Housing Outcomes of Adults Who Were Homeless at Admission to Substance Use Disorder Treatment Programs Nationwide
Abstract
Objective:
Substance use disorders affect 30%−50% of single homeless adults, and specialized homelessness service programs enable homeless persons to exit homelessness at rates of about 80%. However, many such adults are treated in substance use disorder treatment programs. This study examined housing outcomes in these programs.
Methods:
Data from the Treatment Episode Data Set: Discharges database were used to examine housing status at discharge from substance use disorder treatment programs of adults who were homeless at admission. Associations of outcomes with sociodemographic characteristics, treatment programs and processes, and clinical variables were further evaluated with bivariate and multivariate logistic regressions. Odds ratios of ≥1.5 or ≤0.67 were considered meaningful.
Results:
Of 1,200,105 persons admitted to the programs, 192,838 (16.1%) were homeless at admission; 68.7% remained homeless at discharge, 16.3% were discharged to dependent housing, and only 15.0% were discharged to independent housing. Factors associated with remaining homeless included being age ≥55 years, being unemployed, admission for detoxification (vs. rehabilitation or residential treatment or ambulatory treatment), shorter stays, and program noncompletion. Factors associated with discharge to independent versus dependent housing included employment, admission to nonintensive outpatient treatment, and, unexpectedly, shorter stays.
Conclusions:
Most adults experiencing homelessness at admission to substance use disorder treatment programs remained homeless at discharge, and only half of those no longer homeless were independently housed. These outcomes are considerably worse than outcomes typically reported by specialized homelessness service programs. Evidence-based service models that support exit from homelessness could be provided through augmented internal programming or links with specialized programs.
HIGHLIGHTS
•
Homeless adults in the United States experience high rates of substance use disorders, and nearly 200,000 were admitted in 2018 to federally funded programs for substance use disorder treatment.
•
More than two-thirds of adults who were homeless at admission to these programs remained homeless at discharge, a considerably worse outcome than outcomes reported by specialized homelessness service programs.
•
Of adults who obtained housing at discharge, about half entered dependent housing and the other half obtained independent housing.
•
Factors associated with remaining homeless included older age, lack of employment, being admitted for detoxification, shorter length of stay, and noncompletion of the treatment program.
Homelessness represents an ongoing public health crisis in the United States, with >580,000 individuals experiencing homelessness on a given night in 2020 (1). It is estimated that 30%−50% of adults experiencing homelessness have a substance use disorder (2–6), and alcohol and drug use disorders are perhaps the most important health-related risk factors for chronic homelessness (7, 8).
Although substance use disorders involving alcohol and drugs confer substantial risk for homelessness, interventions such as subsidized and supportive housing have been shown to facilitate exits from homelessness as effectively for people with these disorders as for others (9–11). For example, permanent supportive housing and Housing First programs have consistently enabled 80% of participants to obtain and retain independent housing (12, 13) and, to a lesser extent, have reduced alcohol and drug use and improved community functioning and quality of life (14–18).
Programs to treat patients with substance use disorders are a potentially important but underutilized point of contact for housing intervention. Although many homeless adults with substance use disorders are admitted to treatment programs for such disorders, few studies have examined the role and effectiveness of such programs in addressing homelessness (19, 20).
This study used a comprehensive database of patients discharged from federally funded substance use disorder treatment programs nationwide, the 2018 Treatment Episode Data Set: Discharges (TEDS-D), to estimate the proportion of adults admitted to such programs who were homeless and who remained so at discharge. It further examined specific sociodemographic characteristics, treatment programs and processes, and clinical variables that are associated with major housing outcomes, with the goal of identifying opportunities for interventions that could facilitate exit from homelessness.
Methods
Data Sources and Study Sample
Data were retrieved from the TEDS-D, a data set compiled annually by the Substance Abuse and Mental Health Services Administration concerning users of substance use disorder treatment facilities in the United States that receive federal funding. This nationwide data set comprises information on demographic, socioeconomic, clinical, and other characteristics of individuals using these services (21). This study included adults ages ≥18 years who were homeless on admission to a substance use disorder treatment facility and who were discharged during 2018. Institutional review board approval was waived because the data were public and deidentified.
Measures
The TEDS-D data set defines “homeless” as having no fixed address or residing in a shelter (22). “Dependent living” includes supervised settings, such as residential programs, halfway homes, or group homes. “Independent living” is defined as living alone or with others in a private residence and being capable of self-care. Independent living includes permanent supportive housing and scattered-site voucher programs in which case management is available for support.
