Abstract
Background:
The work of addressing veteran homelessness has largely been focused on veterans who are already homeless.
Objectives:
This study aimed to identify factors that can be targeted upstream before military personnel leave the military to prevent veteran homelessness during the critical transition from active duty to civilian life.
Design:
Data were analyzed from a 2001 to 2014 longitudinal cohort study of 418 624 post-9/11 veterans who entered Veterans Affairs (VA) healthcare after leaving the military.
Methods:
Department of Defense (DoD) data on clinical diagnoses, demographics, and military history were linked to VA data on homelessness and neighborhood of residence.
Results:
Homelessness in the 2 years after military discharge was associated with residing in a socioeconomically disadvantaged neighborhood after discharge as well as with younger age; Black race; and diagnoses of substance use disorder (SUD), serious mental illness (SMI), and personality disorder. Veterans with co-occurring SUD, SMI, and personality disorder had 5 times higher incidence of homelessness than veterans with none of these diagnoses, with rates most elevated among veterans residing in disadvantaged neighborhoods.
Limitations and Conclusion:
Several limitations include potential for missed cases of homelessness due to the use of medical records and lack of generalizability as note all veterans utilize VA services. Nevertheless, this large-sample, longitudinal sampling frame revealed critical environment-level and individual-level risk factors predicting homelessness after military separation that can be addressed proactively by policy and programs aimed at improving community reintegration of veterans transitioning to civilian life.
Introduction
Military veterans in the United States (US) have historically experienced elevated rates of homelessness, which have been dramatically reduced in the past decade due to major federally funded initiatives, including those led by the US Department of Veterans Affairs (VA). 1 Research shows that homelessness in veterans is related to various risk factors, including severe mental illness, substance use disorder, adverse childhood experiences, poverty, and other medical and psychosocial factors.2 -4 Despite studies investigating homelessness among veterans overall and within specific cohorts,5,6 little is known about homelessness among service members as they transition from active duty to civilian status.
Given that many empirical investigations of veteran homelessness employ a cross-sectional design,2,3 this research cannot speak to prospective risk factors and does not permit verification of clinical data at the time of military discharge. Moreover, reviews of risk factors for homelessness among veterans have focused relatively more on individual-level (eg, mental disorders) as compared to societal or environment-level predictors. In particular, veterans who have recently left the military and live in more socially disadvantaged neighborhoods may be at higher risk of homelessness, regardless of other individual-level risk factors. While studies have found that on average, veterans first become homeless 5 to 10 years after military discharge, 7 there is some evidence that more recent cohorts of Iraq and Afghanistan veterans are becoming homeless earlier (ie, within 1 or 2 years of military discharge) which may be due to social adjustments issues after military separation and the increased efforts of VA to provide homeless prevention assistance.4,8
There have been a few notable studies that have used prospective designs with unique datasets that hint at the importance of individual and environment-level factors. Two recent studies analyzed data from the Army Study to Assess Risk and Resilience in Servicemembers-Longitudinal Study (Army STARRS-LS) project and found that individual characteristics, such as mental disorders and major life events, were predictive of homelessness.8,9 Importantly, these studies also found that transitioning service members’ address of record upon discharge, were also predictive of their homelessness. More longitudinal studies, of this kind, following recently discharged veterans from military to civilian life, that address social determinants along with individual risk factors are needed to address these crucial gaps in the scientific literature.
The current study links US Department of Defense (DoD) and Department of Veterans Affairs (VA) administrative data to identify environment-level and individual-level risk factors for homelessness during the critical first 2 years after military discharge. Findings would inform and tailor prevention and early intervention programming. Specifically, analyzing longitudinal data of homelessness among recently discharged military veterans would be helpful for 1) identifying veterans at high risk for homelessness, allowing military and veteran agencies to tailor services for service members leaving the military; (2) pinpointing dynamic risk and protective factors that can be targeted to inform prevention efforts of homelessness; and (3) enabling homelessness programs to better gauge what specific level and type of services are needed for each military veteran.
