Mental Health Correlates of Past Homelessness in the National Comorbidity Study Replication
This study uses data from a nationally representative epidemiologic survey, the National Comorbidity Survey Replication, to investigate the association of mental health and substance use disorders, along with other factors, with past homelessness. Approximately 5% of the 5,251 adults reported having been homelessness. Multivariate analysis showed the strongest independent risk factors for past homelessness were past receipt of welfare payments (odds ratio [OR]=5.7), incarceration for 27 or more days (OR=3.9), exposure to personal violence (OR=2.7), lifetime substance use disorder (OR=2.4), and Black race (OR=2.1). Several non-substance use psychiatric disorders were also significantly, if less strongly (OR 1.4 to 1.6), associated with past homelessness. Past homelessness is associated with a broad array of sociodemographic, economic, and mental health problems. While the association of both substance use and psychiatric disorders with past homelessness was quite strong, non-substance use psychiatric disorders was not as strong an independent risk factor as substance abuse disorders.
Homelessness, mental health, substance abuse, incarceration, trauma
Homelessness remains a major problem in the United States. The most recent estimate at the national level of the number of homeless people suggested that on a single night in January 2007 there were just over 670,000 sheltered homeless people.1 The complementary roles of both structural societal factors and individual characteristics in generating high levels of homelessness have long been acknowledged.2-4 While social and structural level factors (e.g., the declining availability of low-income housing or the reduced value of public support payments) may account for the number of homeless people at any given point in time, personal risk factors (i.e., sociodemographic, economic, and mental health characteristics) are likely to explain why some individuals are at greater risk for homelessness than others.2,3,5-9
The most prominent sociodemographic characteristics identified with high risk of homelessness have been male gender,2,4,10-11 Black race,2,4,10-12 being unmarried,2,4,8 and [End Page 1234] being middle aged.2,4,8,10,13 As would be expected, homeless people also suffer from economic disadvantages, such as extreme poverty, low job skills and educational levels, and high rates of unemployment.2,8,10,14
Much has also been reported on the association of mental illness and substance use disorders with homelessness.2,4,9,11,15-16 However, due to methodological differences across studies there has been large variation in the reported estimates of the nature, extent, and severity of mental health and substance use disorders among homeless individuals.3-4,10,14-15 Past estimates have suggested that about 20% to 50% of homeless adults suffer from a serious mental illness while approximately 50% to 80% have a history of substance use disorders.3,4,10,11
Two other individual characteristics that are often reported to be associated with homelessness are criminal justice system involvement, particularly previous incarceration, and exposure to trauma.2,11,13,17-20
While the influence of various characteristics on individual risk for homelessness has been well studied in local populations,10,15,16,21-24 the quality of the mental health diagnostic assessment in past national surveys has been limited2 and as a result it has been difficult to weigh the relative influence of sociodemographic, economic, and mental health characteristics as risk factors.
The National Combordity Study Replication (NCSR) provides a unique opportunity to examine these issues because it includes both comprehensive sociodemographic measures, including reports of past homelessness, and sophisticated measures of mental health and addictive disorders in a representative national sample. This study used NCSR data to examine correlates of past homelessness and especially the independent effects and relative importance of sociodemographic, economic, and psychiatric risk factors.
Source of data and sample.
As described in detail elsewhere25 the NCSR is a nationally representative household survey of respondents 18 years and older in the coterminous United States conducted from February 5, 2001 to April 7, 2003. The survey was limited to English speakers and excluded institutionalized individuals and those living on military bases.