Sociodemographic characteristics documented for patients discharged during the study period included age (18–34, 35–54, or ≥55 years), sex (male or female), race (non-Hispanic White, non-Hispanic Black, Hispanic, or other), marital status (never married; currently married; or separated, divorced, or widowed), U.S. region (Northeast, Midwest, South, or West), education level (less than high school, high school or equivalent, or more than high school), employment status (full-time, part-time, or not employed), veteran status (yes or no), primary source of income (wages or salary; public assistance; retirement, pension, or disability; or other), and arrests in the 30 days before admission (none, once, or two or more times).
Program characteristics examined included service setting at admission (detoxification, rehabilitation or residential, intensive outpatient, or nonintensive outpatient). Patient-level measures of service delivery included reason for discharge (completed treatment, dropped out or terminated, or incarcerated or transferred to another facility), length of stay in treatment (1–30, 31–60, 61–120, or ≥121 days), referral source (individual, health care provider, work, or court or criminal justice), and primary type of substance use at admission (alcohol, opioids, marijuana, cocaine and methamphetamine, or other).
Statistical Analysis
After descriptive analyses of residential status, bivariate logistic regression analyses were used to examine the association of sociodemographic and clinical variables with each of two dependent variables: homeless status at discharge (homeless vs. housed) and the type of housing (independent vs. dependent). Independent variables included the sociodemographic and clinical characteristics noted above.
Because of the large sample sizes, statistical significance was achieved for most variables, even if the magnitude of the associations was small. As a result, effect sizes were used to identify substantially meaningful associations by using published thresholds for odds ratios (ORs) of ≥1.5 or ≤0.67 (23). These thresholds were somewhat lower than the 2.0–0.5 minimum standard recommended by Ferguson (23) as the “minimum effect size representing a ‘practically’ significant effect” and were therefore liberal standards for identifying meaningful associations, making it unlikely that any meaningful relationships were overlooked. Multivariate logistic regression analysis was then used to identify factors independently associated with remaining homeless at discharge and, among those no longer homeless, with becoming independently versus dependently housed at discharge.
Results
The TEDS-D data included 1,200,105 admissions of persons discharged from substance use disorder treatment programs during 2018, including 192,838 (16.1%) who had been homeless at the time of admission. Of these homeless adults, 132,561 (68.7%) remained homeless at discharge, 31,366 (16.3%) were discharged to dependent housing, and 28,911 (15.0%) were discharged to independent housing.
Correlates of Homelessness at Discharge
Bivariate analyses.
Bivariate analysis (Tables 1 and 2) showed that individuals ages ≥55 years and those with either no employment or part-time employment were substantially more likely than those who were younger and employed full-time to be homeless at discharge. Admission to a substance detoxification facility was substantially more likely to be associated with remaining homeless than was admission to residential or ambulatory treatment, as was treatment noncompletion. On the other hand, longer stay (>30 days) in the substance use disorder treatment program was associated with substantially less risk for homelessness at discharge, as was discharge from programs located in the Northeast. Admission primarily for opioid use was, unexpectedly, associated with a reduced risk for homelessness at discharge (OR=0.62), compared with admission for problematic use of alcohol.