Methods
Sample and Data Sources
A cohort of post-9/11 veterans (n = 418 624) was identified who served in the active duty component of the US military and who (1) entered Veterans Health Administration (VHA) care between October 1, 2001 and September 30, 2014; (2) had 2 years of VHA care (recorded in either inpatient, outpatient, or pharmacy records) in at least 2 years between FY2002 and 2018; (3) entered VHA care within 1 year of last date of active duty; and (4) received care in the DoD healthcare system within 1 year before DoD separation. Individuals who entered VHA care before military discharge were excluded. Inclusion criteria were selected to ensure data completeness of outcome data on homelessness. This study received institutional review board approval from the University of Utah with a waiver of informed consent (IRB #00127108). A previous study was conducted with this dataset focused on criminal justice involvement among veterans, 10 and the current study focused on homelessness among veterans.
For this sample, administrative data were acquired from the DoD and VHA. DoD data included demographic characteristics, clinical data obtained via the DoD Health Risk File, and inpatient and outpatient encounter data obtained via the DoD and VA Infrastructure for Clinical Intelligence (DaVINCI). VHA data sources included national inpatient and outpatient encounter data from the VHA Corporate Data Warehouse and mortality data from the VHA vital status files. Physical health conditions were identified using International Classification of Diseases, (both 9th and 10th edition), Clinical Modification (ICD-9-CM/10) codes, and merged demographic characteristics from the DoD Health Risk File with outcomes identified in VHA data for a longitudinal analytic data sets using unique identifiers provided in DaVINCI (electronic data interchange personal identifier).
Measures
Demographics/Military Background
From the DoD, data was gathered on age at entry to VHA care, sex, race/ethnicity, and marital status. Characteristics of the veterans’ military service included branch of service, rank, and history of deployment during military service.
Clinical Status
Variables in DoD’s Health Risk file were used to identify individuals diagnosed with traumatic brain injury (TBI), posttraumatic stress disorder (PTSD), personality disorder (PD), conduct disturbance, and severe mental illness (SMI; depression, schizophrenia, or bipolar disorder) prior to military discharge. DoD classifies these conditions based on Medicaid managed care algorithms requiring either 1 inpatient diagnosis or 2 outpatient diagnoses made by any provider at least 7 days apart. ICD-9-CM/10 were used to identify diagnoses of SUD in DoD data. The Selim physical comorbidity index, 11 developed for military veterans, was used to measure 30 chronic physical diseases using ICD-9-CM/10 codes and grouped into the following categories: 0, 1, 2, 3, or >4 diagnoses.
Socioeconomic Status
We determined veterans’ neighborhood of residence after military discharge by the 9-digit zip code documented in the VHA medical record closest to VHA entry date. The Area Deprivation Index (ADI) is a validated measure of socioeconomic disadvantage built at the Census block group level that ranges from 1 to 100, with a ranking of 1 indicating the lowest level of “disadvantage” and 100 indicating the highest level. The ADI was computed by linking veterans’ zip codes to Census block group crosswalk 12 and was grouped into quartiles (higher scores/quartiles represent higher levels of disadvantage).
Homelessness
Our dependent variable was an indicator of homelessness at any point during the 2 years after the DoD-defined military discharge date. Consistent with previous research,5,6 it was constructed using both clinic stop codes (3-digit location codes that capture use of VA homeless programs) and ICD-9 CM/10 codes drawn from VHA medical records.
Analysis
Analyses were conducted in SAS. Descriptive statistics summarized demographics, military history, clinical status, and socioeconomic disadvantage of the entire sample and veterans who became homeless within 2 years of military separation. Next, multivariable logistic regression analyses were conducted predicting the occurrence of homelessness within 2 years of separation from the military. Finally, to explore potential intersections between individual and environment-level factors, the mean incidence of homelessness was calculated as a function of combinations of 3 risk factors (SMI, SUD, and PD), stratified by ADI quartile.
Results
Table 1 shows characteristics for the entire sample of recently discharged veterans. Eighty-three percent were male, 47% married, 59% White, 18% Black, 10% Hispanic, 3% Asian, 2% Native American/Pacific Islander, and 8% unknown race/ethnicity. With respect to military background, 45% had served in the US Army, 91% were enlisted, and 68% had been deployed. Clinically, 4% met criteria for TBI, 16% met criteria for SMI, 24% had SUD diagnosis, 8% had PTSD diagnosis, 2% had PD, and <1% had conduct disturbance diagnosis. About 53% had 1 or more medical conditions on the Selim physical comorbidity index, with the most common conditions being lower back pain (34.01%), high blood pressure (17.63%), and osteoarthritis (10.34%). Sixteen percent were in the highest ADI quartile.