Respondents to the NCSR were drawn by probability sampling within a multistage clustered area probability sample of households using the sampling frames and sample selection procedures that are common to the University of Michigan Survey Research Center's National Sample design.26-28 Face-to-face interviews were conducted by professional interviewers from the Institute for Social Research at the University of Michigan using laptop computer-assisted personal interview methods. The overall response rate was 70.9%. The survey was carried out in two parts. Part I included a core diagnostic assessment and was administered to all respondents. Part II assessed risk factors, correlates, service use, and additional disorders and was only administered to 5,692 of the 9,282 individuals who answered Part 1. The Part II data were weighted to adjust for different probabilities of selection in households, for differential nonresponse, for differential selection of Part 1 respondents into the Part II sample, and for residual discrepancies with U.S. Census data.25,29 [End Page 1235]
Interviewers obtained verbal informed consent prior to each interview. Study procedures were approved by the human subjects committees of Harvard Medical School and the University of Michigan at Ann Arbor.
Responses to several questions were used to construct a dichotomous indicator representing whether an individual was homeless longer than one week since s/he turned 18. The questions associated with these variables did not operationally define homelessness for the respondent.
A series of dichotomous measures represent gender, marital status, education (at least a high school degree), whether a language other than English was spoken while growing up, as well as whether the individual was born outside the United States. Four dichotomous measures were used to classify individuals as belonging to one of four exclusive racial/ethnic groups (White, Black, Other, and Hispanic). Additionally, four dichotomous represented U.S. regions (i.e., Northeast, Midwest, South, or West). A continuous measure of age was also created (each unit represents a decade).
Economic status was assessed with a continuous measure of current annual household income (in $10,000 increments) and two dichotomous measures representing past or current receipt of welfare payments, and current fulltime employment (i.e., 20 hours or more per week).
Criminal justice system involvement.
Two dichotomous measures represented involvement with the criminal justice system since the age of 18: 1) whether incarcerated from 1 to 27 days in a jail, prison, or a correctional facility; and 2) whether incarcerated for greater than 27 days. Although a cut-off of one month is commonly used to represent the distinction between short and longer periods of incarceration, 27 days was the only cutoff available in the NCSR data set.
Four measures of exposure to trauma were used. The first was a dichotomous indicator of whether the individual had ever participated in combat. We also used 48 items representing 26 traumas to create three measures that indicated whether an individual had experienced one of the following kinds of traumas: 1) passive exposure to a traumatic environment or event; 2) personal violence (for instance, being beaten up, mugged, raped); and/or 3) exposure to death, trauma, or injury of others.
Diagnostic Assessment of Lifetime Mental and Substance Use Disorders was conducted using Version 3.0 of the World Health Organization Composite International Diagnostic Interview (CIDI)30 based on Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria. The CIDI is a fully structured lay-administered diagnostic interview. We created four dichotomous measures that reflected four diagnostic groups suggested by Kessler and associates31 and that consisted of 16 lifetime mental health diagnoses (categorized as 1) anxiety disorders; 2) mood disorders; 3) disorders that feature difficulty with impulse control; and 4) substance use disorders [see Table 1 for the items that make up each measure]). Data collected on schizophrenia with the NCSR were not released by the National Institute of Mental Health because they were not judged to be valid.
DSM-IV organic exclusion rules were used in the making of all diagnoses. Additionally, hierarchical rules were used to make all diagnoses except for substance use diagnoses. Blind clinical re-interviews that used the structured clinical interview for [End Page 1236] [Begin Page 1237]
[End Page 1239]
DSM-IV (SCID or Structured Clinical Interview) found that there was good concordance between anxiety, mood, and substance disorders diagnoses based on the SCID and diagnoses based on the CIDI. The lifetime diagnoses of impulse control were not validated.32-33
One additional dichotomous measure was created that indicated whether an individual was seriously or severely mentally ill in the previous year based on a complex algorithm that used several indicators and measures. See Kessler and associates34 for further details.
There were several steps to our analyses. First, we performed a series of bivariate chi-squared and F tests to examine whether a significant relationship existed between prior homelessness and each of the other measures. Second, we investigated the strength of the relationship between past homelessness and individual characteristics with a series of bivariate logistic regressions in which past homelessness was the dependent variable and the odds ratio of each independent measure was the indicator of effect size. Finally, multivariate logistic regression analysis was used to identify the independent strength of the relationship between past homelessness and each risk factor.
