Health Risk Factors and Differences in Outcomes Between Younger and Older Veterans Using VA Transitional Housing

In 2009, President Obama and U.S. Department of Veterans Affairs (VA) Secretary Eric Shinseki announced the federal government’s goal to end veteran homelessness by 2015. While major programs for homeless veterans are being implemented, little is known about the growing subpopulation of older homeless veterans and its implications for service provision. What is known is that, compared with younger adults, older adults in the general population experience more physical, behavioral, economic, and other risk factors associated with negative health outcomes (1,2). Risk and vulnerability increase with age as people become more susceptible to chronic disease, adverse effects of multiple prescription medications, nutritional problems, loss of functional abilities, and overall life loss and change (3). In addition, older adults are more likely than others to live in poverty and to lack the financial resources to obtain needed services (4). As homelessness in general is associated with increased risk of infection or contagious disease (5) and to exposure to elements (6), injury (7,8), and malnutrition (9,10), older homeless veterans are likely to be at even greater risk of negative outcomes resulting from the combined effects of age-associated decline in health and of changes in socioeconomic status (11,12).

Prior studies have identified differences in the needs and preferences of younger and older homeless populations, with older adults more likely to have chronic medical conditions, less likely to report the need for substance abuse treatment (despite high rates of substance use), and more likely to utilize street outreach and shelter-based services (13). Other research examining age differences found that compared with younger homeless adults, fewer older homeless people reported a history of drug abuse, although no differences existed for problems related to alcohol or mental health (14). In contrast, other research has documented substantial differences in comparisons of older and younger homeless adults in overall mental health, with older homeless people having higher rates of cognitive impairment and psychiatric disorders (15).

A study comparing older homeless adults (≥45 years) from the era of the military draft with younger adults (ages 20–44) from the era of all-volunteer forces found that younger veterans more often had drug and marital problems, spent time in a psychiatric ward, and were homeless for six months or more compared with nonveterans (16). Among older adults, veterans demonstrated a lower proportion of homelessness but a higher proportion of alcohol and psychiatric problems compared with nonveterans. The mixed findings concerning age of cohort may indicate that the veterans from the all-volunteer forces will continue to have problems into older age.

The VA’s Grant Per Diem (GPD) program offers grants to community-based providers to acquire and renovate facilities to create housing for veterans. The program also provides funding to defray operational expenses and the cost of supportive services, which may include vocational, substance abuse, and educational interventions. The GPD program provides housing support for a period of up to two years and is designed as a transitional program leading to permanent housing. Community-based providers differ in their eligibility requirements for admission and the mix of support services provided. Homeless veterans typically enter the GPD program from the street or a shelter but may move into GPD programs directly from a hospital, halfway house, or other short-term housing situation (17). Each veteran entering a GPD program has an individualized treatment program that includes a plan for housing and, as required, for employment, training, general medical care, and behavioral health care. Successful program completion is defined as accomplishment of all individualized treatment program goals.

Despite the number of published studies focusing on homeless veterans, few studies have examined how aging affects veterans’ health outcomes and participation in VA homeless services. In response to this gap in our knowledge about older homeless veterans, this study examined general medical and mental health risk factors and the differences in outcomes between younger (<55) and older (≥55) veterans in the VA transitional supportive housing program. In addition, we examined how age and health affect program completion.

Methods

Sample and Data Source

Approval from the University of South Florida Institutional Review Board was obtained for the purposes of this study. We used a national data set provided by the VA Northeast Program Evaluation Center that consisted of deidentified records of veterans who were admitted into and discharged from the GPD program during federal fiscal years 2003 through 2009. The data set was derived from two sources of information, VA Forms X and D.

Individual characteristics (Form X).

Form X is a relatively comprehensive structured interview administered by program staff to veterans in the community entering specialized VA homelessness service programs. Form X captures sociodemographic (such as gender, race-ethnicity, and marital status), psychosocial, health (including any serious medical condition, hypertension, or liver disease), housing (including length of homelessness, housing arrangement within the past 30 days, and type of housing), and employment (such as recent work history and payment for services) information. In addition, Form X captures mental health diagnosis (presence of psychiatric or mental health issues, such as depression or anxiety) and substance use disorder diagnosis (including substance abuse or dependence) data reported by participants. Their responses to items about the presence of general medical, mental, or substance abuse issues were collected and treated as simple dichotomous responses (yes or no).