| Bivariate logistic regression analysis | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Homeless (N=132,561, 68.7%) | Dependent living (N=31,366, 16.3%) | Independent living (N=28,911, 15.0%) | Homeless vs. dependent or independent living | Independent living vs. dependent living | ||||||||
| Characteristic | N | % | N | % | N | % | OR | 95% CI | p | OR | 95% CI | p |
| Age in years | ||||||||||||
| 18–34 (reference) | 49,614 | 65.4 | 13,799 | 18.2 | 12,429 | 6.4 | ||||||
| 35–54 | 64,378 | 70.0 | 14,336 | 15.6 | 13,426 | 14.6 | 1.23 | 1.21–1.25 | <.001 | 1.04 | 1.01–1.08 | .024 |
| ≥55 | 18,569 | 74.7 | 3,231 | 13.0 | 3,056 | 12.3 | 1.56b | 1.51–1.61 | <.001 | 1.05 | .99–1.11 | .082 |
| Sex | ||||||||||||
| Male (reference) | 94,940 | 69.6 | 22,222 | 16.3 | 19,310 | 14.2 | ||||||
| Female | 37,763 | 66.7 | 9,228 | 16.3 | 9,628 | 17.0 | .88 | .86–.89 | <.001 | 1.20 | 1.16–1.24 | <.001 |
| Unknownc | 45 | 71.4 | 11 | 17.5 | 7 | 11.1 | ||||||
| Race-ethnicity | ||||||||||||
| Non-Hispanic White (reference) | 74,352 | 67.6 | 18,780 | 17.1 | 16,883 | 15.4 | ||||||
| Non-Hispanic Black | 26,994 | 70.0 | 6,148 | 15.9 | 5,450 | 15.4 | 1.12 | 1.09–1.14 | <.001 | .99 | .95–1.03 | .513 |
| Hispanic | 19,861 | 68.8 | 4,365 | 15.1 | 4,653 | 16.1 | 1.06 | 1.03–1.09 | <.001 | 1.19 | 1.13–1.24 | <.001 |
| Other | 10,854 | 73.6 | 2,080 | 14.1 | 1,816 | 12.3 | 1.34 | 1.29–1.39 | <.001 | .97 | .91–1.04 | .387 |
| Unknownc | 687 | 74.8 | 88 | 9.5 | 143 | 15.6 | ||||||
| Marital status | ||||||||||||
| Never married (reference) | 77,534 | 67.8 | 19,758 | 17.3 | 17,036 | 14.9 | ||||||
| Currently married | 6,802 | 67.9 | 1,379 | 13.8 | 1,842 | 18.4 | 1.00 | .96–1.05 | .923 | 1.55b | 1.44–1.67 | <.001 |
| Divorced, separated, or widowed | 30,140 | 72.0 | 6,019 | 14.4 | 5,733 | 13.7 | 1.22 | 1.19–1.25 | <.001 | 1.10 | 1.06–1.15 | <.001 |
| Unknownc | 18,272 | 67.9 | 4,305 | 16.0 | 4,334 | 16.1 | ||||||
| U.S. region | ||||||||||||
| Northeast (reference) | 37,576 | 58.7 | 12,250 | 19.1 | 14,168 | 22.1 | ||||||
| Midwest | 30,974 | 71.6 | 7,472 | 17.3 | 4,805 | 11.1 | 1.77b | 1.73–1.82 | <.001 | .56b | .53–.58 | <.001 |
| South | 25,647 | 69.1 | 6,979 | 18.8 | 4,515 | 12.2 | 1.57b | 1.53–1.61 | <.001 | .56b | .53–.58 | <.001 |
| West | 38,551 | 79.1 | 4,760 | 9.8 | 5,457 | 11.2 | 2.65b | 2.58–2.73 | <.001 | .99 | .95–1.04 | .706 |
| Education | ||||||||||||
| Less than high school (reference) | 37,080 | 69.1 | 8,550 | 15.9 | 8,040 | 15.0 | ||||||
| High school or equivalent | 64,051 | 69.3 | 14,592 | 15.8 | 13,785 | 14.9 | 1.01 | .99–1.03 | .403 | 1.00 | .97–1.04 | .814 |
| More than high school | 30,131 | 67.02 | 7,962 | 17.7 | 6,863 | 15.3 | .91 | .89–.93 | <.001 | .92 | .88–.96 | <.001 |
| Unknownc | 1,486 | 70.8 | 357 | 17.0 | 257 | 12.2 | ||||||
| Employment at discharge | ||||||||||||
| Full-time (reference) | 6,534 | 50.3 | 1,885 | 14.5 | 4,562 | 35.1 | ||||||
| Part-time | 4,749 | 61.6 | 1,072 | 13.9 | 1,891 | 24.5 | 1.58b | 1.49–1.67 | <.001 | .73 | .66–.80 | <.001 |
| Not employed | 119,498 | 70.6 | 27,886 | 16.5 | 21,938 | 13.0 | 2.37b | 2.28–2.45 | <.001 | .33b | .31–.34 | <.001 |
| Unknownc | 1,967 | 62.7 | 618 | 19.7 | 554 | 17.6 | ||||||
| Veteran status | ||||||||||||
| Yes | 4,406 | 72.3 | 854 | 14.0 | 834 | 13.7 | 1.20 | 1.14–1.27 | <.001 | 1.06 | .96–1.17 | .233 |
| No (reference) | 121,045 | 68.4 | 29,075 | 16.4 | 26,770 | 15.1 | ||||||
| Unknownc | 7,297 | 71.8 | 1,532 | 15.1 | 1,341 | 13.2 | ||||||
| Primary source of income | ||||||||||||
| Wages or salary (reference) | 8,205 | 67.1 | 1,800 | 14.7 | 2,216 | 18.1 | ||||||
| Public assistance | 8,384 | 59.9 | 2,873 | 20.5 | 2,731 | 19.5 | .73 | .70–.77 | <.001 | .77 | .71–.84 | <.001 |