The bivariate logistic regression analyses were used to examine the strength of the relationship between each risk factor and homelessness regardless of other factors. The multivariate logistic regression analyses, in contrast, examine the importance of each risk factor independent of other factors (i.e., the unique relationship of each variable to homelessness over and above the association with other variables).
All analyses were performed using Sudaan (Version 9.0.3; Research Triangle Institute, Research Triangle Park, North Carolina), which uses items available in NCSR to adjust for the effects of weighting, clustering, and non-responses on the precision of estimated variance. All significant tests in the logistic regression analyses were two-tailed with a p value of .05 and made using Wald F tests. These tests were based on coefficient variance-covariance matrices that were adjusted for design effects using the Taylor series method.
Approximately 5.5% of the 5,251 individuals for whom residential data were available had been homeless.
Substantial and significant differences were also found between individuals who had experienced homelessness and those who had not (Table 1). Individuals who were married, had graduated from high school, or had been born outside the U.S. were about half as likely to have experienced homelessness as others. The odds of having experienced homelessness also decreased by 2% for every 10 year increase in age (for example, a 40-year old would be 4% less likely to have experienced homelessness than a 20-year old).
In contrast, men were about 50% more likely than woman to have experienced homelessness and westerners had twice the risk of others. Being Black (Odds Ratio [OR] = 2.24) also significantly increased the odds of homelessness, while being White reduced the odds by almost half (OR = .55). [End Page 1240]
Two of the three indicators of economic status were significantly and strongly associated with the likelihood of past homelessness. Individuals who had ever received welfare payments were almost eight times as likely as others to have experienced homelessness; and the odds of experiencing homelessness decreased by 12% for every $10,000 increase in income (for instance, an individual with an income of $30,000 would be 24% less likely to have experienced homelessness than someone with an income of only $10,000).
As expected, individuals who had been incarcerated one to 26 days were 3.0 times more likely to have been homeless, and those incarcerated for 27 or more days were 11.5 times more likely.
Exposure to combat was not significantly associated with past homelessness, but individuals who had been exposed to a traumatic environment or event were 2.6 times as likely to have experienced homelessness, those who experienced personal violence, 6.5 times as likely, and those who had witnessed or caused trauma to others, 3.0 times as likely.
Almost all substance use and mental illness diagnoses were associated with past homelessness. The relationship between past homelessness and the two substance use disorders were particularly strong, with individuals who suffered from drug use being 7.4 times as likely as others to have experienced homelessness and those with an alcohol use disorder being 6.1 times more likely. Past homelessness was less strongly associated with the three psychiatric diagnostic classes: anxiety disorder (OR=2.87), mood disorder (OR = 2.78), and impulse control disorder (OR = 4.44).
It is important to emphasize that unlike the results from the multivariate analyses reported below, the above findings are based on bivariate logistic regression analyses and thus were not adjusted for other factors, such as age, that may have affected the relationship between individual characteristics and past homelessness.
Multivariate logistic regressions.
In multivariate analysis, few of the sociodemographic and economic characteristics were independently associated with past homelessness. Of the 13 sociodemographic measures only being Black (OR=2.08) and being male (OR=1.77) independently increased the odds of past homelessness, while being from the Northeast (OR=.55) or the Midwest (OR=.48), as opposed to the West, reduced the odds of past homelessness (see Table 2). Only one economic characteristic, receipt of welfare payments, was independently associated with past homelessness (OR=5.72).
Although there was no independent association between past homelessness and being incarcerated from one to 26 days, individuals incarcerated for 27 days or more were 3.9 times more likely to have experienced homelessness. Of the four trauma measures, only the experience of personal violence independently increased the odds of past homelessness (OR=2.70).
Past homelessness was significantly and independently associated with a lifetime substance use diagnosis (OR=2.70), as well as with two of the three classes of lifetime mental illness—mood disorder (1.58) and impulse control disorder (1.63).