Outcomes (Form D).

Form D is an administrative form completed by program staff that captures treatment and outcome status at the point of discharge from the GPD program. Information is recorded for the veteran’s living situation on program exit (such as no residence, apartment, incarceration facility, living with friends, or living alone), reason for program exit (such as successful completion, illness, elopement, or program violation), employment on program exit (full-time, part-time, or student), and arrangements for VA financial benefits (such as a pension or VA service-connected disability payments).

Analyses

Data are presented as means with standard deviations, and categorical data are presented as frequencies and percentages. All analyses compared veterans under 55 years of age with those 55 and older. Age 55 was chosen as a cutpoint because it is generally used by the VA and in homelessness research to distinguish older and younger persons (18). Student’s t test or chi square analyses were performed where appropriate. To compensate for potential type I errors due to the large sample size, we calculated effect sizes for all t tests and chi square analyses. Cohen’s d effect sizes were categorized as small (.2–<.5), medium (.5–<.8), and large (≥.8) (19). Absolute values less than .2 were considered meaningless, regardless of level of significance. Cramér’s phi was calculated for chi square effect size estimates (20). Effect sizes of phi were interpreted as small (.1–<.3), medium (.3–<.5), and large (≥.5). Absolute values less than .1 were considered meaningless, regardless of statistical significance. Because our statistical interpretation was based on effect sizes, we did not correct p values for the number of analyses conducted (for example, for Bonferroni adjustments).

Initial comparisons used Student’s t test or chi square analyses, where appropriate. In addition, using unconditional logistic regression, we modeled the probability of program completion (program completion=yes) after adjusting for covariates and potential confounding variables (including age, race-ethnicity, days homeless, general physical and mental well-being, and substance abuse history). Odds ratios (ORs) and 95% confidence intervals are presented for both unadjusted and adjusted models. A univariate assessment of the impact of predictors on program completion was conducted on all relevant parameter predictors through literature review or clinical judgment. Parameters significant in univariate models were subsequently evaluated in multivariate models. Model fitting for all models was assessed by the Homer and Lemeshow goodness-of-fit test and area-under-the-curve (AUC) diagnostics. We considered p<.05, two-tailed, to be statistically significant. All analyses were conducted with SAS, version 9.1.

Results

A total of 40,820 homeless veterans were in the GPD program between 2003 and 2009 and were therefore eligible for this analysis (Table 1). Homeless veterans were predominantly male (96.1%) and nonwhite (53.3%), and their mean±SD age was 49.1. When stratified by age, older homeless veterans were more likely to be white (54.1% versus 44.3%).

A questionnaire assessing health status and a range of health issues—from oral and dental problems to tuberculosis—was administered to every homeless veteran at GPD program enrollment as part of the Form D interview. These data are presented in Table 2. In response to the item, “Do you have any serious medical problems?” a significantly greater proportion of older veterans (63.0%) responded yes, compared with 47.2% of younger veterans (p<.001). Older veterans presented with, on average, one more health problem compared with younger veterans (3.1 versus 2.3, p<.001). Health issues that affected older veterans were those traditionally associated with aging: hypertension (44.9% versus 25.2%, p<.001), heart and cardiovascular issues (21.1% versus 9.3%, p<.001), and chronic obstructive pulmonary disease or emphysema (16.8% versus 8.3%, p<.001). Across items assessing psychiatric problems, older and younger veterans were markedly similar on seven of 11 items. Older veterans were less likely to report either trouble controlling violent behavior (5.6% versus 6.6%, p<.001) or attempted suicide within the past 30 days (2.7% versus 3.3%, p<.003).

