Characteristics and Service Use of Homeless Veterans and Nonveterans Residing in a Low-Demand Emergency Shelter

Table 1 presents demographic and other characteristics of the two study subgroups. Nearly half (N=103, 47%) of the sample was Caucasian, 50% (N=110) were African American, and 3% (N=7) were members of other racial groups. Nine percent (N=19) reported Hispanic ethnicity. Few (N=15, 7%) were currently married. Total lifetime homelessness averaged almost four years. Most participants (N=157, 71%) had a history of adult felony conviction, and an overwhelming majority (N=209, 95%) had been incarcerated; many (N=38, 17%) had a recent arrest with criminal charges. Approximately three-quarters (N=158, 72%) of the sample reported working for pay, but the average income earned by the sample in the past 30 days was very low ($463.3±$392.6). Compared with nonveterans, the veterans were older and more educated, more likely to have ever married, and less likely to have private or public (non-VA) health care insurance. In addition, veterans were significantly older upon first experiencing homelessness.

Compared with a national population norm, the sample’s mean±SD VR-12 scores were significantly better for physical functioning (44.9±8.8 versus 38.4±12.2; t=10.9, df=219, p<.001) but significantly worse for mental functioning (40.0±13.5 versus 51.1±11.4; t=–12.2, df=219, p<.001) (22). Approximately two-thirds (N=140, 64%) of the sample had an identified alcohol problem, and a vast majority of the sample (N=181, 82%) had an identified drug problem. More than half of the sample (N=127, 58%) reported having used illicit drugs in the past 30 days; among those reporting drug use, cocaine and marijuana were the most frequently used substances. No significant differences were found between veterans and nonveterans in physical or mental functioning scores, alcohol or drug problems, or use of specific substances.

Veteran participants had served most often in the Army and during the Vietnam and post-Vietnam service eras. Slightly fewer than one-third served in a war zone, and just over two-thirds had received an honorable discharge. Fewer than one-fifth received a non–service-connected pension or service-connected disability benefits. A clear majority of veterans (N=101, 92%) reported that they were able to access VA clinical services in the past year.

Service use data are provided in Table 2. Almost three-quarters (N=158, 72%) of the sample reported some use of general medical services, especially emergency and outpatient care, in the past 12 months. Just over one-third (N=86, 39%) reported use of psychiatric services, usually outpatient services (86% of those using any psychiatric services). One-quarter (N=57, 26%) reported use of substance abuse services, especially inpatient services (71% of those using any substance abuse services).

Significantly more veterans than nonveterans used any medical or any psychiatric services, but there were no differences between the groups in the proportion that used any substance abuse services. Significantly more veterans than nonveterans used medical outpatient services; psychiatric emergency, inpatient, and outpatient services; and substance abuse inpatient services. No differences were found in the proportions of veterans and nonveterans who used non-VA medical, psychiatric, or substance abuse services. However, more nonveterans than veterans used two subcomponents of non-VA medical services (inpatient and outpatient services). Because veterans and nonveterans did not differ in the proportions using non-VA medical, psychiatric, or substance abuse services, the finding that more veterans than nonveterans used any medical, psychiatric, and substance abuse services was accounted for by the veterans’ additional use of VA services in these sectors.

Table 3 presents the results of multiple logistic regression models predicting use of services (dependent variable) in the general medical, psychiatric, and substance abuse sectors on the basis of need-based, predisposing, and enabling variables (independent covariates). Use of services in the medical sector was predicted by veteran status and having insurance and also by poor physical and mental functioning. Use of services in the psychiatric sector was predicted by younger age, veteran status, and poor physical and mental functioning. Use of services in the substance abuse sector was predicted by veteran status, lower mental functioning, an alcohol problem, and 30-day cocaine use.

The results of similar multiple logistic regression models for predicting the use of specific subcomponents of services in the medical, psychiatric, and substance abuse sectors were almost completely consistent with the results described above (data not shown). The only exception was that a lower level of education (p=.027) and having insurance (p=.016) also predicted use of psychiatric inpatient services.

This sample of low-demand shelter users was similar in age and duration of lifetime homelessness to samples of shelter users studied elsewhere, although this sample had fewer participants who were members of racial-ethnic minority groups (10). The high rate of participants with an alcohol problem in the current study’s low-demand sample (64%) was identical to the rate for a sample of unsheltered homeless persons in New York City (23), but it was considerably higher than the rates for two sheltered samples of homeless persons in New York City (41% and 38%) (24,25).

A comparison of the veteran and nonveteran subgroups in this study found that veterans were older, first experienced homelessness later in life, were better educated, and were more likely to have ever been married. These results are consistent with differences between veterans and nonveterans, both in the general population and other homeless samples, that have been noted in other research (26–28). Veterans were also found to be less likely than nonveterans to possess health insurance. Both groups reported extensive criminal histories, which can impede access to critical services (including housing) and other resources (29). A majority of both groups reported working for pay, but mean monthly incomes reported by either group were extremely low. The prevalence of a current alcohol problem and cocaine use and levels of physical or mental functioning were similar for veterans and nonveterans. A notable finding was that the sample’s mean physical functioning score was higher than that of the general population norm. This finding reflects the demands of living in a low-demand shelter, which require that individuals be ambulatory and be able to transport all of their belongings, tolerate the daytime conditions outside the shelter, and negotiate the admission process. Despite a comparable need for services, however, considerably more veterans than nonveterans accessed services in all three sectors.