To summarize, although being of the Other racial/ethnic group, age, being married, being a high school graduate, having spoke a language other than English growing up, and income were found to be significant in bivariate analyses, multivariate logistic [End Page 1241]
[End Page 1242]
regression analyses indicated that they were not independently associated with past homelessness. Thus, few of the sociodemographic and economic characteristics had an independent relationship to homelessness over and above other characteristics. Similarly, while three measures of trauma were significantly associated with past homelessness in the bivariate analyses only the experience of personal violence was independently associated with past homelessness. Of the mental health and substance abuse disorders in both models, only anxiety disorder was significant in bivariate analyses but not in the multivariate analyses.
Seriously or severely mentally ill.
In further multivariate analyses we substituted the dichotomous indicator of whether an individual was seriously or severely mentally ill for the three dichotomous diagnostic indicators representing anxiety, mood, or impulse control disorder. Although the measure was strongly associated with homelessness in the bivariate analyses, in the multivariate analyses the measure was barely significant and had roughly the same effect on the odds of being homeless as having a mood or impulse control disorder (OR: 1.47; Confidence Interval [CI]: 1.01-2.13).
Published estimates of the rate of lifetime homelessness (defined as living on the street or in a shelter) based on nationwide telephone surveys range between 6.2-8.1%, modestly higher than 5.5% estimated with NCSR data.35-37 However, the NCSR estimate is smaller than lifetime homelessness estimates that include being doubled up or moving in with someone else which range between 11.7% and 14.0%.35-37 The NCSR estimate may be lower because it only counted individuals as homeless who reported being homeless for at least a week, while earlier surveys did not specify a minimum period of homelessness.
Consistent with previous studies, the multivariate regression analysis indicated that past homelessness is associated with being Black, being male, having received welfare payments, criminal justice system involvement, and exposure to personal violence as well as mental health problems. While the association of both substance use and psychiatric disorders with past homelessness was quite strong, non-substance use psychiatric disorders were not as prominent independent risk factors as were substance abuse disorders.
The most likely explanation for this difference from prior studies is that this study had several methodological advantages. In particular, rather than surveying currently homeless individuals, we used a national representative survey to assess the past homelessness and other characteristics of currently domiciled individuals. Surveys of the currently homeless are notoriously difficult to conduct in that many homeless individuals may not be found and if found may refuse to be interviewed. Additionally, the chronically homeless and service users tend to be over-represented in surveys of currently homeless individuals, particularly of individuals using formal services for the homeless or living in traditional skid-row areas. A face-to-face survey of domiciled individuals is more likely to include individuals who experienced relatively short or intermittent episodes of homelessness and less likely to over-represent service users who have health problems. [End Page 1243]
This survey methodology also has the advantage of providing a representative comparison group of individuals who have never been homeless, thus allowing for the investigation of specific risk factors most strongly associated with homelessness. Most past studies have used prevalence rates of risk factors, such as mental illness, and compared them with survey data on the general population, data that are not collected using the same measures, at the same time, or using the same sampling frame. Other methodological advantages of this survey include its having well-validated diagnostic measures, a wide variety of salient covariates for use in multivariate analyses, a representative non-homeless comparison group, and a large sample with broad geographic coverage.
A more substantive explanation for the stronger relationship we found between lifetime substance use diagnoses and past homelessness, compared with a psychiatric diagnosis, is that individuals with serious psychiatric diagnoses have greater access to income assistance that allows them to remain housed. Passage of Public Law 104-121 in 1996 terminated Social Security benefits for individuals who are disabled primarily by a substance abuse disorder. Thus, it is possible that some NCSR survey respondents at one point qualified for and received welfare payments, and later lost this source of income or their benefits were reduced, increasing their risk of homelessness.
