Google Scholar]
  71. Zhang H, & Singer B (n.d.). Recursive Partitioning in the Health Sciences (1999. ed.). New York: Springer. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
NIHMS1732561-supplement-Supplemental_Material.pdf (89.8KB, pdf)

RESOURCES