- Email: [email protected]
Demographic Characteristics Associated with Homelessness and Risk Among Female and Male Veterans Accessing VHA Outpatient Care
Women's Health Issues 25-1 (2015) 42–48
www.whijournal.com
Original article
Demographic Characteristics Associated with Homelessness and Risk Among Female and Male Veterans Accessing VHA Outpatient Care Ann Elizabeth Montgomery, PhD a,*, Melissa E. Dichter, PhD, MSW b, Arwin M. Thomasson, PhD b, Xiaoying Fu, MS b, Christopher B. Roberts, MPH b a b
U.S. Department of Veterans Affairs, National Center on Homelessness Among Veterans, Philadelphia, Pennsylvania U.S. Department of Veterans Affairs, Center for Health Equity Research and Promotion, Philadelphia, Pennsylvania
Article history: Received 27 January 2014; Received in revised form 2 July 2014; Accepted 8 October 2014
a b s t r a c t Background: This study explored demographic influences on veterans’ reports of homelessness or imminent risk of homelessness with a particular focus on gender. Methods: We analyzed data for a cohort of veterans who responded to the U.S. Department of Veterans Affairs (VA), Veterans Health Administration (VHA) universal screener for homelessness and risk during a 3-month period. Multinomial mixed effects modelsdstratified by genderdpredicted veterans’ reports of homelessness or risk based on age, race, marital status, and receipt of VA compensation. Findings: The proportion of positive screensdhomelessness or riskdwas 2.7% for females and 1.7% for males. Women more likely to report being at risk of homelessness were aged 35 to 54 years, Black, and unmarried; those more likely to experience homelessness were Black and unmarried. Among male veterans, the greatest predictors of both homelessness and risk were Black race and unmarried status. Among both genders, receiving VA disability compensation was associated with lesser odds of being homeless or at risk. Conclusions: The findings describe the current population of veterans using VHA health care services who may benefit from homelessness prevention or intervention services, identify racial differences in housing stability, and distinguish subpopulations who may be in particular need of intervention. Interventions to address these needs are described. Published by Elsevier Inc.
Homelessnessda longstanding “public problem” in the United States (Stern, 1984)dis associated with poor health, economic, and social outcomes (Burt, 2001; Hawkins & Abrams, 2007; Hwang, 2001; Lee & Farrell, 2003; Lee & Greif, 2008; Wolitski, Kidder, & Fenton, 2007; Zerger, 2002). At a single point in time in January 2014, more than 578,000 people in the United States were homeless, staying in shelters or other temporary housing, or on the streets or other places not meant for
This study was funded by the U.S. Department of Veterans Affairs, National Center on Homelessness Among Veterans and Center for Health Equity Research and Promotion (Grant no. LIP 72-061). The contents of this article do not necessarily represent the views of the U.S. Department of Veterans Affairs or the U.S. Government. * Correspondence to: Ann Elizabeth Montgomery, PhD, U.S. Department of Veterans Affairs, National Center on Homelessness Among Veterans, 4100 Chester Avenue, Suite 201, Philadelphia, PA 19146. Phone: 215-823-5800x5067; Fax: 215-222-2591. E-mail address: [email protected] (A.E. Montgomery). 1049-3867/$ - see front matter Published by Elsevier Inc. http://dx.doi.org/10.1016/j.whi.2014.10.003
human habitation; approximately 11% of homeless adults were veterans (U.S. Department of Housing and Urban Development [HUD], 2014). Over the course of a year, roughly 1 in every 156 veterans will experience homelessness (HUD, 2013). Although the rate of veteran homelessness has declined during recent years, the problem continues to impact a substantial proportion of the veteran population, including younger veterans and female veterans (Fargo et al., 2012): During federal fiscal year 2012, 8.5% of veterans who accessed emergency shelter or transitional housing were younger than 30 years and almost 7.8% were female (HUD, 2013). The U.S. Department of Veterans Affairs (VA) has intensified its efforts to address homelessness among veterans, as evidenced by its 5-year plan to end homelessness as well as substantial new investments in programs intended to prevent and end homelessness (U.S. Interagency Council on Homelessness, 2010). VA’s strategy places particular emphasis on homelessness prevention (U.S. Interagency Council on Homelessness, 2013), which is in line
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
with a broader shift in homeless assistance policies in the United States that increasingly favor prevention and rapid rehousing approaches (Burt, Pearson, & Montgomery, 2005, 2007; Culhane, Metraux, & Byrne, 2011); that is, either ensuring that a homeless episode does not occur or, if it does, ending it as rapidly as possible. In support of this prevention-oriented strategy, the VA established the Supportive Services for Veteran Families (SSVF) program in 2011 to promote housing stability among homeless and at-risk veterans and their families. Through this program, VA awards grants to nonprofit organizations that provide supportive services to eligible very low-income veteran families who are either recently homeless or at risk, including outreach, case management, assistance in obtaining VA benefits, and help in accessing and coordinating other public benefits. SSVF grantees can also provide temporary financial assistance to cover rent, utilities, security deposits, and moving costs (VA, 2013a). To improve the VA’s ability to identify veterans who are at risk of homelessness, or experiencing homelessness but are not accessing services through the Veterans Health Administration’s (VHA) Homeless Program, the VA National Center on Homelessness Among Veterans developed the Homelessness Screening Clinical Reminder (HSCR) to conduct an ongoing, universal screen for homelessness and risk among veterans accessing outpatient health care services (Montgomery, Fargo, Byrne, Kane, & Culhane, 2013; Montgomery, Fargo, Kane, & Culhane, 2014). The HSCR is embedded in veterans’ medical records; when they present for services, health care staff ask veterans the screening questions and record their responses in the medical record. The objective of this national, health system-based screening instrument is to enhance the rapid identification of veterans who have very recently become homeless or are at imminent risk of homelessness, and to ensure that they are referred for appropriate assistance. Implemented in October 2012, the HSCR is the first attempt by a health system or other national-level service providerdin the United States or elsewheredto systematically assess whether patients are homeless or at risk. An initial analysis of data collected through the HSCR between October 1, 2012, and January 10, 2013, identified the following rates of housing instability: 0.9% of veteran respondents reported current homelessness, 1.2% reported being at risk, and 97.9% screened negative for both. This analysis found that, compared with men, significantly more female veterans reported being homeless (1.1% for women vs. 0.9% for men) and at risk (1.9% for women vs. 1.2% for men). This initial analysis identified significant gender-based differences in positive screens, calling for future research to inform outreach, targeting, and homelessness prevention interventions (Montgomery et al., 2013). The objective of the present study was to explore the demographic associations of veterans’ reports of being homeless or at risk with a particular focus on how this varies by gender; therefore, we conducted analyses separately for men and women (Bird, 2013). Because VHA’s patient population is predominantly maledcomprising 93.2% of the cohort studied heredgender differences may not be particularly evident when conducting analyses that include all veterans rather than conducting gender-specific analyses. Additionally, the literature has identified differences between the characteristics associated with homelessness by gender and it may be helpful to identify the particular characteristics for each of these groups (Blackstock, Haskell, Brandt, & Desai, 2012; HUD & VA, 2010, 2011; Leda, Rosenheck, & Gallup, 1992; Tsai, Rosenheck, & McGuire, 2012).