A third possible explanation for the stronger association between a substance use diagnoses and past homelessness is that individuals with a substance use disorder may lose access to public housing as a result of the 1988 Anti-Drug Abuse Act. This act, which was strengthened by the Housing Opportunity Extension Act of 1996, requires public housing agencies to use leases that allow for eviction of tenants if the tenant, tenant's family member, or guest engaged in a drug-related crime.
Although we lacked a clear explanation of why psychiatric disorders are not as strongly correlated with past homelessness as substance abuse disorders, we found robust independent associations between both types of illness and past homelessness. These findings confirm in a nationally representative sample findings from a number of single-site studies that have also demonstrated the over-representation of both mental illness and substance abuse disorders in homeless populations,22,23,38 albeit using general community surveys based on different measures for comparison. These findings also highlight the simultaneous need of homeless individuals for access to not only subsistence services (i.e., shelter, meals, and income supports) but also mental health and substance abuse services.
Approximately 58% of all spending for mental health and substance abuse services is funded by federal, state, and local governments39 and such services for the homeless are likely to depend even more on public funding since homeless people generally lack private insurance. However, behavioral health spending has been subject to growing resource constraints,40 which are likely to be exacerbated by the current recession. The possibility of improving access to substance abuse services for homeless people is further impeded by public ambivalence as to whether substance abuse represents bad behavior (poor self control), or a true illness that requires treatment that is effective, medically necessary, and deserving of taxpayer support.41 One example of such policy ambivalence is the restrictions on access to public assistance among individuals with substance abuse problems. Another example is that while there has been extensive [End Page 1244] interest in the development and dissemination of Housing First programs for people with serious psychiatric disorders (with and without co-morbid addictions)42-43 there has been far less effort to develop such models for people with addictive disorders. A recently published study that clearly demonstrated the potential cost-effectiveness of such models for homeless alcoholics actually highlights the fact that such studies have been so few in number.44
Another finding that is of special importance is that individuals who experienced more than 27 days of incarceration in total were much more likely to have experienced homelessness. Incarceration may increase the risk of homelessness by weakening community and family ties as well as by limiting opportunities for employment and access to public housing.45-46 Homelessness itself may also increase the risk for incarceration because homeless individuals may commit crimes in an effort to survive with limited resources. Previous research has made note of this bidirectional relationship by suggesting that many homeless jail inmates became homeless at least in part as a result of prior incarceration.47 Thus, it is possible the strong relationship we found between having experienced at least 27 days incarcerated and past homelessness reflects a population of individuals that cycles through periods of homelessness and incarceration and suggests that there may be a need for more services to assist former inmates in the transition to community settings.
One last finding of this study deserves further comment. This is that the unadjusted odds ratios for the mental health diagnoses (i.e., the results from the bivariate logistic regression analyses) were substantially greater than the odds ratios found after adjusting for other factors. The three measures that most strongly attenuated the strength of the association between psychiatric diagnoses and past homelessness were substance abuse diagnoses, exposure to personal violence, and past receipt of welfare payments. These three measures are thus possible partial mediators of the relationship between mental illness and past homelessness.
This study has a number of methodological limitations. Most importantly, it used a cross-sectional rather than prospective study design. Previously homeless individuals were interviewed at unknown but presumably variable lengths of time since they were last homeless. Many individual characteristics, such as sociodemographic characteristics as well as psychiatric and substances abuse illnesses, are relatively stable over time. However, there might be greater variation over time in characteristics such as employment status and income (i.e., risk factors that precipitated an episode of homelessness may no longer be present once an individual has become domiciled). Other characteristics that may appear to increase the risk for homelessness, such as receipt of welfare payments may have emerged after, rather than before, homelessness. However, given that homelessness is a rare event and thus very difficult to study prospectively, a cross-sectional approach remains a necessary tool for the study of homelessness and can add to our knowledge of potential contributors to homelessness.