Sample Characteristics of Transitioning Service Members.
Abbreviation: ADI, Area Deprivation Index.
Among recently discharged veterans who became homeless within 2 years of separation from the military, 82% were male, only 32% married, 51% were White, 30% Black, 12% Hispanic, 2% Asian, 3% Native American/Pacific Islander, and 3% unknown race/ethnicity (Table 1). With respect to military background, 60% had served in the US Army, 98% were enlisted, and 77% had been deployed. Clinically, 6% met criteria for TBI, 31% met criteria for SMI, 46% SUD, 15% PTSD, 6% PD, and 1% conduct disturbance. Among these veterans, the most common medical conditions on the Selim physical comorbidity index were lower back pain (35.54%), high blood pressure (11.34%) and chronic obstructive pulmonary disease (6.62%). Twenty-four percent lived in neighborhoods in the highest quartile of disadvantage.
Multivariable analysis demonstrated that nearly all measured demographic and military service characteristics were associated with homelessness (Table 2). SUD (OR = 2.17, CI = 2.10-2.24; P < .001), SMI (OR = 2.02, CI = 1.94-2.10; P < .001), PD (OR = 1.84, CI = 1.71-1.98; P < .001), and Black race (OR = 2.34, CI = 2.26-2.43; P < .001) evidenced strong associations with incident homelessness. Being younger was associated with homelessness; specifically, for age at VA entry, veterans aged 40 to 49 years (OR = 0.42, CI = 0.39-0.45; P < .001) and aged 50 (OR = 0.25, CI = 0.20-0.32; P < .001) were significantly less likely to become homeless compared to Veterans aged 17 to 29 years.
Multivariable Models of Homelessness Within 2 Years of Military Separation.
Abbreviations: ADI, Area Deprivation Index; DoD, Department of Defense.
For branch of service, using the Army as the reference group, the odds ratio for the Air Force is 0.50 (CI = 0.48-0.53; P < .001), for the Marines it is 0.65 (CI = 0.63-0.68; P < .001), for the Navy/Coast Guard it is 0.70 (CI = 0.67-0.73; P < .001), and for other branches it is 0.24 (CI = 0.10-0.57; P < .001). Regarding the highest rank, using enlisted personnel as the reference group, the odds ratio for officers is 0.41 (CI = 0.37-0.46; P < .001), and for warrant officers it is 0.33 (CI = 0.26-0.43; P < .001).
Other correlates of homelessness included: male sex (OR = 0.91, CI = 0.87-0.95; P < .001), Hispanic ethnicity (OR = 1.35, CI = 1.29-1.42; P < .001), Native American or Pacific Island race (OR = 1.59, CI = 1.45-1.73; P < .001), PTSD (OR = 1.14, CI = 1.09-1.20; P < .001), military deployment history (OR = 1.19, CI = 1.15-1.24; P < .001), TBI (OR = 1.18, CI = 1.11-1.26; P < .001), conduct disturbance (OR = 1.31, CI = 1.13-1.52; P < .001), and residing in neighborhoods with the greatest disadvantage (OR = 1.19, CI = 1.13-1.25; P < .001).
Figure 1 illustrates the mean incidence of homelessness in the 2 years following military separation, as a function of combinations of risk factors (SMI, SUD, and PD), stratified by ADI quartile. Veterans in the highest ADI quartile had higher rates of homelessness. Veterans with SUD, SMI, and PD had more than 5 times higher mean incidence of becoming homeless compared to veterans with none of these risk factors regardless of ADI quartile. In the highest ADI quartile, the mean incidence of homelessness was 0.25 for veterans with SMI + SUD + PD compared to 0.04 for veterans with none of these risk factors. Figure 2 illustrates the co-occurrence of SMI, SUD, and PD in the total sample of veterans with any 2-year incidence of homelessness.
Mean incidence of homelessness within 2 years of military separation.
Venn diagram showing co-occurrence of severe mental illness, substance use disorder, and personality disorders in veterans with any 2-year incidence of homelessness.