43
Methods This study was approved by the local institutional review board. Data were extracted from the VA Corporate Data Warehouse, a repository of data from VHA clinical and administrative systems (VA Information Resource Center, 2012). The VA Corporate Data Warehouse contains records of all inpatient and outpatient stays occurring at any VA medical facility, as well as demographic information about patients receiving services. Sample The cohort for this study is comprised of all veterans who completed the HSCR during a VHA outpatient visit between November 1, 2012, and January 31, 2013, excluding those who declined screening or had missing, incomplete, or duplicate responses to the HSCR (n ¼ 23,177; 1.4%). In addition, patients with no outpatient visit 6 months before screening (n ¼ 37,690; 2.3%) were also excluded, because this study focuses on individuals who were actively engaged in VHA care. In total, 4.4% of females and 3.6% of males were excluded from the analyses. Measures The dependent variable is defined by veterans’ responses to the following 2 questions in the HSCR: 1. In the past 2 months, have you been living in stable housing that you own, rent, or stay in as part of a household? (“No” response indicates veteran is positive for homelessness and screening ends.) 2. Are you worried or concerned that in the next 2 months you may not have stable housing that you own, rent, or stay in as part of a household? (“Yes” response indicates veteran is at risk for homelessness.) Veterans are categorized as either homeless, at risk, or negative for both homelessness and risk. If a veteran is positive for homelessness, risk is not assessed; if a veteran is positive for neither homelessness nor risk, the veteran is considered negative. The HSCR is administered annually with all veterans who access outpatient health care services with the exception of those who received homeless assistance through VA at any time during the previous 6 months; those who report already receiving homeless services or assistance (unknown to VA at the time of screening); long-term residents of a nursing home or long-term care facility; and veterans who decline or are unable to answer. Veterans screening positive are rescreened semiannually. Veterans in palliative or long-term care are rescreened biannually. Once a veteran screens negative during 3 consecutive screens, the veteran is only screened biannually. The independent variables include: Age, a continuous measure categorized into 5 groups; Race, which we collapsed into 3 categories (White, Black, and other); Marital status, composed of 6 categories that we collapsed to create a binary variable (married/unmarried); and VHA Enrollment Priority Group, which indicates the extent to which a veteran is receiving compensation owing to a disability incurred during military service as well as whether a veteran is very low income. We collapsed the groups as follows: 50% or greater service-connected (receiving the
44
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
most VA financial compensation); less than 50% serviceconnected disability (less, but some, VA financial compensation); no service-connected disability but Medicaid eligible (no VA financial compensation and low-income); and no service-connected disability but not Medicaid eligible (no VA financial compensation and not low-income; U.S. Department of Veterans Affairs, 2012). Analysis We conducted all analyses using SAS version 9.2 (SAS Institute, Inc., 2011), including multinomial mixed effects models, with a facility-specific random intercept. Univariate models utilized all available data (i.e., patients with missing values for a given covariate were not excluded from other univariate models). We estimated four multivariable modelsdfor at-risk females, homeless females, at-risk males, and homeless malesdwhich included only patients with complete data; we chose this approach to describe the cohort as thoroughly as possible. The multivariable models contained the covariate of interest, as well as age and race. Results Sample Description The study sample included 1,582,125 veterans: 107,504 (6.8%) were women and 1,474,621 (93.2%) were men. Table 1 describes the demographic characteristics of the veterans in the sample by gender. The majority of women (62.9%) were younger than 55 years, whereas the majority of men (79.7%) were older than 55 years. The majority of women (58.4%) and men (69.9%) reported identifying as White, followed by Black. Approximately one third of women veterans and almost two thirds of men veterans reported being married. Nearly two thirds of female veterans received VA compensation for a service-connected disability, whereas one half of men received VA compensation for a service-connected disability. Overall, 0.8% of the veterans who accessed VHA outpatient health care during the 3-month study period reported current homelessness, 1.1% reported being at risk, and the remaining 98.2% were negative for both homelessness and risk. The rate of positive screens among women was 2.7% compared with 1.7% for men. (These rates are slightly different than those reported in the initial analysis of HSCR data owing to differences in the inclusion criteria for the cohort [Montgomery et al., 2013]). Homelessness and Risk among Female Veterans Table 2 presents the unadjusted and adjusted odds ratios for the association of women veterans screening positive for homelessness or being at risk and demographic characteristics. Compared with female veterans younger than 35, those in the 45- to 54-year age range had the greatest odds of screening positive for risk, whereas those aged 65 or older were more than 3 times less likely to screen positive for risk. Female veterans who identified as Black or other race (American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, Mixed) were more likely than Whites to report risk of homelessness. Marital status was also a significant predictor of female veterans reporting being at risk of homelessness: Those who were unmarried had almost twice the odds of reporting homelessness risk as their married counterparts. Finally, receiving VA compensation for a service-connected disabilitydcompared
Table 1 Characteristics of Veterans Health Administration Outpatients, by Gender: November 1, 2012–January 30, 2013 Female n
Male %
n
Total %
n
%
Total number 107,504 6.8 1,474,621 93.2 1,582,125 100.0 Age (y) <35 19,495 18.1 69,973 4.7 89,468 5.6 35–44 19,092 17.8 75,323 5.1 94,415 5.6 45–54 28,977 27.0 153,854 10.4 182,831 11.6 55–64 25,967 24.2 408,636 27.7 434,603 27.5 65 13,973 13.0 766,832 52.0 780,805 49.4 Missing 3 0.0 3 0.0 Race White 62,730 58.4 1,030,549 69.9 1,093,279 69.1 Black 27,008 25.1 196,451 13.3 223,459 14.1 American Native 743 0.7 6,841 0.5 7,584 0.5 Asian 830 0.8 8,439 0.6 9,269 0.6 Hawaiian 998 0.9 10,375 0.7 11,373 0.7 Mixed 1,464 1.4 14,306 1.0 15,770 1.0 Missing 13,731 12.8 207,660 14.1 221,391 14.0 Marital status Married 35,464 33.0 860,976 58.4 896,440 56.7 Divorced 36,714 34.2 324,674 22.0 361,388 22.8 Never married 23,319 21.7 142,573 9.7 165,892 10.5 Separated 4,832 4.5 43,693 3.0 48,525 3.1 Single 347 0.3 2,848 0.2 3,195 0.2 Widowed 6,325 5.9 95,099 6.4 101,424 6.4 Missing 503 0.5 4,758 0.3 5,261 0.3 VA Enrollment Priority Group 50% SC disability 39,629 36.9 427,605 29.0 467,234 29.5 <50% SC disability 24,851 23.1 285,943 19.4 310,794 19.6 Not SC/Medicaid 25,327 23.6 368,920 25.0 394,247 24.9 eligible Not SC/Not Medicaid 17,560 16.3 391,506 26.5 409,066 25.9 eligible Missing 137 0.1 647 0.0 784 0.0 Response to HSCR Homeless 1,033 1.0 10,876 0.7 11,909 0.8 At risk 1,809 1.7 14,845 1.0 16,654 1.1 Negative 104,662 97.4 1,448,900 98.3 1,553,562 98.2 Abbreviation: SC, service-connected.
with not having a service-connected disability and not being Medicaid eligibledwas associated with lower odds of homelessness risk among women veterans. Although younger age was not associated with a positive screen for risk, women veterans older than 55 years had significantly lower odds of reporting homelessness compared with those younger than 35. Women veterans who identified as Black were more likely than Whites to report homelessness and unmarried female veterans were more than twice as likely as married female veterans to report homelessness. Again, receipt of benefits related to a service-connected disability was associated with lower odds of homelessness among women. The profiles of women veterans who reported being at risk were somewhat distinct from those reporting homelessness. Specifically, women between 35 and 54 years were more likely to report being at risk than those younger than 35 years; however, this did not hold true for homelessness. Generally, older age was associated with a lower odds of both homelessness and risk among women. Women veterans identifying as Black were more likely than those identifying as White to screen positive for either homelessness or risk. Unmarried women had a greater likelihood of reporting homelessness than being at risk. Marital status and older age were the predictors for both homelessness and risk with the greatest effect size. Finally, both homelessness
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
45
Table 2 Unadjusted and Adjusted Odds of Screening Positive for Homelessness or at Risk among Female VHA Outpatients: November 1, 2012–January 30, 2013 Effect
Age (y) 35–44 45–54 55–64 65 Race Other Black Marital status – not married VA Enrollment Priority Group 50% SC disability <50% SC disability Not SC/not Medicaid eligible
Reference
At Risk (n ¼ 1,809)
Homeless (n ¼ 1,033)
OR (95% CI)
AOR (95% CI)
OR (95% CI)
AOR (95% CI)
1.20 1.30 1.05 0.31
1.21 1.24 0.96 0.27
0.92 1.07 0.81 0.27
0.98 1.03 0.73 0.25
<35 (1.03, (1.13, (0.91, (0.24,
1.40)* 1.50)* 1.21) 0.40)*
(1.03, (1.07, (0.82, (0.20,
1.42)* 1.43)* 1.12) 0.35)*
(0.76, (0.90, (0.68, (0.19,
1.12) 1.26) 0.98)* 0.37)*
(0.80, (0.86, (0.60, (0.18,
1.21) 1.24) 0.89)* 0.36)*
White
Married Not SC/Medicaid eligible
1.34 (1.06, 1.69)* 1.58 (1.41, 1.77)* 1.89 (1.69, 2.12)*
1.30 (1.03, 1.63)* 1.42 (1.27, 1.59)* 1.71 (1.51, 1.94)*
1.23 (0.90, 1.69) 1.63 (1.40, 1.88)* 2.80 (2.36, 3.33)*
1.19 (0.87, 1.63) 1.45 (1.24, 1.68)* 2.38 (1.98, 2.87)*
0.53 (0.47, 0.59)* 0.71 (0.62, 0.80)* 0.57 (0.49, 0.66)*
0.47 (0.42, 0.54)* 0.69 (0.60, 0.79)* 0.69 (0.59, 0.81)*
0.46 (0.39, 0.53)* 0.58 (0.50, 0.69)* 0.42 (0.34, 0.52)*
0.44 (0.37, 0.52)* 0.55 (0.46, 0.66)* 0.51 (0.41, 0.64)*
Abbreviations: AOR, adjusted odds ratio; OR, odds ratio; SC, service-connected. Notes. Reference category for outcomes is negative. AOR adjust for race and age. * p < .0001.