| Retirement, pension, or disability | 6,995 | 67.3 | 1,969 | 18.9 | 1,436 | 13.8 | 1.01 | .95–1.06 | .847 | .59b | .54–.65 | <.001 |
| Other | 62,808 | 66.9 | 16,151 | 17.2 | 14,947 | 15.9 | .99 | .95–1.03 | .574 | .75 | .70–.90 | <.001 |
| Unknownc | 46,356 | 74.0 | 8,668 | 13.8 | 7,615 | 12.2 | ||||||
| Arrests in 30 days before admission | ||||||||||||
| 0 (reference) | 121,358 | 69.1 | 28,165 | 16.0 | 26,231 | 14.9 | ||||||
| 1 | 9,149 | 65.7 | 2,661 | 19.1 | 2,107 | 15.1 | .86 | .83–.89 | <.001 | .85 | .80–.90 | <.001 |
| ≥2 | 1,710 | 64.9 | 415 | 15.8 | 509 | 19.3 | .83 | .77–.90 | <.001 | 1.32 | 1.16–1.50 | <.001 |
| Unknownc | 531 | 62.5 | 220 | 25.9 | 98 | 11.5 | ||||||
a
Data are from the Treatment Episode Data Set: Discharges. For the descriptive statistics, the percentages are row percentages.
b
Substantial effect (odds ratio ≥1.5 or ≤.67).
c
Category was not considered in the bivariate analyses using a listwise deletion process.
| Bivariate logistic regression analysis | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Homeless (N=132,561, 68.7%) | Dependent living (N=31,366, 16.3%) | Independent living (N=28,911, 15.0%) | Homeless vs. dependent or independent living | Independent living vs. dependent living | ||||||||
| Characteristic | N | % | N | % | N | % | OR | 95% CI | p | OR | 95% CI | p |
| Type of service setting at admission | ||||||||||||
| Detoxification (reference) | 53,593 | 81.0 | 7,621 | 11.5 | 4,956 | 7.5 | ||||||
| Rehabilitation or residential | 39,057 | 57.3 | 17,113 | 25.1 | 11,988 | 17.6 | .31b | .31–.32 | <.001 | 1.08 | 1.03–1.12 | .001 |
| Ambulatory, intensive outpatient | 9,935 | 59.6 | 3,254 | 19.5 | 3,477 | 20.9 | .35b | .33–.36 | <.001 | 1.64b | 1.55–1.74 | <.001 |
| Ambulatory, nonintensive outpatient | 30,163 | 71.5 | 3,473 | 8.2 | 8,524 | 20.2 | .59b | .57–.61 | <.001 | 3.77b | 3.58–3.98 | <.001 |
| Reason for discharge | ||||||||||||
| Treatment completed (reference) | 56,160 | 62.3 | 18,379 | 20.4 | 15,570 | 17.3 | ||||||
| Dropped out of treatment or terminated by facility | 38,405 | 72.4 | 5,860 | 11.1 | 8,790 | 16.6 | 1.58b | 1.55–1.62 | <.001 | 1.77b | 1.70–1.84 | <.001 |
| Incarcerated or transferred to another facility | 29,893 | 75.6 | 6,406 | 16.2 | 3,221 | 8.2 | 1.88b | 1.83–1.93 | <.001 | .59b | .57–.62 | <.001 |
| Other | 8,290 | 79.2 | 816 | 7.8 | 1,364 | 13.0 | 2.30b | 2.19–2.41 | <.001 | 1.97b | 1.80–2.16 | <.001 |
| Length of stay in treatment, days | ||||||||||||
| 1–30 (reference) | 23,188 | 94.1 | 560 | 2.3 | 904 | 3.7 | ||||||
| 31–60 | 88,719 | 69.7 | 22,976 | 18.1 | 15,539 | 12.2 | .15b | .14–.15 | <.001 | .42b | .38–.47 | <.001 |
| 61–120 | 10,798 | 53.3 | 4,609 | 2.8 | 4,855 | 24.0 | .07b | .07–.08 | <.001 | .65b | .59–.73 | <.001 |
| ≥121 | 10,043 | 47.8 | 3,316 | 15.8 | 7,647 | 36.4 | .06b | .05–.06 | <.001 | 1.43 | 1.28–1.60 | <.001 |
| Referral source | ||||||||||||
| Individual, including self-referral (reference) | 63,237 | 69.6 | 14,740 | 16.2 | 12,893 | 14.2 | ||||||
| Health care provider | 26,626 | 63.2 | 8,738 | 20.8 | 6,752 | 16.0 | .75 | .73–.77 | <.001 | .88 | .85–.92 | <.001 |
| Work | 18,849 | 74.7 | 3,008 | 11.9 | 3,385 | 13.4 | 1.29 | 1.25–1.33 | <.001 | 1.29 | 1.22–1.36 | <.001 |
| Court or criminal justice, DUI or DWIc | 21,232 | 67.1 | 4,838 | 15.3 | 5,582 | 17.6 | .89 | .87–.92 | <.001 | 1.32 | 1.26–1.38 | <.001 |
| Unknownd | 2,804 | 85.6 | 137 | 4.2 | 333 | 10.2 | ||||||
| Primary substance use at admission | ||||||||||||
| Alcohol (reference) | 52,221 | 74.0 | 9,632 | 13.6 | 8,757 | 12.4 | ||||||
| Opioids | 40,804 | 63.7 | 12,146 | 19.0 | 11,114 | 17.4 | .62b | .60–.63 | <.001 | 1.01 | .97–1.05 | .744 |
| Marijuana | 7,063 | 68.6 | 1,217 | 11.8 | 2,019 | 19.6 | .77 | .74–.80 | <.001 | 1.82b | 1.69–1.97 | <.001 |