Table 3 presents a tabulated comparison of GPD program, housing, and employment outcomes, stratified by age. Time in the program and program cost were significantly different across age groups. Younger homeless veterans averaged 19 fewer days within the program (148.8 versus 130.1, p<.001) and approximately $500 less in program costs ($3,891.10 versus $3,374.20, p<.001). However, older veterans were less likely to complete the program (38.0% versus 50.5%, p<.001). Among homeless veterans with residence at discharge from the GPD program, older veterans reported a greater percentage of single-room occupancy (5.9% versus 4.6%) and living in an apartment or house (51.6% versus 48.1%), but a greater percentage of younger veterans left the program without sharing their plan for housing (17.1% versus 14.2%).

Unconditional adjusted and unadjusted ORs and 95% confidence intervals for health and psychiatric characteristics predictive of program completion are presented in Table 4. GPD program completion was defined as yes or no and served as the dependent variable in all logistic regression models. Among self-reported health issues within unadjusted models, only the presence of liver disease (OR=.88), significant skin problems (OR=1.15), and significant trauma (OR=.86) were statistically significant in successfully predicting program completion when analyses were restricted to older homeless veterans.

When unadjusted logistic regression models included all veterans, liver disease, significant skin problems and significant trauma remained significant positive predictors of program completion, and additional predictors included hypertension (OR=1.08), heart and cardiovascular issues (OR=1.08), gastrointestinal issues (OR=1.07), and seizure disorder (OR=.81). Among self-reported psychiatric issues within unadjusted models for older veterans, a previous psychiatric hospitalization (OR=.81), receiving VA general medical or psychiatric care within the past six months (OR=.91), trouble controlling violent behavior (OR=.80), and being prescribed psychological medication (OR=.92) were statistically significant negative predictors of program completion. When unadjusted logistic regression models included all veterans, those psychiatric variables remained significant predictors, and additional psychiatric variables that were significant predictors included current psychiatric or emotional problems (OR=.94), serious depression (OR=.93), hallucinations (OR=.89), and serious thoughts of suicide (OR=.95).

After adjustment for potential confounders or moderators among older homeless veterans only, significant skin problem (OR=1.16) and significant trauma (OR=.89) were the only variables that remained statistically significant predictors of program completion among self-reported health problems. Among psychiatric issues, only previous psychiatric hospitalizations (OR=.82), trouble controlling violent behavior (OR=.84), and prescribed medication for a psychological problem (OR=.92) remained statistically significant predictors of program completion.

When adjusted logistic regression models included all veterans, all self-reported health issues significantly predicted program completion, with seizure disorder (OR=.82) and significant trauma (OR=.91) as the most predictive. Among psychiatric issues, all issues were significantly associated with program completion. Among the strongest predictors were prior psychiatric hospitalization (OR=.84) and trouble controlling violent behavior (OR=.86).

Discussion

In one of the largest studies to date to assess the effects of homelessness and aging, the most significant findings were the differences in race and the number of serious general medical problems found among older veterans. Of interest is that the study covered a period of economic downturn, which may have influenced why some veterans were more inclined to enter the GPD program. However, homelessness in this period was found to decrease slightly, likely due to substantial investment by the federal government to prevent homelessness and help those who were homeless to find housing (21).

As in previous studies (16), older homeless veterans were more likely than younger homeless veterans to be white. This may simply be a reflection of generational differences within the U.S. Armed Forces; a systemic issue within the VA homelessness program, such as a failure to provide adequate services to nonwhite homeless veterans; or a lack of voluntary participation within this group for services such as these.

The U.S. armed forces have often been referred to as the most integrated institution within the United States. Current estimates suggest that racial-ethnic minority populations are significantly larger within the armed forces compared with the civilian population, accounting for approximately one-third of the military; but this was not always the case, and a vast majority of older veterans (>80%) are white. Our observed racial disparity may simply reflect an insufficient amount of time to demonstrate change. It is possible that the difference in racial groups using the GPD program is a more systemic issue in that the program is not tailored to the needs or preferences of nonwhite veterans, particularly those of older age. After conducting a survey of homeless assistance programs, the U.S. Department of Housing and Urban Development reported that 59% of homeless individuals were from racial-ethnic minority groups (40% black), but no racial-ethnic differences were observed among individuals currently homeless versus previously homeless (http://homeless.samhsa.gov).