Discussion
This paper examines risk factors of homelessness among recently discharged veterans during their transition from military to civilian life. Drawing on DoD and VHA data sources, our findings corroborated prior research linking veteran homelessness to SMI,2,13 substance use disorder,6,14 and misconduct.2,15 We found that individual (eg, race/ethnicity, military rank) and environment-level markers (eg, ADI) were salient homeless risk factors, consistent with the few past studies that exist on the topic.4,8,9 These factors might be better understood within the context of a social vulnerability framework,16,17 in which a combination of social, cultural, economic, and institutional-level factors may shape and influence recovery from major life events like housing instability and homelessness. With this framework and new data sharing and cross-agency collaboration, the DoD and VA may be able to address these factors to improve veterans’ successful community reintegration. There may also be opportunities to track whether these factors change over time and particular trends in the military-to-civilian transition or onset of homelessness after military discharge with new DoD-VA collaborative efforts.
These results have important implications for community reintegration programs and homelessness prevention efforts of military veterans. First, this is one of the few studies to investigate community deprivation and homelessness among recently discharged veterans. Many of the individual-level risk factors identified in Table 1 likely interact with one’s neighborhood of residence. This is consistent with recent research finding that residential neighborhoods may influence homeless veterans’ use of healthcare utilization as a function of public transportation use 18 and one’s residential address after military discharge may predict risk of homelessness. 9 As a result, the current study findings speak to the need to consider these environment-level factors of where veterans will be living after separation in policies and clinical practices intending to facilitate community reintegration. The DoD’s Transition Assistance Program (TAP) and programs focused on connecting military personnel preparing for discharge to VA programs may better utilize this information.
Second, consistent with recent analyses of veterans 5 and the general U.S. population, 19 Black veterans were at higher risk of homelessness after military discharge, followed by Hispanic, Native American, and veterans from Pacific Island origins. Pathways to homelessness may be influenced by factors related to cultural background, that may influence individual-level (eg, untreated psychiatric problems, lack of intergenerational wealth) and environment-level (eg, impoverished neighborhood) factors. 5 Additional research is needed to understand cultural differences and community reintegration among veterans. We also found in our data that male veterans were at greater risk for homelessness, but it may be important to consider potential sex differences by type of homelessness as women veterans have been noted in other findings as particularly at high risk of unsheltered homelessness in recent years. 20 There may be race/ethnicity factors to consider in their interaction with sex and type of homelessness in this context as well, that can serve as the basis for future study.
Third, a new finding was that a chart diagnosis of PD significantly predicted future homelessness in veterans. Nearly 1 in twenty veterans with homeless experience were diagnosed with PD, which may exist alone or comorbid with SMI and SUD. Our findings underscore the importance of considering PD in the context of military separation. In addition to efforts to screen for psychiatric and substance use disorders, the results point to a need for routine screening of behaviors indicative of PD. Adding this will help identify veterans who may require treatment, case management, or other assistance to prevent homelessness during this transition period.
Fourth, a striking percentage (46%) of veterans who became homeless within 2 years of military discharge had a documented SUD diagnosis before military discharge. Untreated SUD contributes to prolonged homelessness and poor outcomes and is a major concern among homeless veterans due to their high risk for overdose deaths 21 and continued substance use disorder that may contribute to their return to homelessness. 22 We do want to caveat our findings though that our total sample had an unusually high rate of SUD (25% in total sample, regardless of any homeless experience or not) so our sample may not be fully representative of this population. Nonetheless, our findings indicate the importance of connecting veterans with SUD treatment as a critical strategy to prevent homelessness among recently discharged veterans.
Fifth, an unexpected finding in our study indicates that while having 1 physical condition is associated with greater odds of experiencing homelessness, having 4 or more physical conditions is associated with lower odds of homelessness. We hypothesize that Veterans with multiple physical health conditions have a greater need to utilize VA medical services, which can facilitate access to treatments for mental and substance use disorders, ultimately helping to prevent homelessness.
Study limitations should be considered. Homelessness was identified with ICD-10 codes and stop codes, but did not include a few indicators used in other studies so some cases of homelessness may have been missed, 23 and our homelessness indicator might not fully capture those who are informally homeless. Although the binary homelessness indicator aligns with existing literature, it is also the case that a binary outcome does not capture the timing or duration of homelessness. Related, it is difficult to determine whether the incidence of homelessness was influenced by greater attention to homelessness, more available services and programs, or increasing risk of homelessness among veterans at earlier time periods; as a result, future analyses should examine the incidence of homeless over the years post-separation, ideally in increments (eg, 5-year time periods).
