To the Editor: Researchers have documented mental illness profiles (1), homelessness trajectories (2), and intervention outcomes (3) among homeless veterans with mental illness. Few recent studies, however, have examined the influence of mental illness on homelessness risk in a general veteran population receiving care in the Department of Veteran Affairs (VA) health care system. Although homeless veterans report a high prevalence of mental illness (4) and incidental findings of recent studies indicate that psychoses (1) and posttraumatic stress disorder (PTSD) (2) influence homelessness risk, few studies have systematically examined various mental illnesses to identify the ones that contribute most to homelessness risk.

Addressing this need, this study examined mental illness and homelessness among 6,819 VA patients in the Veterans Aging Cohort Study, an eight-site study of veterans in VA care in Atlanta, Baltimore, New York, Houston, Los Angeles, Pittsburgh, and Washington, D.C. (5).

Homelessness was measured as lifetime homelessness, homelessness in the past month, and shelter use in the past month. Mental illness diagnoses were obtained from hospital administrative data. The effects of depression, PTSD, anxiety, and schizophrenia were assessed in our analysis. We conducted bivariate analyses and entered significant correlates into logistic regression models.

Almost 40% of the sample (N=2,693) had experienced past homelessness; 13% (N=866) reported recent homelessness, and 11% (N=714) reported recent shelter use. Almost 11% (N=780) had a diagnosis of major depressive disorder, 9% (N=603) had a PTSD diagnosis, 6% (N=378) had an anxiety disorder diagnosis, and 4% (N=271) had a diagnosis of schizophrenia. Compared with participants without depression, a significantly higher proportion of depressed veterans reported past homelessness (60%, N=468, versus 37%, N=2,234), current homelessness (24%, N=187, versus 11%, N=682), and current shelter use (22%, N=169, versus 9%, N=544; chi square test, p<.01 for all). Similarly, veterans with PTSD and schizophrenia reported significantly higher rates of past and current homelessness than veterans without these diagnoses. When the logistic regression models controlled for age, race, gender, socioeconomic status, as well as frequency of alcohol and drug use, depression emerged as one of the strongest risk factors, doubling or nearly doubling the risk across all three measures of homelessness (past homelessness, odds ratio [OR]=2.12, 95% confidence interval [CI]=1.79–2.50; current homelessness, OR=1.95, CI=1.60–2.38; and current shelter use, OR=2.08, CI=1.67–2.58). Compared with veterans without schizophrenia, those with this diagnosis were more likely to report past homelessness (OR=2.01, CI=1.52–2.64), current homelessness (OR=1.39, CI=1.01–1.92), and current shelter use (OR=1.77, CI=1.27–2.47). PTSD and anxiety were not associated with homelessness.

The results establish priorities for mental illness treatment as a potentially effective approach to preventing homelessness among veterans: depression and schizophrenia warrant treatment first, because other mental illnesses were not associated with homelessness. Researchers have found that PTSD is usually comorbid with other psychiatric illnesses among veterans (1), which may account for the fact that its bivariate association with homelessness disappeared after the analysis controlled for other mental illness.

Given the high prevalence of homelessness in this population and the contribution of mental illness to homelessness risk, a significant step toward the government's goal of eradicating homelessness among veterans is to target veterans with depression and schizophrenia who are already in VA care. Future research needs to examine how services mediate the association between mental illness and homelessness.

The authors report no competing interests.