The Behavioral Model for Vulnerable Populations proved to be a useful framework for this assessment of low-demand emergency shelter users and their use of services. Service sector use was predicted by indicators of need linked to each sector. For example, use of general medical services was linked to physical functioning, use of psychiatric services was linked to mental functioning, and use of substance abuse services was linked to an alcohol problem and recent cocaine use. These linkages point to the appropriate use of services by the study sample. Mental functioning, however, further predicted the use of medical and substance abuse services, and physical functioning predicted the use of psychiatric services, suggesting that service use is a complex phenomenon characterized by multiple viable pathways for obtaining services in various sectors. It was somewhat surprising that an alcohol problem and cocaine use did not predict the use of medical services, given the negative physical effects of problematic alcohol use and cocaine use and their prevalence in this sample.

In the multivariate model, the enabling variable of insurance predicted use of services in the medical sector. Only one-quarter of the sample reported possessing insurance, however, indicating that many individuals may not be able to afford needed medical services, limiting their use of much-needed services. The enabling factor of veteran status was also found to be predictive of use of all service sectors, which confirms that in the midst of comparable rates of non-VA service use by veterans and nonveterans, the ability to access the parallel service system provided by the VA was a significant source of the services used by the veteran subgroup.

This study had a number of methodological strengths. Users of low-demand emergency shelters, particularly veterans, have not been characterized in previous studies. Other methodological strengths included the random sampling of the study sample and the study’s construction around the Behavioral Model for Vulnerable Populations, which informed the inclusion of predisposing, enabling, and need variables relevant to the study.

This study was not without noteworthy limitations. The study sample did not represent the more general homeless population, and thus the comparisons between veterans and nonveterans and other findings do not necessarily generalize beyond the population of low-demand shelter users. The study also did not verify veteran status or the use of services or use objective diagnostic measures to assess general medical, mental, and substance abuse problems, which may have limited the accuracy of some of the data. Additional limitations were that the study did not gather information about the type of employment of study participants or the length of time between military discharge and first experience of homelessness.

Although the data collected by this study predate the more recently stated VA goal of ending homelessness, the findings have direct relevance for assessing current VA policy regarding homelessness among veterans, in particular the focus on housing veterans who are chronically homeless. With a mean lifetime duration of homelessness of almost four years and a current homeless episode duration of greater than a year, this sample of veterans from a low-demand shelter likely met the current VA definition of chronic homelessness. The finding that these veterans were relatively involved with VA health care services suggests that they were not completely alienated from the system and thus might be engaged in VA housing services, validating the VA’s current emphasis on community outreach.

Together, the fundamental similarities of the veteran and nonveteran homeless groups using low-demand shelter in this study and the finding that veterans were somewhat engaged in VA services (especially medical) suggest possibilities for improving services for homeless veterans. At the time that data were collected, local VA housing options required that individuals meet requirements regarding program participation or treatment compliance, for example, sobriety. An example of this service approach is the VA grant and per diem transitional housing program housed within the low-demand shelter facility, which required veterans to participate in substance abuse treatment or VA employment programs before being linked with permanent housing. At a conceptual level, it might be argued that homeless veterans in low-demand shelters have rejected or have been unable to comply with these programmatic demands, but their continued use of VA medical services indicates that they are still receptive to seeking care from the VA. Therefore, opportunities may still exist to engage this population, and the shelter may serve an important function by keeping this population linked to the assistance network—if only for shelter and food (30).

Although this study provided important information relative to policy and current practices, the limitations noted earlier point to some clear next steps for future research. The recent changes in VA policy on homelessness, including the adoption of the low-demand Housing First intervention, invite subsequent investigation of whether these policy shifts have affected veterans’ use of low-demand shelters. This information would be especially important considering the prevalence of cocaine use among the participants of this study, a practice that undermines housing stability, including in Housing First programs (31,32). Now that a new policy has been implemented, repeating the current study using identical methods would allow comparison of findings immediately before and after the shift in policy. One other promising direction for future research is to conduct longitudinal studies of low-demand shelter users. Examining housing trajectories as these individuals engage (or fail to engage) in housing services might provide important information about the predictors of their success in achieving permanent housing.

Given the prevalence of alcohol and drug problems and extensive criminal history among users of low-demand emergency shelters, these individuals may represent a distinct subpopulation of people who are homeless. The association between use of general medical, psychiatric, and substance abuse services by the study sample and the need for services was appropriate. However, overall use of substance abuse and psychiatric services seemed low. Despite some basic differences, the veteran and nonveteran subgroups were remarkably similar in terms of substance abuse, mental health, and general medical problems and factors that may contribute to or perpetuate homelessness (criminal histories and low income). Access to the VA service system, however, seemed to offer veterans an advantage that drove greater use of services.

This study was funded through the U.S. Department of Veterans Affairs Pre-Doctoral Social Work Research Fellowship Program. The views expressed in this article are those of the authors and do not necessarily represent the official position of the United States government.

The authors report no competing interests.