A second limitation is that since homelessness was not clearly defined for the respondents (i.e., the respondent was simply asked if they had ever been homeless), there may have been some variability across respondents in how they interpreted the homelessness questions. This data limitation also prevented us from specifically examining subgroups with more significant housing challenges. [End Page 1245]
A third limitation is that the survey depended on the respondent's memory, potentially reducing the reliability of the data. A fourth limitation is that NCSR data on obsessive compulsive disorder and schizophrenia were not publicly available because of doubts about their validity. Although schizophrenia is a rare disorder among the population at large it is more common among homeless individuals.22-23 One final potential limitation of this study is that the institutionalized population was not surveyed. However, the formerly institutionalized were included in the sample, and we thus believe it is unlikely that the results would have been substantially different from those presented here if institutional settings had also been surveyed.
Despite these limitations, the NCSR is the largest national survey of which we are aware that includes well validated measures of psychiatric and substance use disorders and information on past homelessness. Our results suggest that past homelessness is associated with a broad array of sociodemographic, economic, and mental health disadvantages, especially substance abuse. Although prior research has shown that societal-level factors, such as differences in welfare program rules and housing market conditions, are associated with trends over time and differences across regions in homelessness rates,48-50 additional research is needed on how the personal vulnerabilities examined here might interact with societal-level factors to produce homelessness in specific areas of the country and during specific time periods.
There are no potential conflicts of interest in that this work was done as part of the authors' duties as employees of the Veterans Health Administration.
Dr. Greenberg is affiliated with the VA New England Mental Illness, Research, Education and Clinical Center (VA-NE-MIRECC), VAMC West Haven, Conn., and is a Lecturer at Yale University in the Department of Psychiatry. Dr. Rosenheck is affiliated with the VA-NE-MIRECC and is a Professor in the Department of Psychiatry and Yale University School of Epidemiology and Public Health.
1. U.S. Department of Housing and Urban Development Office of Community Planning and Development (HUD). The 2007 annual homeless assessment report: a summary of findings. Washington, DC: HUD, 2008.
2. Burt M, Aron LY, Lee E, et al. Helping America's homeless: emergency shelter or affordable housing? Washington, DC: Urban Institute Press, 2001.
3. Koegel P, Burnam MA, Baumohl J. The causes of homelessness. In: Baumohl J, ed. Homelessness in America. Phoenix, AZ: Oryx Press, 1996.
4. Scott J. Homelessness and mental illness. British Journal of Psychiatry. 1993 Mar;162:314-24.
5. Greenblatt M. Deinstitutionalization and reinstitutionalization of the mentally ill. In: Robertson MJ, Greenblatt M, eds. Homelessness: a national perspective. New York, NY: Plenum Publishing, 1992; 47-56.
6. Leginski W. Historical and contextual influences on the U.S. response to contemporary homelessness. In: Dennis D, Locke G, Khadduri J, eds. Toward understanding homelessness : the 2007 national symposium on homelessness research. Washington, DC: U.S. Department of Housing and Urban Development Office of Policy Development and Research, 2007. Available at: http://aspe.hhs.gov/hsp/homelessness/symposium07/index.htm.
7. Leavitt J. Homelessness and the housing crisis. In: Robertson MJ, Greenblatt M, eds. Homelessness: a national perspective. New York, NY: Plenum Press, 1992. [End Page 1246]
8. Martens WH. A review of physical and mental health in homeless persons. Public Health Rev. 2001;29(1):13-33.
9. Morse GA. Causes of homelessness. In: Robertson M, Greenblatt M, eds. Homelessness: a national perspective. New York, NY: Plenum Press, 1992.
10. Koegel P, Burnam MA. Problems in the assessment of mental illness among the homeless: an empirical approach. In: Robertson M, Greenblatt M, eds. Homelessness: a national perspective. New York, NY: Plenum Press, 1992.
11. Toro PA. International perspectives on homelessness in developed nations. Hoboken, NJ: Wiley-Blackwell, 2007.
12. Hopper K, Milburn N. Homelessness among African Americans: a historical and contemporary perspective. In: Baumohl J, ed. Homelessness in America. Phoenix, AZ: Oryx Press, 1996.