| Cocaine and methamphetamine | 28,646 | 67.6 | 7,517 | 17.7 | 6,222 | 14.7 | .73 | .72–.75 | <.001 | .91 | .87–.95 | <.001 |
| Other | 3,918 | 69.3 | 931 | 16.5 | 809 | 14.3 | .79 | .75–.84 | <.001 | .96 | .87–1.05 | .368 |
| Unknownd | 106 | 71.6 | 18 | 12.2 | 24 | 16.2 | ||||||
a
Data are from the Treatment Episode Data Set: Discharges. For the descriptive statistics, the percentages are row percentages.
b
Substantial effect (odds ratio ≥1.5 or ≤.67).
c
DUI or DWI, driving under the influence or driving while intoxicated.
d
Category was not considered in the bivariate analyses using a listwise deletion process.
Multivariate analyses of homelessness.
Multivariate logistic regression showed that being age ≥55 remained an independent risk factor for homelessness at discharge (OR=1.51) (Table 3). Lack of employment (OR=1.63) was again associated with homelessness at discharge, but working part-time no longer correlated substantially with homelessness at discharge in the multivariate analysis. Residing in the Midwest region (OR=1.98) and especially the West region (OR= 4.90) of the United States was associated with greater likelihood of remaining homeless, compared with residing in the Northeast or South. Being admitted for detoxification was also associated with a higher risk for remaining homeless. Successful treatment completion (i.e., not dropping out of the substance use disorder treatment program or being transferred to another facility) remained protective against continued homelessness, as did length of stay of >30 days, especially when the stay was ≥121 days (OR=0.08). The primary substance of use at admission was no longer substantially associated with homelessness in the multivariate analysis.
| Variable | OR | 95% CI | p |
|---|---|---|---|
| Age (reference: 18–34 years) | |||
| 35–54 | 1.23 | 1.20–1.27 | <.001 |
| ≥55 | 1.51b | 1.44–1.59 | <.001 |
| Race-ethnicity (reference: non-Hispanic White) | |||
| Non-Hispanic Black | 1.38 | 1.34–1.43 | <.001 |
| Hispanic | 1.18 | 1.14–1.23 | <.001 |
| Other | 1.18 | 1.12–1.25 | <.001 |
| Region (reference: Northeast) | |||
| Midwest | 1.98b | 1.91–2.06 | <.001 |
| South | 1.36 | 1.31–1.42 | <.001 |
| West | 4.90b | 4.66–5.16 | <.001 |
| Employment at discharge (reference: full-time) | |||
| Part-time | 1.28 | 1.18–1.38 | <.001 |
| Not employed | 1.63b | 1.55–1.73 | <.001 |
| Primary source of income (reference: wages or salary) | |||
| Public assistance | .85 | .80–.91 | <.001 |
| Retirement, pension, or disability | .71 | .66–.76 | <.001 |
| Other | .99 | .94–1.04 | .789 |
| Arrests in 30 days before admission (reference: none) | |||
| 1 | .86 | .82–.91 | <.001 |
| ≥2 | .73 | .65–.82 | <.001 |
| Type of service setting at admission (reference: detoxification) | |||
| Rehabilitation or residential | .53b | .51–.55 | <.001 |
| Ambulatory, intensive outpatient | .47b | .44–.49 | <.001 |
| Ambulatory, nonintensive outpatient | 1.08 | 1.02–1.13 | .003 |
| Reason for discharge (reference: treatment completed) | |||
| Dropped out of treatment or terminated by facility | 2.28b | 2.21–2.36 | <.001 |
| Incarcerated or transferred to another facility | 2.09b | 2.00–2.18 | <.001 |
| Other | 3.15b | 2.93–3.38 | <.001 |
| Length of stay in treatment, days (reference: 1–30) | |||
| 31–60 | .22b | .20–.23 | <.001 |
| 61–120 | .10b | .10–.11 | <.001 |
| ≥121 | .08b | .08–.09 | <.001 |
| Primary substance use at admission (reference: alcohol) | |||
| Opioids | .71 | .69–.73 | <.001 |
| Marijuana | 1.07 | 1.01–1.14 | .024 |
| Cocaine and methamphetamine | .83 | .80–.86 | <.001 |
| Other | .97 | .89–1.05 | .389 |
a
Results of multivariable analysis; data are from the Treatment Episode Data Set: Discharges.