Many veterans do not utilize VHA services after military discharge, limiting generalizability to the entire US veteran population. Additionally, relying on static neighborhood data might not accurately reflect changes in mobility after discharge. Future research should include qualitative analyses about changes in mobility after discharge and themes veterans identify as contributing to housing instability and homelessness. Given the large sample size, a power analysis for sample size calculation was not done. Finally, other unmeasured factors may contribute to veteran homelessness, including household income, employment stability, and housing affordability. Greater attention to these factors in future studies are needed to examine not only veteran homelessness, but use of any available services and programs.
Examining homelessness during the transition from military to civilian life, this study identified risk factors that can be prioritized by the DoD, VA, and other healthcare systems to improve the likelihood that veterans reintegrate successfully into the community. Such empirical data have direct implications for programs serving military personnel preparing for military discharge, such as the DoD Transition Assistance Program (TAP), which provides vocational opportunities and training to meet post-military goals. The findings that rank and military branch predicted future homelessness are critical for the DOD to integrate into the TAP program, especially for enlisted personnel and for those who served in the Army. TAP counselors could improve community success of Veterans by using the risk factors in Table 2 to gauge the level of need among Service Members to ensure they have solid plans for housing upon separation from the military.
Findings are also relevant to the VA Military to Civilian Readiness Pathway (M2C Ready), which supports recently discharged veterans through 365 days post-separation. The current data is also relevant to the VA Solid Start Program, in which every newly separated service member is called 3 times during their first year of separation to help them get a home loan, to health care, return to work, or access mental health care. Findings pertain to VHA homeless programs as well, which provide housing and other psychosocial support for veterans experiencing homelessness. By identifying service members at highest risk for homelessness in advance, these programs can gauge what specific level of services may be needed for a given service member when they reintegrate into the community.
Conclusion
This study provides valuable information that specific environment-level factors, like a veteran’s residential address, and specific individual-level factors, like a serious mental illness, substantive use disorder, and personality disorders, are predictive of homelessness after military discharge. Despite limitations, the study has numerous strengths including large sample size, use of ADI, analysis of new variables such as personality disorder, and focus on the transition from military to civilian life. As a result, DOD and VA policies and programs can use this information to address risk factors proactively among service members and to improve the likelihood that they reintegrate successfully into the community when they become veterans.
Footnotes
Acknowledgements
Thanks to Linda Southcott, MA, for her review of this work for its operational utility.
Ethical Considerations
This study received institutional review board approval from the University of Utah (IRB #00127108).
Consent to Participate
A waiver of informed consent was approved by the University of Utah.
Author Contributions
Eric B. Elbogen led conceptualization and writing of the paper. Mary Jo Pugh helped provide funding and administrative support. Megan Amuan led data analysis. Shannon M. Blakey and Robert C. Graziano helped interpret the results and write the paper. Richard E. Nelson, and Audrey L. Jones helped review and edit the paper. Jack Tsai helped conceptualize the study, provide administrative support, and write the paper.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Assistant Secretary of Defense for Health Affairs endorsed by the US Department of Defense, through the Psychological Health/Traumatic Brain Injury Research Program Long-Term Impact of Military-Relevant Brain Injury Consortium (LIMBIC) Award/W81XWH-18-PH/TBIRP-LIMBIC under Awards No W81XWH1920067 and W81XWH-13-2-0095, and by the US Department of Veterans Affairs Awards No I01 CX002097, I01 CX002096, I01 HX003155, I01 RX003444, I01 RX003443, I01 RX003442, I01 CX001135, I01 CX001246, I01 RX001774, I01 RX 001135, I01 RX 002076, I01 RX 001880, I01 RX 002172, I01 RX 002173, I01 RX 002171, I01 RX 002174, and I01 RX 002170. The US Army Medical Research Acquisition Activity, 839 Chandler Street, Fort Detrick MD 21702-5014 is the awarding and administering acquisition office. This work was also developed in collaboration with the US Department of Veterans Affairs National Center on Homelessness among Veterans. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the US Department of Defense. Mary Jo Pugh is also supported by a VA Health Services Research and Development (HSR&D) Research Career Scientist Award (RCS 17-297, Award No, IK6HX002608). Any opinions, findings, interpretations, conclusions, and recommendations expressed in this publication are those of the authors. They do not necessarily reflect the views of the US Government, the US Department of Veterans Affairs, or the US Department of Defense, and no official endorsement should be inferred.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
Data is available upon request from the corresponding author and with proper institutional approvals.