and risk were less likely among women veterans receiving benefits related to service-connected disabilities. Homelessness and Risk among Male Veterans Table 3 presents the unadjusted and adjusted odds ratios for the association of male veterans screening positive for homelessness or risk and a set of demographic characteristics. Among male veterans, those aged 55 years and older were significantly less likely to screen positive for risk. Male veterans identifying as Black or some other race were more likely than their White counterparts to report homelessness risk and unmarried male veterans were around twice as likely to report homelessness risk compared with those who were married. Receiving financial compensation for a service-connected disability was associated with a lower homelessness risk for men; those without a serviceconnected disability who were not low income were least likely to screen positive for risk. Male veterans older than age 55 years were less likely than their younger counterparts to report current homelessness; those between 35 and 54 years were slightly more likely than those younger than 35 to report homelessness. Compared with veterans who were White, those identifying as Black had approximately a 2 times higher odds of reporting homelessness. Unmarried male veterans were more than 3 times more likely than married male veterans to report homelessness. Finally, service connectedness was associated with lower odds of reporting homelessness. The only demographic variable that seemed to have a different pattern regarding the likelihood of being homeless or at risk among men was marital status. Although there was a significant, positive relationship between unmarried status and both homelessness and at risk, there was a stronger relationship with homelessness than at risk. Discussion This study identified the demographic correlates of current homelessness as well as risk of imminent homelessness among a large cohort of female and male veterans receiving care through the VHA. The analyses conducted here identified that, for women
veterans, being in the middle age range (35–54 years), Black race, and being unmarried were associated with higher odds of screening at risk, whereas only Black race and unmarried status were associated with higher odds of screening positive for homelessness. The effect size of unmarried status was approximately 30% greater for predicting homelessness than risk. Among male veterans, although older age seemed to be associated with lower odds of screening positive for risk or homelessness, the greatest predictors of both were Black race and unmarried status. Among both men and womendand for both outcomesdhaving access to compensation related to a serviceconnected disability (compared with being non-service connected and Medicaid eligible) was associated with lower odds of homelessness and risk. These findings shed light on the current population of veterans using VHA health care services who may benefit from homelessness prevention or intervention services and inform our understanding of housing stability in this population. Uniquely, these findings can be useful for early intervention efforts to prevent homelessness among those individuals who are at imminent risk but not (yet) experiencing homelessness. Women aged 35 to 54 years have an increased odds of being at risk for homelessness, compared with their younger peers, but not of current homelessness; the youngest group of women veterans (those younger than 35 years) may be most likely to return to their parents’ homes after military service, thus affording at least temporary housing stability. The vulnerable, middle-age population of women may represent those who are parenting children and adolescents or caring for aging parents, which may increase their household expenses while limiting their employment opportunities during their peak earning years. Women older than 54 years may be more stably housed and partnered and/or employed or retired and less likely than their younger peers to experience intimate partner violence, a common pathway into homelessness among women (Baker, Cook, & Norris, 2003; Hamilton, Poza, & Washington, 2011; Pavao, Alvarez, Baumrind, Induni, & Kimerling, 2007; Washington et al., 2010). Black race and unmarried status were strong predictors of homelessness and risk, particularly among men. Consistent with previous findings, we found that Black race was related to increased risk of housing instability (HUD, 2013). Similarly,
46
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
Table 3 Unadjusted and Adjusted Odds of Screening Positive for Homelessness or at Risk among Male VHA Outpatients: November 1, 2012 - January 30, 2013 Effect
Age (y) 35–44 45–54 55–64 65 Race Other Black Marital status – not married VA Enrollment Priority Group 50% SC disability <50% SC disability Not SC/not Medicaid eligible
Reference
At Risk (n ¼ 14,845)
Homeless (n ¼ 10,876)
OR (95% CI)
AOR (95% CI)
OR (95% CI)
AOR (95% CI)
1.10 1.23 0.73 0.18
1.12 1.10 0.69 0.20
0.97 1.25 0.68 0.16
1.05 1.09 0.64 0.18
<35 (1.02, (1.15, (0.68, (0.17,
1.18)* 1.31)* 0.77)* 0.19)*
(1.04, (1.03, (0.64, (0.18,
1.22)* 1.18)* 0.73)* 0.21)*
(0.89, (1.16, (0.64, (0.15,
1.05) 1.34)* 0.73)* 0.17)*
(0.96, (1.01, (0.60, (0.17,
1.15) 1.18)* 0.69)* 0.20)*
White
Married Not SC/Medicaid eligible
1.32 (1.21, 1.45)* 2.31 (2.21, 2.41)* 2.53 (2.44, 2.62)*
1.18 (1.07, 1.29)* 1.66 (1.59, 1.73)* 1.74 (1.67, 1.80)*
1.33 (1.20, 1.48)* 2.59 (2.47, 2.72)* 5.04 (4.81, 5.28)*
1.23 (1.11, 1.37)* 1.78 (1.69, 1.87)* 3.31 (3.14, 3.48)*
0.41 (0.40, 0.43)* 0.66 (0.63, 0.69)* 0.37 (0.35, 0.39)*
0.42 (0.40, 0.44)* 0.67 (0.64, 0.70)* 0.53 (0.50, 0.56)*
0.31 (0.30, 0.33)* 0.48 (0.46, 0.51)* 0.31 (0.29, 0.32)*
0.37 (0.35, 0.40)* 0.54 (0.51, 0.58)* 0.51 (0.48, 0.54)*
Abbreviations: AOR, adjusted odds ratio; OR, odds ratio; SC, service connected. Notes. Reference category for outcomes is negative. AOR adjust for race and age. * p < .0001.
earlier studies have found higher rates of homelessness among those who are unmarried (Jencks, 1994); however, in the sample studied herein, there are likely different explanations for unmarried status posing risk to women and men, including the amount and type of social support each receives from their spouse. The women in the sample were younger and more likely to be unmarried, whereas the men in the sample were older and more likely to be married. Unmarried status for the women may signify single parenthood, and for both genders it may signify a lack of social support. Several limitations of this study should be considered. First, the study included only veterans who accessed outpatient health care services through the VHA at least twice during the 6-month period before the visit at which they responded to the HSCR; therefore, the rates reported herein cannot be considered prevalence rates of homelessness or risk among all veterans. Second, the constructs of homelessness and risk are based on veterans’ self-report and their own perceptions of whether they are experiencing unstable housing, which may vary by individual. Third, because the HSCR is a national-level screen administered by many types of staff throughout VHA, there may be distinctions in how veterans are screened; variations in screening practices may influence response. Although the data were based on veterans’ self-report, and analyses control for location of screening, the variations in screening procedures may be systematic, leading to inaccurate rates of homelessness and risk as well as relationships between these rates and veterans’ demographic characteristics. In addition, the administrative data may include possible errors in documentation of marital status and race. To date, the research literature has provided little explanation for findings of differences between female and male veteran homelessness, and veteran and non-veteran homelessness. Women, in particular, may be vulnerable to housing instability owing to social factors that could not be measured within the data used for this study. For example, several researchers have identified violence and abuse as precursors to homelessness among women in general (Baker et al., 2003; Browne & Bassuk, 1997; Goodman, 1991; Pavao et al., 2007; Rollins, Saris, & Johnston-Robledo, 2001) and female veterans in particular (Hamilton et al., 2011; Washington et al., 2010). Other factors related to homelessness among veteran and non-veteran women include adolescent childbearing, low-wage employment or
unemployment, lack of affordable housing, and disability, poor health, and diagnosis of posttraumatic stress disorder or an anxiety disorder (Goodman, 1991; Hamilton et al., 2011; Rollins et al., 2001; Washington et al., 2010). Future research should include as predictive variables sources of risk beyond demographic characteristics. Implications for Policy and/or Practice The veterans in this study who reported current homelessness or risk represent the population receiving care in the VHA who have not been reached with homeless services (those already receiving such services are excluded from the screening) but may benefit from such programs. Not only do the findings presented herein begin to identify specific subpopulations of veterans who may be in particular need of intervention, but they also hint at particular interventions specific to each subpopulation. Younger, unmarried females; older, unmarried males; and veterans who do not seem to have access to income through VA compensation are the most likely to report current homelessness or imminent risk. We also continue to see racial disparities with increased risk among non-White populations. Efforts need to be made to target these subpopulations, not only for screening and assessment, but for interventions that prevent homelessness or quickly end a current episode of homelessness. Younger, unmarried female veterans may be best assisted through access to childcare, educational resources, temporary financial assistance for rent or utilities, and perhaps supportive housing, depending on other needs such as mental or behavioral health conditions. Although the SSVF program provides these resources throughout hundreds of communities in the United Statesdand has provided services to families (30% of all households served in federal fiscal year 2012) and a disproportionate number of female veterans (15% of all veterans served in federal fiscal year 2012)dperhaps greater outreach and education about these services is necessary within the health care system (VA, 2013b). In addition, the provision of services to the children of younger, unmarried women may prevent involvement in other social services systems, such as foster care, in the future (Culhane, Metraux, Byrne, Stino, & Bainbridge, 2013). Efforts to provide prevention services to unmarried, older male veterans who are homeless or at risk should be increased,
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
both in the community and within the health care system. Although not a focus of the present study, we note that, as these veterans continue to age, they may experience chronic medical conditions that require additional support, both medically and socially (Culhane et al., 2013). Programs that enable veterans to remain in housing, affordably, as their needs increase may increase their housing stability. Although we do not have information about employment status or receipt of other income or benefits among veterans screening positive for homelessness and at risk, the findings from the present study indicate that receiving some financial compensation is protective against housing instability. Efforts to increase income among veterans who are homeless or at riskdthrough jobs programs or application for benefitsdcould increase the likelihood that veterans attain or maintain permanent housing. Current work is considering the role that physical, mental, and behavioral health play in female and male veterans’ report of homelessness and risk as well as the likelihood of veterans to access VA services after a positive screen. Conducting these analyses from a gender-based perspective will increase our understanding about the specific needs of female and male veterans and how interventions may be tailored. To determine which interventions are most effective at mitigating homelessness and risk, for which populations, future studies could test the effectiveness of various interventions using randomized designs and assess for demographic variation. Further qualitative research to uncover the ways in which various factors may influence veterans’ experiences of housing instability will help to inform our understanding of pathways to homelessness and potential points of intervention for this population.