b
Substantial effect (odds ratio ≥1.5 or ≤.67).
Correlates of Independent Housing Among Those Not Homeless at Discharge
Bivariate analyses.
Among those who exited homelessness at discharge, being currently married was associated with greater likelihood of obtaining independent housing in bivariate analyses (Tables 1 and 2). Participation in substance use disorder treatment programs in the Midwest or South was associated with less likelihood of obtaining independent housing, compared with participation in programs in the West or Northeast, as was nonemployment (compared with full-time employment) and receipt of retirement, pension, or disability income (compared with receipt of wages or salary). Admission to either intensive or nonintensive outpatient treatment was substantially associated with obtaining independent housing at discharge (OR=1.64 and OR=3.77, respectively). Those who dropped out of treatment or were terminated prematurely or discharged early for another reason (other than incarceration or transfer) but who obtained housing at discharge were more likely to obtain independent housing than those who completed treatment or who were incarcerated or transferred. Although results indicated that longer stays in treatment were associated with a reduced likelihood of remaining homeless, longer stays were not associated with being discharged to independent living. Compared with alcohol use, only marijuana use, but not use of other drugs, was associated with greater odds of obtaining independent living at discharge (OR=1.82).
Multivariate analysis of independent housing.
Results from multivariate logistic regression indicated that after adjustment for all of the factors that were substantially associated in bivariate analyses with being housed at discharge or being independently housed (see details of these analyses above), being married was no longer associated with independent living at discharge (Table 4). Those discharged from programs in the West had the greatest likelihood of obtaining independent living, compared with any other region (OR=1.53). Similarly, those admitted to a nonintensive outpatient ambulatory program had the highest likelihood of obtaining independent living, compared with admission to other program types (OR=2.28). As in the bivariate analysis, nonemployment reduced the likelihood of obtaining independent living at discharge, as did early termination of treatment because of incarceration or transfer (OR=0.56). Shorter stays in treatment (1–30 days) were more strongly associated with obtaining independent housing at discharge, compared with longer stays.
| Variable | OR | 95% CI | p |
|---|---|---|---|
| Female (reference: male) | 1.28 | 1.22–1.33 | <.001 |
| Marital status (reference: never married) | |||
| Currently married | 1.31 | 1.21–1.42 | <.001 |
| Divorced, separated, or widowed | 1.16 | 1.11–1.22 | <.001 |
| U.S. region (reference: Northeast) | |||
| Midwest | .47b | .45–.50 | <.001 |
| South | .53b | .51–.56 | <.001 |
| West | 1.53b | 1.39–1.68 | <.001 |
| Employment at discharge (reference: full-time) | |||
| Part-time | .68 | .61–.77 | <.001 |
| Not employed | .40b | .37–.43 | <.001 |
| Arrests in 30 days before admission (reference: none) | |||
| 1 | .91 | .85–.98 | .009 |
| ≥2 | 1.44 | 1.23–1.69 | <.001 |
| Type of service setting at admission (reference: detoxification) | |||
| Rehabilitation or residential | .94 | .89–.99 | .030 |
| Ambulatory, intensive outpatient | 1.42 | 1.31–1.53 | <.001 |
| Ambulatory, nonintensive outpatient | 2.28b | 2.11–2.45 | <.001 |
| Reason for discharge (reference: treatment completed) | |||
| Dropped out of treatment or terminated by facility | 1.59b | 1.52–1.67 | <.001 |
| Incarcerated or transferred to another facility | .56b | .52–.59 | <.001 |
| Other | 1.12 | 1.00–1.25 | .044 |
| Length of stay in treatment, days (reference: 1–30) | |||
| 31–60 | .51b | .45–.57 | <.001 |
| 61–120 | .54b | .47–.62 | <.001 |
| ≥121 | .89 | .78–1.02 | .096 |
| Referral source (reference: individual, including self-referral) | |||
| Health care provider | .77 | .74–.81 | <.001 |
| Work | .93 | .87–.99 | .022 |
| Court or criminal justice referral, DUI or DWIc | .97 | .91–1.03 | .288 |
| Primary substance use at admission (reference: alcohol) | |||
| Opioids | .90 | .86–.94 | <.001 |
| Marijuana | 1.23 | 1.12–1.36 | <.001 |
| Cocaine and methamphetamine | .94 | .88–.99 | .020 |
| Other | .94 | .84–1.05 | .255 |
a
Results of multivariable analysis; data are from the Treatment Episode Data Set: Discharges.