However, among homeless people in general, the age of death for people from racial-ethnic minority groups compared with whites is substantially lower, particularly for black homeless persons (22), which may have an impact on the study findings. In addition, veteran participation in the GPD program is voluntary, and nonwhite veterans may be less inclined to take advantage of the services provided by the GPD program, perhaps because they are not aware of them, prefer alternative community services, have additional family support, or the program does not meet their needs.

In these data, older homeless veterans were less likely than younger veterans to complete the program, perhaps reflecting a mix of services offered by community providers that may be more tailored to younger veterans. Compared with younger veterans, older veterans reported on average one more significant general medical problem. In addition, older veterans were 34% more likely to report at least one serious medical problem. The greater prevalence of general medical issues among older homeless veterans appears to be driven by the number of oral or dental, hypertension, heart or cardiovascular, and orthopedic problems. The prevalence of oral and severe dental problems among homeless individuals and those in poverty has been documented (23). Dental problems place older homeless veterans at an increased risk of nutritional deficits, pain, infections, and other negative health outcomes (23,24). Increased age has been associated with an increased risk of hypertension (25) and cardiovascular issues (26), as has diminished access to health care (27) in conjunction with other health issues (28).

Between 2003 and 2009, the number of general medical conditions reported by homeless veterans increased by approximately 24%, regardless of age. Further, the percentage of homeless veterans reporting at least one serious medical condition increased 7% during the study period. These increases have been driven mainly by oral and dental, hypertension, and orthopedic conditions. In addition, although some conditions decreased slightly from 2003 to 2009 among homeless veterans under age 55, all general medical conditions except significant skin problems increased for homeless veterans age 55 and older.

The proportion of homeless veterans agreeing to the question “Do you think that you have any current psychiatric or emotional problem(s) other than alcohol or drug?” increased approximately 20% from 2003 to 2009. Both younger and older homeless veterans reported similar rates of psychiatric conditions for each year and trends over time. A slight increase for both groups was observed with the exception of thoughts of suicide, which decreased by 10%. However, the reported use of prescribed medication increased markedly (24%) for both age groups during the study period. In addition, older homeless veterans with psychiatric issues were significantly less likely than other veterans to complete the program, indicating the need to improve delivery of services and better engage this population in mental health care.

Potential limitations with this project were those associated with any use of secondary analyses and interviews utilizing convenience sample and self-report techniques. Limitations associated with secondary analyses include potential coding and data entry errors, lack of specificity to a targeted population, and relevance at time of final analysis. Limitations associated with the use of data generated from Forms X and D may result from the data collection methodology. Data from homeless veterans agreeing to participate and provide self-reported information via Form X may not generalize to other homeless veterans but rather the subset of homeless veterans participating in the GPD program compared with homeless veterans who participate in non-VA community-based programs. In addition, Form X and Form D data do not allow for the comparison of older homeless veterans with nonhomeless veterans to determine differences or similarities in general medical and health needs. Further, this study may not generalize to homeless nonveterans or veterans in other countries. The format of Form X and Form D items limit the comprehensiveness of the admission and outcome evaluations, using primarily self-report (Form X) and observation (Form D) formats, often with simple dichotomous response options.

Conclusions

Homelessness amplifies existing vulnerabilities and places older adults at increased risk of morbidity, mortality, and other adverse events, such as theft and violence. Older homeless veterans are at increased risk of serious medical issues (including orthopedic and cardiovascular problems). Although homelessness has been more prevalent among younger adults, recent data reveal that homelessness among older adults is steadily increasing (18). Current data regarding the homeless veteran population suggest an increasing proportion of the homeless population consists of individuals over 55. Coupled with their increasing needs for both general medical and mental health services over time, this older age group is especially vulnerable to experiencing negative consequences related to homelessness. VA programs and services may require tailoring to meet older veterans’ cumulative and complex needs.