13. Rosenheck R, Fontana A. A model of homelessness among male veterans of the Vietnam-War generation. Am J Psychiatry. 1994 Mar;151(3):421-7.
14. Morse GA, Calsyn R. Mental health and other human service needs and homeless people. In: Robertson M, Greenblatt M, eds. Homelessness: a national perspective. New York, NY: Plenum Press, 1992.
15. Folsom D, Jeste DV. Schizophrenia in homeless persons: a systematic review of the literature. Acta Psychiatr Scand. 2002 Jun;105(6):404-13.
16. Schutt RK, Garrett GR. The homeless alcoholic: past and present. In: Robertson M, Greenblatt M, eds. Homelessness: a national perspective. New York, NY: Plenum Press, 1992.
17. Caton C, Wilkins C, Anderson J. People who experience long-term homelessness: characteristics and interventions. In: Dennis D, Locke G, Khadduri J, eds. Toward understanding homelessness: the 2007 national symposium on homelessness research. Washington, DC: U.S. Department of Housing and Urban Development Office of Policy Development and Research, 2007. Available at: http://aspe.hhs.gov/hsp/homelessness/symposium07/index.htm.
18. Metraux S, Roman C, Cho R. Incarceration and homelessness. In: Dennis D, Locke G, Khadduri J, eds. Toward understanding homelessness: the 2007 national symposium on homelessness research. Washington, DC: U.S. Department of Housing and Urban Development Office of Policy Development and Research, 2007. Available at: http://aspe.hhs.gov/hsp/homelessness/symposium07/index.htm.
19. O'Flaherty B. Making room: the economics of homelessness. Cambridge, MA: Harvard University Press, 1996.
20. Weitzman BC, Knickman JR, Shinn M. Predictors of shelter use among low-income families: psychiatric history, substance abuse, and victimization. Am J Public Health. 1992 Nov;82(11):1547-50.
21. Breakey WR, Fischer PJ, Kramer M, et al. Health and mental health problems of homeless men and women in Baltimore. JAMA. 1989 Sep 8;262(10):1352-7.
22. Koegel P, Burnam MA, Farr RK. The prevalence of specific psychiatric disorders among homeless individuals in the inner city of Los Angeles. Arch Gen Psychiatry. 1988 Dec;45(12):1085-92.
23. Susser E, Struening EL, Conover S. Psychiatric problems in homeless men. Lifetime psychosis, substance use, and current distress in new arrivals at New York City shelters. Arch Gen Psychiatry. 1989 Sep;46(9):845-50.
24. Susser E, Moore R, Link B. Risk factors for homelessness. Epidemiol Rev. 1993;15(2):546-56. [End Page 1247]
25. Kessler RC, Merikangas KR. The National Comorbidity Survey Replication (Ncs-R): background and aims. Int J Methods Psychiatr Res. 2004;13(2):60-8.
26. Heeringa SG, Connor JH, Darrah D. The 1980 SRC/NORC national sample. Ann Arbor, MI: Survey Methodology Program, Survey Research Center, University of Michigan, 1984.
27. Heeringa SG, Connor J, Redmond G. The 1990 SRC national sample. Ann Arbor, MI: Survey Methodology Program, Survey Research Center, University of Michigan, 1994.
28. Heeringa SG, Wagner J, Torres M, et al. Sample designs and sampling methods for the Collaborative Psychiatric Epidemiology Studies (CPES). Int J Methods Psychiatr Res. 2004;13(4):221-40.
29. Druss BG, Wang PS, Sampson NA, et al. Understanding mental health treatment in persons without mental diagnoses: results from the National Comorbidity Survey Replication. Arch Gen Psychiatry. 2007 Oct;64(10):l196-203.
30. Kessler RC, Ustun TB. The World Mental Health (WMH) Survey Initiative Version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI). Int J Methods Psychiatr Res. 2004;13(2):93-121.