b
Substantial effect (odds ratio ≥1.5 or ≤.67).
c
DUI or DWI, driving under the influence or driving while intoxicated.
Discussion
Of 1,200,105 patients admitted to federally funded substance use disorder treatment facilities, 16.1% were homeless at admission, and 68.7% of these patients remained homeless at discharge. Among the approximately 31% who obtained housing, fewer than half were discharged to an independent living setting. Those remaining homeless tended to be ages ≥55 years, had no employment, were admitted for detoxification, had shorter lengths of stay in the treatment program, or did not complete the program. Those discharged to independent housing tended to be employed and admitted to nonintensive outpatient treatment and, unexpectedly, had shorter stays in treatment.
These outcomes contrast with published outcomes from specialized homelessness service programs that have shown ability to house approximately 80% of homeless individuals admitted, most consistently through subsidized housing or Housing First models but also through time-limited residential treatment programs (12, 13). Our results are similar to evidence indicating transient but not long-term housing benefits of residential substance use disorder treatment among veterans (19); the few other reports show relatively poor housing outcomes as well (20).
Consistent with literature suggesting that unemployment is a major risk factor for homelessness (7), a lack of employment was an independent risk factor for remaining homeless and led to a lower likelihood of obtaining independent living. Recipients of retirement, pension, or disability income were less likely to obtain independent living than were individuals receiving wages or salary, perhaps reflecting the inadequate funds associated with disability programs such as Social Security (24) or the more severe disabilities required to qualify for such programs.
Given that affordability of housing is associated with increased exit rates from homelessness (25) and that affordable housing shortages affect especially the western United States (26), it was not surprising that being located in the West conferred the greatest risk of remaining homeless at discharge; however, those who became housed were more likely to obtain independent housing rather than dependent housing. It is unclear whether this latter finding reflects the greater availability of government subsidies for independent versus dependent living arrangements or other regional public policies in the West.
Our finding that a longer stay in treatment protected against homelessness at discharge is consistent with previous studies that have shown that longer stays allow time for obtaining housing before discharge (12) and that it can take on average 3 months to arrange subsidies, even when a Section 8 housing voucher is available (27). Detoxification facilities tend to have limited goals and shorter stays, which may explain why these programs had poorer housing outcomes.
However, longer stays were not associated with achieving independent housing, pointing to the possibly greater availability of dependent housing situations for formerly homeless adults or greater severity of illnesses among those requiring a longer stay. Terminating treatment prematurely led to a higher risk for remaining homeless; however, premature termination was unexpectedly associated with a greater likelihood of obtaining independent housing on discharge rather than dependent housing. Perhaps some of those who left treatment early did so because they had found independent housing. Those who terminated treatment early may also have had less severe clinical conditions, which may have made it easier to obtain independent housing.
It was unexpected that older patients would be more likely than others to remain homeless on discharge, because being age >65 is typically protective against homelessness (28), in part because of access to income supports such as Social Security. Many chronically ill adults ages >55 receive Social Security benefits early through the Social Security Disability program. Older age is, on the other hand, a specific risk factor for chronic homelessness, which could have impeded exit from homelessness, but the TEDS-D data set did not specify chronicity of homelessness (29).