References Baker, C. K., Cook, S. L., & Norris, F. H. (2003). Domestic violence and housing problems: A contextual analysis of women’s helpseeking, received informal support, and formal system response. Violence Against Women, 9(7), 754–783. Bird, C. E. (2013). Spotlight on women’s health: Gender-based analysis. Retrieved 10/9/2013 from. http://www.hsrd.research.va.gov/for_researchers/cyber_ seminars/archives/video_archive.cfm?SessionID¼755. Blackstock, O. J., Haskell, S. G., Brandt, C. A., & Desai, R. A. (2012). Gender and the use of Veterans Health Administration homeless services programs among Iraq/Afghanistan Veterans. Medical Care, 50(4), 347–352. Browne, A.,, & Bassuk, S. S. (1997). Intimate violence in the lives of homeless and poor housed women: Prevalence and patterns in an ethnically diverse sample. American Journal of Orthopsychiatry, 67, 261–278. Burt, M. R. (2001). Helping America’s homeless: Emergency housing or affordable housing? Washington, DC: The Urban Institute. Burt, M. R., Pearson, C. L., & Montgomery, A. E. (2005). Strategies for preventing homelessness. Washington, DC: U.S. Department of Housing and Urban Development. Burt, M. R., Pearson, C. L., & Montgomery, A. E. (2007). Community-wide strategies for preventing homelessness: Recent evidence. Journal of Primary Prevention, 28, 213–228. Culhane, D. P., Metraux, S., & Byrne, T. H. (2011). A prevention-centered approach to homelessness assistance: A paradigm shift? Housing Policy Debate, 21(2), 295–315. Culhane, D. P., Metraux, S., Byrne, T. H., Stino, M., & Bainbridge, J. (2013). The age structure of contemporary homelessness: Evidence and implications for public policy. Analyses of Social Issues and Public Policy, 13(1), 228–244. Fargo, J. D., Metraux, S., Byrne, T. H., Munley, E., Montgomery, A. E., Jones, H., et al. (2012). Prevalence and risk of homelessness among U.S. Veterans: A multisite investigation. Preventing Chronic Disease, 9, 110–112. Goodman, L. (1991). The prevalence of abuse in the lives of homeless and housed poor mothers: A comparison study. American Journal of Orthopsychiatry, 61, 489–500. Hamilton, A. B., Poza, I., & Washington, D. L. (2011). “Homelessness and trauma go hand-in-hand”: Pathways to homelessness among women Veterans. Women’s Health Issues, 21(4S), S203–S209.
47
Hawkins, R. L., & Abrams, C. (2007). Disappearing acts: The social networks of formerly homeless individuals with co-occurring disorders. Social Science & Medicine, 65(10), 2031–2042. Hwang, S. W. (2001). Homelessness and health. Canadian Medical Association Journal, 164(2), 229–233. Jencks, C. (1994). The homeless. Cambridge, MA: Harvard University Press. Leda, C., Rosenheck, R. A., & Gallup, P. (1992). Mental illness among homeless female Veterans. Hospital Community Psychiatry, 43(10), 1026–1028. Lee, B. A.,, & Farrell, C. R. (2003). Buddy, can you spare a dime? Urban Affairs Review, 38(3), 299–324. Lee, B. A.,, & Greif, M. J. (2008). Homelessness and hunger. Journal of Health and Social Behavior, 49(1), 3–19. Montgomery, A. E., Fargo, J. D., Byrne, T. H., Kane, V., & Culhane, D. P. (2013). Universal screening for homelessness and risk for homelessness in the Veterans Health Administration. American Journal of Public Health, 103(S2), S210–S211. Montgomery, A. E., Fargo, J. D., Kane, V., & Culhane, D. P. (2014). Development and validation of an instrument to assess imminent risk of homelessness among Veterans. Public Health Reports, 129, 439–447. Pavao, J., Alvarez, J., Baumrind, N., Induni, M., & Kimerling, R. (2007). Intimate partner violence and housing instability. American Journal of Preventive Medicine, 32(2), 143–146. Rollins, J. H., Saris, R. N., & Johnston-Robledo, I. (2001). Low-income women speak about housing: A high-stakes game of musical chairs. Journal of Social Issues, 57, 277–298. SAS Institute Inc. (2011). SAS/STATÒ 9.3 user’s guide. Cary, NC: Author. Stern, M. J. (1984). The emergence of the homeless as a public problem. Social Service Review, 58(2), 291–301. Tsai, J., Rosenheck, R. A., & McGuire, J. F. (2012). Comparison of outcomes of homeless female and male veterans in transitional housing. Community Mental Health Journal, 48(6), 705–710. U.S. Department of Housing and Urban Development (HUD). (2014). The 2014 Annual Homeless Assessment Report (AHAR) to Congress. Washington, DC: Author. U.S. Department of Housing and Urban Development (HUD). (2013). The 2012 Annual Homeless Assessment Report to Congress. Washington, DC: Author. U.S. Department of Housing and Urban Development (HUD), & U.S. Department of Veterans Affairs (VA). (2011). Veteran homelessness: A supplemental report to the 2010 annual homeless assessment report to Congress. Washington, DC: Author. U.S. Department of Housing and Urban Development (HUD), & U.S. Department of Veterans Affairs (VA). (2010). Veteran homelessness: A supplemental report to the 2009 annual homeless assessment report to Congress. Washington, DC: Author. U.S. Department of Veterans Affairs (VA). (2012). Priority Groups Table. Retrieved 1/16/2014 from: http://www.va.gov/healthbenefits/resources/priority_ groups.asp. U.S. Department of Veterans Affairs (VA). (2013a). Department of Veterans Affairs Supportive Services for Veteran Families (SSVF) Program Guide. Retrieved 1/16/2014 from: http://www.va.gov/HOMELESS/docs/SSVF/SSVFUniversity/ ProgramRequirementsPage/Program_Guide_Revised_12_19_2013_FINAL.pdf. U.S. Department of Veterans Affairs (VA). (2013b). Effectiveness of permanent housing program FY 2012 report. Retrieved 1/23/2014 from: http://www.va.gov/ HOMELESS/docs/SSVF/Effectiveness_of_SSVF_Program_Report_FY2012.pdf. U.S. Interagency Council on Homelessness. (2010). Opening doors: Federal strategic plan to prevent and end homelessness. Washington, DC: Author. U.S. Interagency Council on Homelessness. (2013). Ending homelessness among veterans: A report by the United States Interagency Council on Homelessness. Retrieved 1/16/2014 from: http://usich.gov/resources/uploads/asset_library/ USICH_Ending_Homelessness_Among_Veterans_Rpt_February_2013_FINAL. pdf. VA Information Resource Center. (2012). VA Corporate Data Warehouse (CDW): A VIReC resource guide. Retrieved 10/15/2013 from: http://vaww.virec.research. va.gov/CDW/RG-CDW-CY12-RA.pdf. Washington, D. L., Yano, E. M., McGuire, J., Hines, V., Lee, M., & Gelberg, L. (2010). Risk factors for homelessness among women veterans. Journal of Health Care for the Poor and Underserved, 21, 82–91. Wolitski, R. J., Kidder, D. P., & Fenton, K. A. (2007). HIV, homelessness, and public health: Critical issues and a call for increased action. AIDS and Behavior, 11, 167–171. Zerger, S. (2002). A preliminary review of literature: Chronic medical illness and homelessness. Nashville, TN: National Health Care for the Homeless Council.
Author Descriptions Ann Elizabeth Montgomery, PhD, conducts research with the U.S. Department of Veterans Affairs, National Center on Homelessness Among Veterans. Dr. Montgomery’s work emphasizes homelessness prevention, interventions to end homelessness among high-need populations, and the demography and epidemiology of homelessness among veterans.
48
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
Melissa E. Dichter, PhD, MSW, is a health services researcher at the U.S. Department of Veterans Affairs, Center for Health Equity Research and Promotion. Dr. Dichter’s research focuses on women’s experiences with intimate partner violence and related psychosocial health care needs.
Xiaoying Fu, MS, is a programmer at the U.S. Department of Veterans Affairs, Center for Health Equity Research and Promotion. Mr. Fu assembles and analyzes data for research studies that seek to uncover and solve health disparities in the veteran population.
Arwin M. Thomasson, PhD, is a biostatistician at the U.S. Department of Veterans Affairs, Center for Health Equity Research and Promotion. Dr. Thomasson’s applied research focuses on health disparities and end-of-life care. Her statistical research interests include joint modeling and mixed-effects methods.