31. Kessler RC, Chiu WT, Colpe L, et al. The prevalence and correlates of serious mental illness (SMI) in the National Comorbidity Survey Replication (NCS-R). In: Manderscheid RW, Berry JT, eds. Mental health, United States, 2004. Rockville, MD: Substance Abuse and Mental Health Services Administration, 2006; 134-48.
32. First MB, Spitzer RL, Gibbon M, et al. Structured clinical interview for DSM-IV-TR Axis I disorders, research version, non-patient edition (SCID-I/NP). New York, NY: Biometrics Research, New York State Psychiatric Institute, 2002.
33. Kessler RC, Berglund P, Demler O, et al. Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication (NCS-R). Arch Gen Psychiatry. 2005 Jun;62(6):593-602.
34. Kessler RC, Chi WT, Demler O, et al. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry. 2005 Jun;62(6):617-27.
35. Link BG, Susser E, Stueve A, et al. Lifetime and five-year prevalence of homelessness in the United States. Am J Public Health. 1994 Dec;84(12):1907-12.
36. Toro P, Tompsett C, Philipott P, et al. Homelessness in Europe and the United States: a comparison of prevalence and public opinion. Hoboken, NJ: Wiley-Blackwell, 2007.
37. Tompsett CJ, Toro PA, Guzicki M, et al. Homelessness in the United States: assessing changes in prevalence and public opinion, 1993-2001. Am J Community Psychol. 2006 Mar;37(1-2):47-61.
38. Rossi PH. Down and out in America: the origins of homelessness. Chicago, IL: University of Chicago Press, 1991.
39. Levit KR, Kassed CA, Coffey RM. Future funding for mental health and substance abuse: increasing burdens for the public sector. Health Aff (Millwood). 2008 Nov-Dec;27(6):w513-22. Epub 2008 Oct 7.
40. Appelbaum PS. Responses to the presidential address—the systematic defunding of psychiatric care: a crisis at our doorstep. Am J Psychiatry. 2002 Oct;159(10):1638-40.
41. Wilhite J. Public policy and the homeless alcoholic: rethinking our priorities for treatment programs. In: Robertson MJ, Greenblatt M, eds. Homelessness: a national perspective. New York, NY: Plenum Press, 1992.
42. Tsemberis S, Gulcur L, Nakae M. Housing First, consumer choice, and harm reduction [End Page 1248] for homeless individuals with a dual diagnosis. Am J Public Health. 2004 Apr;94(4):651-6.
43. U.S. Department of Housing and Urban Development, Office of Policy Development and Research (HUD). The applicability of housing first models to homeless persons with serious mental illness. Washington, DC: HUD, 2007.
44. Larimer ME, Malone DK, Garner MD, et al. Health care and public service use and costs before and after provision of housing for chronically homeless persons with severe alcohol problems. JAMA. 2009 Apr 1;301(13):1349-57.
45. Kushel MB, Hahn JA, Evans JL, et al. Revolving doors: imprisonment among the homeless and marginally housed population. Am J Public Health. 2005 Oct;95(10):1747-52.
46. Travis J, Solomon AL, Waul M. From prison to home: the dimensions and consequences of prisoner reentry. Washington, DC: The Urban Institute, 2001.
47. Metraux S, Culhane DP. Recent incarceration history among a sheltered homeless population. Crime and Delinquency. 2006 Jul;52(3):504-17.
48. Burt MR. Over the edge: the growth of homelessness in the 1980s. Washington, DC: Urban Institute Press, 1992.
49. Curtis SE. Social welfare as a dimension of regional development. In: Townroe P, Martin R, eds. Regional development in the 1990s: the British Isles in transition: regional policy and development 4. London, United Kingdom: Spoon Press, 2002.
50. Park JYS. Increased homelessness and low rent housing vacancy rates. J Hous Econ. 2000 Mar;9:76-103. [End Page 1249]