Christopher B. Roberts, MPH, is a Data Administrator for the National Center on Homelessness Among Veterans and the Center for Health Equity Research and Promotion. Mr. Roberts mines, manages, and analyzes data for research studies in support of ending homelessness among veterans.
www.whijournal.com
Original article
Demographic Characteristics Associated with Homelessness and Risk Among Female and Male Veterans Accessing VHA Outpatient Care Ann Elizabeth Montgomery, PhD a,*, Melissa E. Dichter, PhD, MSW b, Arwin M. Thomasson, PhD b, Xiaoying Fu, MS b, Christopher B. Roberts, MPH b a b
U.S. Department of Veterans Affairs, National Center on Homelessness Among Veterans, Philadelphia, Pennsylvania U.S. Department of Veterans Affairs, Center for Health Equity Research and Promotion, Philadelphia, Pennsylvania
Article history: Received 27 January 2014; Received in revised form 2 July 2014; Accepted 8 October 2014
a b s t r a c t Background: This study explored demographic influences on veterans’ reports of homelessness or imminent risk of homelessness with a particular focus on gender. Methods: We analyzed data for a cohort of veterans who responded to the U.S. Department of Veterans Affairs (VA), Veterans Health Administration (VHA) universal screener for homelessness and risk during a 3-month period. Multinomial mixed effects modelsdstratified by genderdpredicted veterans’ reports of homelessness or risk based on age, race, marital status, and receipt of VA compensation. Findings: The proportion of positive screensdhomelessness or riskdwas 2.7% for females and 1.7% for males. Women more likely to report being at risk of homelessness were aged 35 to 54 years, Black, and unmarried; those more likely to experience homelessness were Black and unmarried. Among male veterans, the greatest predictors of both homelessness and risk were Black race and unmarried status. Among both genders, receiving VA disability compensation was associated with lesser odds of being homeless or at risk. Conclusions: The findings describe the current population of veterans using VHA health care services who may benefit from homelessness prevention or intervention services, identify racial differences in housing stability, and distinguish subpopulations who may be in particular need of intervention. Interventions to address these needs are described. Published by Elsevier Inc.
Homelessnessda longstanding “public problem” in the United States (Stern, 1984)dis associated with poor health, economic, and social outcomes (Burt, 2001; Hawkins & Abrams, 2007; Hwang, 2001; Lee & Farrell, 2003; Lee & Greif, 2008; Wolitski, Kidder, & Fenton, 2007; Zerger, 2002). At a single point in time in January 2014, more than 578,000 people in the United States were homeless, staying in shelters or other temporary housing, or on the streets or other places not meant for
This study was funded by the U.S. Department of Veterans Affairs, National Center on Homelessness Among Veterans and Center for Health Equity Research and Promotion (Grant no. LIP 72-061). The contents of this article do not necessarily represent the views of the U.S. Department of Veterans Affairs or the U.S. Government. * Correspondence to: Ann Elizabeth Montgomery, PhD, U.S. Department of Veterans Affairs, National Center on Homelessness Among Veterans, 4100 Chester Avenue, Suite 201, Philadelphia, PA 19146. Phone: 215-823-5800x5067; Fax: 215-222-2591. E-mail address: [email protected] (A.E. Montgomery). 1049-3867/$ - see front matter Published by Elsevier Inc. http://dx.doi.org/10.1016/j.whi.2014.10.003
human habitation; approximately 11% of homeless adults were veterans (U.S. Department of Housing and Urban Development [HUD], 2014). Over the course of a year, roughly 1 in every 156 veterans will experience homelessness (HUD, 2013). Although the rate of veteran homelessness has declined during recent years, the problem continues to impact a substantial proportion of the veteran population, including younger veterans and female veterans (Fargo et al., 2012): During federal fiscal year 2012, 8.5% of veterans who accessed emergency shelter or transitional housing were younger than 30 years and almost 7.8% were female (HUD, 2013). The U.S. Department of Veterans Affairs (VA) has intensified its efforts to address homelessness among veterans, as evidenced by its 5-year plan to end homelessness as well as substantial new investments in programs intended to prevent and end homelessness (U.S. Interagency Council on Homelessness, 2010). VA’s strategy places particular emphasis on homelessness prevention (U.S. Interagency Council on Homelessness, 2013), which is in line
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
with a broader shift in homeless assistance policies in the United States that increasingly favor prevention and rapid rehousing approaches (Burt, Pearson, & Montgomery, 2005, 2007; Culhane, Metraux, & Byrne, 2011); that is, either ensuring that a homeless episode does not occur or, if it does, ending it as rapidly as possible. In support of this prevention-oriented strategy, the VA established the Supportive Services for Veteran Families (SSVF) program in 2011 to promote housing stability among homeless and at-risk veterans and their families. Through this program, VA awards grants to nonprofit organizations that provide supportive services to eligible very low-income veteran families who are either recently homeless or at risk, including outreach, case management, assistance in obtaining VA benefits, and help in accessing and coordinating other public benefits. SSVF grantees can also provide temporary financial assistance to cover rent, utilities, security deposits, and moving costs (VA, 2013a). To improve the VA’s ability to identify veterans who are at risk of homelessness, or experiencing homelessness but are not accessing services through the Veterans Health Administration’s (VHA) Homeless Program, the VA National Center on Homelessness Among Veterans developed the Homelessness Screening Clinical Reminder (HSCR) to conduct an ongoing, universal screen for homelessness and risk among veterans accessing outpatient health care services (Montgomery, Fargo, Byrne, Kane, & Culhane, 2013; Montgomery, Fargo, Kane, & Culhane, 2014). The HSCR is embedded in veterans’ medical records; when they present for services, health care staff ask veterans the screening questions and record their responses in the medical record. The objective of this national, health system-based screening instrument is to enhance the rapid identification of veterans who have very recently become homeless or are at imminent risk of homelessness, and to ensure that they are referred for appropriate assistance. Implemented in October 2012, the HSCR is the first attempt by a health system or other national-level service providerdin the United States or elsewheredto systematically assess whether patients are homeless or at risk. An initial analysis of data collected through the HSCR between October 1, 2012, and January 10, 2013, identified the following rates of housing instability: 0.9% of veteran respondents reported current homelessness, 1.2% reported being at risk, and 97.9% screened negative for both. This analysis found that, compared with men, significantly more female veterans reported being homeless (1.1% for women vs. 0.9% for men) and at risk (1.9% for women vs. 1.2% for men). This initial analysis identified significant gender-based differences in positive screens, calling for future research to inform outreach, targeting, and homelessness prevention interventions (Montgomery et al., 2013). The objective of the present study was to explore the demographic associations of veterans’ reports of being homeless or at risk with a particular focus on how this varies by gender; therefore, we conducted analyses separately for men and women (Bird, 2013). Because VHA’s patient population is predominantly maledcomprising 93.2% of the cohort studied heredgender differences may not be particularly evident when conducting analyses that include all veterans rather than conducting gender-specific analyses. Additionally, the literature has identified differences between the characteristics associated with homelessness by gender and it may be helpful to identify the particular characteristics for each of these groups (Blackstock, Haskell, Brandt, & Desai, 2012; HUD & VA, 2010, 2011; Leda, Rosenheck, & Gallup, 1992; Tsai, Rosenheck, & McGuire, 2012).
43
Methods This study was approved by the local institutional review board. Data were extracted from the VA Corporate Data Warehouse, a repository of data from VHA clinical and administrative systems (VA Information Resource Center, 2012). The VA Corporate Data Warehouse contains records of all inpatient and outpatient stays occurring at any VA medical facility, as well as demographic information about patients receiving services. Sample The cohort for this study is comprised of all veterans who completed the HSCR during a VHA outpatient visit between November 1, 2012, and January 31, 2013, excluding those who declined screening or had missing, incomplete, or duplicate responses to the HSCR (n ¼ 23,177; 1.4%). In addition, patients with no outpatient visit 6 months before screening (n ¼ 37,690; 2.3%) were also excluded, because this study focuses on individuals who were actively engaged in VHA care. In total, 4.4% of females and 3.6% of males were excluded from the analyses. Measures The dependent variable is defined by veterans’ responses to the following 2 questions in the HSCR: 1. In the past 2 months, have you been living in stable housing that you own, rent, or stay in as part of a household? (“No” response indicates veteran is positive for homelessness and screening ends.) 2. Are you worried or concerned that in the next 2 months you may not have stable housing that you own, rent, or stay in as part of a household? (“Yes” response indicates veteran is at risk for homelessness.) Veterans are categorized as either homeless, at risk, or negative for both homelessness and risk. If a veteran is positive for homelessness, risk is not assessed; if a veteran is positive for neither homelessness nor risk, the veteran is considered negative. The HSCR is administered annually with all veterans who access outpatient health care services with the exception of those who received homeless assistance through VA at any time during the previous 6 months; those who report already receiving homeless services or assistance (unknown to VA at the time of screening); long-term residents of a nursing home or long-term care facility; and veterans who decline or are unable to answer. Veterans screening positive are rescreened semiannually. Veterans in palliative or long-term care are rescreened biannually. Once a veteran screens negative during 3 consecutive screens, the veteran is only screened biannually. The independent variables include: Age, a continuous measure categorized into 5 groups; Race, which we collapsed into 3 categories (White, Black, and other); Marital status, composed of 6 categories that we collapsed to create a binary variable (married/unmarried); and VHA Enrollment Priority Group, which indicates the extent to which a veteran is receiving compensation owing to a disability incurred during military service as well as whether a veteran is very low income. We collapsed the groups as follows: 50% or greater service-connected (receiving the
44
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
most VA financial compensation); less than 50% serviceconnected disability (less, but some, VA financial compensation); no service-connected disability but Medicaid eligible (no VA financial compensation and low-income); and no service-connected disability but not Medicaid eligible (no VA financial compensation and not low-income; U.S. Department of Veterans Affairs, 2012). Analysis We conducted all analyses using SAS version 9.2 (SAS Institute, Inc., 2011), including multinomial mixed effects models, with a facility-specific random intercept. Univariate models utilized all available data (i.e., patients with missing values for a given covariate were not excluded from other univariate models). We estimated four multivariable modelsdfor at-risk females, homeless females, at-risk males, and homeless malesdwhich included only patients with complete data; we chose this approach to describe the cohort as thoroughly as possible. The multivariable models contained the covariate of interest, as well as age and race. Results Sample Description The study sample included 1,582,125 veterans: 107,504 (6.8%) were women and 1,474,621 (93.2%) were men. Table 1 describes the demographic characteristics of the veterans in the sample by gender. The majority of women (62.9%) were younger than 55 years, whereas the majority of men (79.7%) were older than 55 years. The majority of women (58.4%) and men (69.9%) reported identifying as White, followed by Black. Approximately one third of women veterans and almost two thirds of men veterans reported being married. Nearly two thirds of female veterans received VA compensation for a service-connected disability, whereas one half of men received VA compensation for a service-connected disability. Overall, 0.8% of the veterans who accessed VHA outpatient health care during the 3-month study period reported current homelessness, 1.1% reported being at risk, and the remaining 98.2% were negative for both homelessness and risk. The rate of positive screens among women was 2.7% compared with 1.7% for men. (These rates are slightly different than those reported in the initial analysis of HSCR data owing to differences in the inclusion criteria for the cohort [Montgomery et al., 2013]). Homelessness and Risk among Female Veterans Table 2 presents the unadjusted and adjusted odds ratios for the association of women veterans screening positive for homelessness or being at risk and demographic characteristics. Compared with female veterans younger than 35, those in the 45- to 54-year age range had the greatest odds of screening positive for risk, whereas those aged 65 or older were more than 3 times less likely to screen positive for risk. Female veterans who identified as Black or other race (American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, Mixed) were more likely than Whites to report risk of homelessness. Marital status was also a significant predictor of female veterans reporting being at risk of homelessness: Those who were unmarried had almost twice the odds of reporting homelessness risk as their married counterparts. Finally, receiving VA compensation for a service-connected disabilitydcompared
Table 1 Characteristics of Veterans Health Administration Outpatients, by Gender: November 1, 2012–January 30, 2013 Female n
Male %
n
Total %
n
%
Total number 107,504 6.8 1,474,621 93.2 1,582,125 100.0 Age (y) <35 19,495 18.1 69,973 4.7 89,468 5.6 35–44 19,092 17.8 75,323 5.1 94,415 5.6 45–54 28,977 27.0 153,854 10.4 182,831 11.6 55–64 25,967 24.2 408,636 27.7 434,603 27.5 65 13,973 13.0 766,832 52.0 780,805 49.4 Missing 3 0.0 3 0.0 Race White 62,730 58.4 1,030,549 69.9 1,093,279 69.1 Black 27,008 25.1 196,451 13.3 223,459 14.1 American Native 743 0.7 6,841 0.5 7,584 0.5 Asian 830 0.8 8,439 0.6 9,269 0.6 Hawaiian 998 0.9 10,375 0.7 11,373 0.7 Mixed 1,464 1.4 14,306 1.0 15,770 1.0 Missing 13,731 12.8 207,660 14.1 221,391 14.0 Marital status Married 35,464 33.0 860,976 58.4 896,440 56.7 Divorced 36,714 34.2 324,674 22.0 361,388 22.8 Never married 23,319 21.7 142,573 9.7 165,892 10.5 Separated 4,832 4.5 43,693 3.0 48,525 3.1 Single 347 0.3 2,848 0.2 3,195 0.2 Widowed 6,325 5.9 95,099 6.4 101,424 6.4 Missing 503 0.5 4,758 0.3 5,261 0.3 VA Enrollment Priority Group 50% SC disability 39,629 36.9 427,605 29.0 467,234 29.5 <50% SC disability 24,851 23.1 285,943 19.4 310,794 19.6 Not SC/Medicaid 25,327 23.6 368,920 25.0 394,247 24.9 eligible Not SC/Not Medicaid 17,560 16.3 391,506 26.5 409,066 25.9 eligible Missing 137 0.1 647 0.0 784 0.0 Response to HSCR Homeless 1,033 1.0 10,876 0.7 11,909 0.8 At risk 1,809 1.7 14,845 1.0 16,654 1.1 Negative 104,662 97.4 1,448,900 98.3 1,553,562 98.2 Abbreviation: SC, service-connected.
with not having a service-connected disability and not being Medicaid eligibledwas associated with lower odds of homelessness risk among women veterans. Although younger age was not associated with a positive screen for risk, women veterans older than 55 years had significantly lower odds of reporting homelessness compared with those younger than 35. Women veterans who identified as Black were more likely than Whites to report homelessness and unmarried female veterans were more than twice as likely as married female veterans to report homelessness. Again, receipt of benefits related to a service-connected disability was associated with lower odds of homelessness among women. The profiles of women veterans who reported being at risk were somewhat distinct from those reporting homelessness. Specifically, women between 35 and 54 years were more likely to report being at risk than those younger than 35 years; however, this did not hold true for homelessness. Generally, older age was associated with a lower odds of both homelessness and risk among women. Women veterans identifying as Black were more likely than those identifying as White to screen positive for either homelessness or risk. Unmarried women had a greater likelihood of reporting homelessness than being at risk. Marital status and older age were the predictors for both homelessness and risk with the greatest effect size. Finally, both homelessness
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
45
Table 2 Unadjusted and Adjusted Odds of Screening Positive for Homelessness or at Risk among Female VHA Outpatients: November 1, 2012–January 30, 2013 Effect
Age (y) 35–44 45–54 55–64 65 Race Other Black Marital status – not married VA Enrollment Priority Group 50% SC disability <50% SC disability Not SC/not Medicaid eligible
Reference
At Risk (n ¼ 1,809)
Homeless (n ¼ 1,033)
OR (95% CI)
AOR (95% CI)
OR (95% CI)
AOR (95% CI)
1.20 1.30 1.05 0.31
1.21 1.24 0.96 0.27
0.92 1.07 0.81 0.27
0.98 1.03 0.73 0.25
<35 (1.03, (1.13, (0.91, (0.24,
1.40)* 1.50)* 1.21) 0.40)*
(1.03, (1.07, (0.82, (0.20,
1.42)* 1.43)* 1.12) 0.35)*
(0.76, (0.90, (0.68, (0.19,
1.12) 1.26) 0.98)* 0.37)*
(0.80, (0.86, (0.60, (0.18,
1.21) 1.24) 0.89)* 0.36)*
White
Married Not SC/Medicaid eligible
1.34 (1.06, 1.69)* 1.58 (1.41, 1.77)* 1.89 (1.69, 2.12)*
1.30 (1.03, 1.63)* 1.42 (1.27, 1.59)* 1.71 (1.51, 1.94)*
1.23 (0.90, 1.69) 1.63 (1.40, 1.88)* 2.80 (2.36, 3.33)*
1.19 (0.87, 1.63) 1.45 (1.24, 1.68)* 2.38 (1.98, 2.87)*
0.53 (0.47, 0.59)* 0.71 (0.62, 0.80)* 0.57 (0.49, 0.66)*
0.47 (0.42, 0.54)* 0.69 (0.60, 0.79)* 0.69 (0.59, 0.81)*
0.46 (0.39, 0.53)* 0.58 (0.50, 0.69)* 0.42 (0.34, 0.52)*
0.44 (0.37, 0.52)* 0.55 (0.46, 0.66)* 0.51 (0.41, 0.64)*
Abbreviations: AOR, adjusted odds ratio; OR, odds ratio; SC, service-connected. Notes. Reference category for outcomes is negative. AOR adjust for race and age. * p < .0001.
and risk were less likely among women veterans receiving benefits related to service-connected disabilities. Homelessness and Risk among Male Veterans Table 3 presents the unadjusted and adjusted odds ratios for the association of male veterans screening positive for homelessness or risk and a set of demographic characteristics. Among male veterans, those aged 55 years and older were significantly less likely to screen positive for risk. Male veterans identifying as Black or some other race were more likely than their White counterparts to report homelessness risk and unmarried male veterans were around twice as likely to report homelessness risk compared with those who were married. Receiving financial compensation for a service-connected disability was associated with a lower homelessness risk for men; those without a serviceconnected disability who were not low income were least likely to screen positive for risk. Male veterans older than age 55 years were less likely than their younger counterparts to report current homelessness; those between 35 and 54 years were slightly more likely than those younger than 35 to report homelessness. Compared with veterans who were White, those identifying as Black had approximately a 2 times higher odds of reporting homelessness. Unmarried male veterans were more than 3 times more likely than married male veterans to report homelessness. Finally, service connectedness was associated with lower odds of reporting homelessness. The only demographic variable that seemed to have a different pattern regarding the likelihood of being homeless or at risk among men was marital status. Although there was a significant, positive relationship between unmarried status and both homelessness and at risk, there was a stronger relationship with homelessness than at risk. Discussion This study identified the demographic correlates of current homelessness as well as risk of imminent homelessness among a large cohort of female and male veterans receiving care through the VHA. The analyses conducted here identified that, for women
veterans, being in the middle age range (35–54 years), Black race, and being unmarried were associated with higher odds of screening at risk, whereas only Black race and unmarried status were associated with higher odds of screening positive for homelessness. The effect size of unmarried status was approximately 30% greater for predicting homelessness than risk. Among male veterans, although older age seemed to be associated with lower odds of screening positive for risk or homelessness, the greatest predictors of both were Black race and unmarried status. Among both men and womendand for both outcomesdhaving access to compensation related to a serviceconnected disability (compared with being non-service connected and Medicaid eligible) was associated with lower odds of homelessness and risk. These findings shed light on the current population of veterans using VHA health care services who may benefit from homelessness prevention or intervention services and inform our understanding of housing stability in this population. Uniquely, these findings can be useful for early intervention efforts to prevent homelessness among those individuals who are at imminent risk but not (yet) experiencing homelessness. Women aged 35 to 54 years have an increased odds of being at risk for homelessness, compared with their younger peers, but not of current homelessness; the youngest group of women veterans (those younger than 35 years) may be most likely to return to their parents’ homes after military service, thus affording at least temporary housing stability. The vulnerable, middle-age population of women may represent those who are parenting children and adolescents or caring for aging parents, which may increase their household expenses while limiting their employment opportunities during their peak earning years. Women older than 54 years may be more stably housed and partnered and/or employed or retired and less likely than their younger peers to experience intimate partner violence, a common pathway into homelessness among women (Baker, Cook, & Norris, 2003; Hamilton, Poza, & Washington, 2011; Pavao, Alvarez, Baumrind, Induni, & Kimerling, 2007; Washington et al., 2010). Black race and unmarried status were strong predictors of homelessness and risk, particularly among men. Consistent with previous findings, we found that Black race was related to increased risk of housing instability (HUD, 2013). Similarly,
46
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
Table 3 Unadjusted and Adjusted Odds of Screening Positive for Homelessness or at Risk among Male VHA Outpatients: November 1, 2012 - January 30, 2013 Effect
Age (y) 35–44 45–54 55–64 65 Race Other Black Marital status – not married VA Enrollment Priority Group 50% SC disability <50% SC disability Not SC/not Medicaid eligible
Reference
At Risk (n ¼ 14,845)
Homeless (n ¼ 10,876)
OR (95% CI)
AOR (95% CI)
OR (95% CI)
AOR (95% CI)
1.10 1.23 0.73 0.18
1.12 1.10 0.69 0.20
0.97 1.25 0.68 0.16
1.05 1.09 0.64 0.18
<35 (1.02, (1.15, (0.68, (0.17,
1.18)* 1.31)* 0.77)* 0.19)*
(1.04, (1.03, (0.64, (0.18,
1.22)* 1.18)* 0.73)* 0.21)*
(0.89, (1.16, (0.64, (0.15,
1.05) 1.34)* 0.73)* 0.17)*
(0.96, (1.01, (0.60, (0.17,
1.15) 1.18)* 0.69)* 0.20)*
White
Married Not SC/Medicaid eligible
1.32 (1.21, 1.45)* 2.31 (2.21, 2.41)* 2.53 (2.44, 2.62)*
1.18 (1.07, 1.29)* 1.66 (1.59, 1.73)* 1.74 (1.67, 1.80)*
1.33 (1.20, 1.48)* 2.59 (2.47, 2.72)* 5.04 (4.81, 5.28)*
1.23 (1.11, 1.37)* 1.78 (1.69, 1.87)* 3.31 (3.14, 3.48)*
0.41 (0.40, 0.43)* 0.66 (0.63, 0.69)* 0.37 (0.35, 0.39)*
0.42 (0.40, 0.44)* 0.67 (0.64, 0.70)* 0.53 (0.50, 0.56)*
0.31 (0.30, 0.33)* 0.48 (0.46, 0.51)* 0.31 (0.29, 0.32)*
0.37 (0.35, 0.40)* 0.54 (0.51, 0.58)* 0.51 (0.48, 0.54)*
Abbreviations: AOR, adjusted odds ratio; OR, odds ratio; SC, service connected. Notes. Reference category for outcomes is negative. AOR adjust for race and age. * p < .0001.
earlier studies have found higher rates of homelessness among those who are unmarried (Jencks, 1994); however, in the sample studied herein, there are likely different explanations for unmarried status posing risk to women and men, including the amount and type of social support each receives from their spouse. The women in the sample were younger and more likely to be unmarried, whereas the men in the sample were older and more likely to be married. Unmarried status for the women may signify single parenthood, and for both genders it may signify a lack of social support. Several limitations of this study should be considered. First, the study included only veterans who accessed outpatient health care services through the VHA at least twice during the 6-month period before the visit at which they responded to the HSCR; therefore, the rates reported herein cannot be considered prevalence rates of homelessness or risk among all veterans. Second, the constructs of homelessness and risk are based on veterans’ self-report and their own perceptions of whether they are experiencing unstable housing, which may vary by individual. Third, because the HSCR is a national-level screen administered by many types of staff throughout VHA, there may be distinctions in how veterans are screened; variations in screening practices may influence response. Although the data were based on veterans’ self-report, and analyses control for location of screening, the variations in screening procedures may be systematic, leading to inaccurate rates of homelessness and risk as well as relationships between these rates and veterans’ demographic characteristics. In addition, the administrative data may include possible errors in documentation of marital status and race. To date, the research literature has provided little explanation for findings of differences between female and male veteran homelessness, and veteran and non-veteran homelessness. Women, in particular, may be vulnerable to housing instability owing to social factors that could not be measured within the data used for this study. For example, several researchers have identified violence and abuse as precursors to homelessness among women in general (Baker et al., 2003; Browne & Bassuk, 1997; Goodman, 1991; Pavao et al., 2007; Rollins, Saris, & Johnston-Robledo, 2001) and female veterans in particular (Hamilton et al., 2011; Washington et al., 2010). Other factors related to homelessness among veteran and non-veteran women include adolescent childbearing, low-wage employment or
unemployment, lack of affordable housing, and disability, poor health, and diagnosis of posttraumatic stress disorder or an anxiety disorder (Goodman, 1991; Hamilton et al., 2011; Rollins et al., 2001; Washington et al., 2010). Future research should include as predictive variables sources of risk beyond demographic characteristics. Implications for Policy and/or Practice The veterans in this study who reported current homelessness or risk represent the population receiving care in the VHA who have not been reached with homeless services (those already receiving such services are excluded from the screening) but may benefit from such programs. Not only do the findings presented herein begin to identify specific subpopulations of veterans who may be in particular need of intervention, but they also hint at particular interventions specific to each subpopulation. Younger, unmarried females; older, unmarried males; and veterans who do not seem to have access to income through VA compensation are the most likely to report current homelessness or imminent risk. We also continue to see racial disparities with increased risk among non-White populations. Efforts need to be made to target these subpopulations, not only for screening and assessment, but for interventions that prevent homelessness or quickly end a current episode of homelessness. Younger, unmarried female veterans may be best assisted through access to childcare, educational resources, temporary financial assistance for rent or utilities, and perhaps supportive housing, depending on other needs such as mental or behavioral health conditions. Although the SSVF program provides these resources throughout hundreds of communities in the United Statesdand has provided services to families (30% of all households served in federal fiscal year 2012) and a disproportionate number of female veterans (15% of all veterans served in federal fiscal year 2012)dperhaps greater outreach and education about these services is necessary within the health care system (VA, 2013b). In addition, the provision of services to the children of younger, unmarried women may prevent involvement in other social services systems, such as foster care, in the future (Culhane, Metraux, Byrne, Stino, & Bainbridge, 2013). Efforts to provide prevention services to unmarried, older male veterans who are homeless or at risk should be increased,
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
both in the community and within the health care system. Although not a focus of the present study, we note that, as these veterans continue to age, they may experience chronic medical conditions that require additional support, both medically and socially (Culhane et al., 2013). Programs that enable veterans to remain in housing, affordably, as their needs increase may increase their housing stability. Although we do not have information about employment status or receipt of other income or benefits among veterans screening positive for homelessness and at risk, the findings from the present study indicate that receiving some financial compensation is protective against housing instability. Efforts to increase income among veterans who are homeless or at riskdthrough jobs programs or application for benefitsdcould increase the likelihood that veterans attain or maintain permanent housing. Current work is considering the role that physical, mental, and behavioral health play in female and male veterans’ report of homelessness and risk as well as the likelihood of veterans to access VA services after a positive screen. Conducting these analyses from a gender-based perspective will increase our understanding about the specific needs of female and male veterans and how interventions may be tailored. To determine which interventions are most effective at mitigating homelessness and risk, for which populations, future studies could test the effectiveness of various interventions using randomized designs and assess for demographic variation. Further qualitative research to uncover the ways in which various factors may influence veterans’ experiences of housing instability will help to inform our understanding of pathways to homelessness and potential points of intervention for this population.
References Baker, C. K., Cook, S. L., & Norris, F. H. (2003). Domestic violence and housing problems: A contextual analysis of women’s helpseeking, received informal support, and formal system response. Violence Against Women, 9(7), 754–783. Bird, C. E. (2013). Spotlight on women’s health: Gender-based analysis. Retrieved 10/9/2013 from. http://www.hsrd.research.va.gov/for_researchers/cyber_ seminars/archives/video_archive.cfm?SessionID¼755. Blackstock, O. J., Haskell, S. G., Brandt, C. A., & Desai, R. A. (2012). Gender and the use of Veterans Health Administration homeless services programs among Iraq/Afghanistan Veterans. Medical Care, 50(4), 347–352. Browne, A.,, & Bassuk, S. S. (1997). Intimate violence in the lives of homeless and poor housed women: Prevalence and patterns in an ethnically diverse sample. American Journal of Orthopsychiatry, 67, 261–278. Burt, M. R. (2001). Helping America’s homeless: Emergency housing or affordable housing? Washington, DC: The Urban Institute. Burt, M. R., Pearson, C. L., & Montgomery, A. E. (2005). Strategies for preventing homelessness. Washington, DC: U.S. Department of Housing and Urban Development. Burt, M. R., Pearson, C. L., & Montgomery, A. E. (2007). Community-wide strategies for preventing homelessness: Recent evidence. Journal of Primary Prevention, 28, 213–228. Culhane, D. P., Metraux, S., & Byrne, T. H. (2011). A prevention-centered approach to homelessness assistance: A paradigm shift? Housing Policy Debate, 21(2), 295–315. Culhane, D. P., Metraux, S., Byrne, T. H., Stino, M., & Bainbridge, J. (2013). The age structure of contemporary homelessness: Evidence and implications for public policy. Analyses of Social Issues and Public Policy, 13(1), 228–244. Fargo, J. D., Metraux, S., Byrne, T. H., Munley, E., Montgomery, A. E., Jones, H., et al. (2012). Prevalence and risk of homelessness among U.S. Veterans: A multisite investigation. Preventing Chronic Disease, 9, 110–112. Goodman, L. (1991). The prevalence of abuse in the lives of homeless and housed poor mothers: A comparison study. American Journal of Orthopsychiatry, 61, 489–500. Hamilton, A. B., Poza, I., & Washington, D. L. (2011). “Homelessness and trauma go hand-in-hand”: Pathways to homelessness among women Veterans. Women’s Health Issues, 21(4S), S203–S209.
47
Hawkins, R. L., & Abrams, C. (2007). Disappearing acts: The social networks of formerly homeless individuals with co-occurring disorders. Social Science & Medicine, 65(10), 2031–2042. Hwang, S. W. (2001). Homelessness and health. Canadian Medical Association Journal, 164(2), 229–233. Jencks, C. (1994). The homeless. Cambridge, MA: Harvard University Press. Leda, C., Rosenheck, R. A., & Gallup, P. (1992). Mental illness among homeless female Veterans. Hospital Community Psychiatry, 43(10), 1026–1028. Lee, B. A.,, & Farrell, C. R. (2003). Buddy, can you spare a dime? Urban Affairs Review, 38(3), 299–324. Lee, B. A.,, & Greif, M. J. (2008). Homelessness and hunger. Journal of Health and Social Behavior, 49(1), 3–19. Montgomery, A. E., Fargo, J. D., Byrne, T. H., Kane, V., & Culhane, D. P. (2013). Universal screening for homelessness and risk for homelessness in the Veterans Health Administration. American Journal of Public Health, 103(S2), S210–S211. Montgomery, A. E., Fargo, J. D., Kane, V., & Culhane, D. P. (2014). Development and validation of an instrument to assess imminent risk of homelessness among Veterans. Public Health Reports, 129, 439–447. Pavao, J., Alvarez, J., Baumrind, N., Induni, M., & Kimerling, R. (2007). Intimate partner violence and housing instability. American Journal of Preventive Medicine, 32(2), 143–146. Rollins, J. H., Saris, R. N., & Johnston-Robledo, I. (2001). Low-income women speak about housing: A high-stakes game of musical chairs. Journal of Social Issues, 57, 277–298. SAS Institute Inc. (2011). SAS/STATÒ 9.3 user’s guide. Cary, NC: Author. Stern, M. J. (1984). The emergence of the homeless as a public problem. Social Service Review, 58(2), 291–301. Tsai, J., Rosenheck, R. A., & McGuire, J. F. (2012). Comparison of outcomes of homeless female and male veterans in transitional housing. Community Mental Health Journal, 48(6), 705–710. U.S. Department of Housing and Urban Development (HUD). (2014). The 2014 Annual Homeless Assessment Report (AHAR) to Congress. Washington, DC: Author. U.S. Department of Housing and Urban Development (HUD). (2013). The 2012 Annual Homeless Assessment Report to Congress. Washington, DC: Author. U.S. Department of Housing and Urban Development (HUD), & U.S. Department of Veterans Affairs (VA). (2011). Veteran homelessness: A supplemental report to the 2010 annual homeless assessment report to Congress. Washington, DC: Author. U.S. Department of Housing and Urban Development (HUD), & U.S. Department of Veterans Affairs (VA). (2010). Veteran homelessness: A supplemental report to the 2009 annual homeless assessment report to Congress. Washington, DC: Author. U.S. Department of Veterans Affairs (VA). (2012). Priority Groups Table. Retrieved 1/16/2014 from: http://www.va.gov/healthbenefits/resources/priority_ groups.asp. U.S. Department of Veterans Affairs (VA). (2013a). Department of Veterans Affairs Supportive Services for Veteran Families (SSVF) Program Guide. Retrieved 1/16/2014 from: http://www.va.gov/HOMELESS/docs/SSVF/SSVFUniversity/ ProgramRequirementsPage/Program_Guide_Revised_12_19_2013_FINAL.pdf. U.S. Department of Veterans Affairs (VA). (2013b). Effectiveness of permanent housing program FY 2012 report. Retrieved 1/23/2014 from: http://www.va.gov/ HOMELESS/docs/SSVF/Effectiveness_of_SSVF_Program_Report_FY2012.pdf. U.S. Interagency Council on Homelessness. (2010). Opening doors: Federal strategic plan to prevent and end homelessness. Washington, DC: Author. U.S. Interagency Council on Homelessness. (2013). Ending homelessness among veterans: A report by the United States Interagency Council on Homelessness. Retrieved 1/16/2014 from: http://usich.gov/resources/uploads/asset_library/ USICH_Ending_Homelessness_Among_Veterans_Rpt_February_2013_FINAL. pdf. VA Information Resource Center. (2012). VA Corporate Data Warehouse (CDW): A VIReC resource guide. Retrieved 10/15/2013 from: http://vaww.virec.research. va.gov/CDW/RG-CDW-CY12-RA.pdf. Washington, D. L., Yano, E. M., McGuire, J., Hines, V., Lee, M., & Gelberg, L. (2010). Risk factors for homelessness among women veterans. Journal of Health Care for the Poor and Underserved, 21, 82–91. Wolitski, R. J., Kidder, D. P., & Fenton, K. A. (2007). HIV, homelessness, and public health: Critical issues and a call for increased action. AIDS and Behavior, 11, 167–171. Zerger, S. (2002). A preliminary review of literature: Chronic medical illness and homelessness. Nashville, TN: National Health Care for the Homeless Council.
Author Descriptions Ann Elizabeth Montgomery, PhD, conducts research with the U.S. Department of Veterans Affairs, National Center on Homelessness Among Veterans. Dr. Montgomery’s work emphasizes homelessness prevention, interventions to end homelessness among high-need populations, and the demography and epidemiology of homelessness among veterans.
48
A.E. Montgomery et al. / Women's Health Issues 25-1 (2015) 42–48
Melissa E. Dichter, PhD, MSW, is a health services researcher at the U.S. Department of Veterans Affairs, Center for Health Equity Research and Promotion. Dr. Dichter’s research focuses on women’s experiences with intimate partner violence and related psychosocial health care needs.
Xiaoying Fu, MS, is a programmer at the U.S. Department of Veterans Affairs, Center for Health Equity Research and Promotion. Mr. Fu assembles and analyzes data for research studies that seek to uncover and solve health disparities in the veteran population.
Arwin M. Thomasson, PhD, is a biostatistician at the U.S. Department of Veterans Affairs, Center for Health Equity Research and Promotion. Dr. Thomasson’s applied research focuses on health disparities and end-of-life care. Her statistical research interests include joint modeling and mixed-effects methods.
Christopher B. Roberts, MPH, is a Data Administrator for the National Center on Homelessness Among Veterans and the Center for Health Equity Research and Promotion. Mr. Roberts mines, manages, and analyzes data for research studies in support of ending homelessness among veterans.