These findings have implications for policy making, pointing to a need for more resources for both dependent and independent housing, especially in regions where affordable housing access is limited. Substance use disorder treatment programs should consider collaborating with existing housing programs or implementing integrated housing services; indeed, it has been shown that increased integration of mental health and housing services is positively correlated with housing outcomes (30). Housing support through grant funding from the Continuum of Care Program is available for nonprofit organizations, state and local governments, and housing agencies. Substance use disorder treatment programs may consider pursuing these funding streams to achieve greater success in housing outcomes.
This study had several limitations. The TEDS-D database captures only individuals receiving services at substance use disorder treatment facilities that are directly federally funded, and these results therefore may not be generalizable to private, nonprofit, or Veterans Affairs programs. Second, limited data were available on the services provided. Some programs may, in fact, have offered housing services, the nature of which remained unknown. Housing outcomes may also be affected by the local availability of low-income housing; federal, state, or local housing subsidies; and insurance coverage limits on length of stay; however, this information was not documented. Even when adequate housing services were available, patients may not have moved into their new homes at the time of discharge but may have completed the process some time thereafter. Reasons for early treatment termination were not fully described in TEDS-D; therefore, only limited hypotheses can be made about why early termination was associated with obtaining independent housing. In addition, the data did not differentiate between single homeless adults and household heads of homeless families, a distinction that can significantly affect the availability of housing assistance. It is notable that in this very large sample, most observed differences were highly significant, even after adjustment for multiple comparisons. We therefore used effect sizes as measures to identify substantial differences by using published norms. Finally, data on psychiatric comorbid conditions, clinical acuity, and level of disability were not available but are likely important risk factors for remaining homeless at discharge.
Conclusions
Despite its limitations, this study’s findings show that of almost 200,000 homeless adults treated in 2018 in federally funded substance use disorder treatment programs, more than two-thirds had disappointing housing outcomes, especially when compared with outcomes achieved by specialized homelessness service programs that emphasize linkage with housing resources. Administrators and policy makers responsible for such programs may improve outcomes by augmenting their provision of housing services or by developing collaborations with programs that have experience providing effective, evidence-based housing services.
Footnote
The authors report no financial relationships with commercial interests.
References
1.
HUD 2019 Continuum of Care Homeless Assistance Programs Homeless Populations and Subpopulations. Washington, DC, US Department of Housing and Urban Development, 2019. https://files.hudexchange.info/reports/published/CoC_PopSub_NatlTerrDC_2019.pdf
2.
Greenberg GA, Rosenheck RA: Correlates of past homelessness in the National Epidemiological Survey on Alcohol and Related Conditions. Adm Policy Ment Health 2010; 37:357–366
3.
Tsai J, Kasprow WJ, Rosenheck RA: Alcohol and drug use disorders among homeless veterans: prevalence and association with supported housing outcomes. Addict Behav 2014; 39:455–460
4.
Fazel S, Khosla V, Doll H, et al: The prevalence of mental disorders among the homeless in western countries: systematic review and meta-regression analysis. PLoS Med 2008; 5:e225
5.
Tsai J: Lifetime and 1-year prevalence of homelessness in the US population: results from the National Epidemiologic Survey on Alcohol and Related Conditions–III. J Public Health 2018; 40:65–74
6.
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; 45:1085–1092
7.
Tsai J, Rosenheck RA: Risk factors for homelessness among US veterans. Epidemiol Rev 2015; 37:177–195
8.
Baggett TP, Hwang SW, O’Connell JJ, et al: Mortality among homeless adults in Boston: shifts in causes of death over a 15-year period. JAMA Intern Med 2013; 173:189–195
9.
Hurlburt MS, Hough RL, Wood PA: Effects of substance abuse on housing stability of homeless mentally Ill persons in supported housing. Psychiatr Serv 1996; 47:731–736
10.
O’Connell MJ, Kasprow W, Rosenheck RA: Rates and risk factors for homelessness after successful housing in a sample of formerly homeless veterans. Psychiatr Serv 2008; 59:268–275
11.
Edens EL, Mares AS, Tsai J, et al: Does active substance use at housing entry impair outcomes in supported housing for chronically homeless persons? Psychiatr Serv 2011; 62:171–178
12.
McGuire J, Rosenheck RA, Kasprow WJ: Patient and program predictors of 12-month outcomes for homeless veterans following discharge from time-limited residential treatment. Adm Policy Ment Health 2011; 38:142–154
13.
Tsemberis S, Gulcur L, Nakae M: Housing First, consumer choice, and harm reduction for homeless individuals with a dual diagnosis. Am J Public Health 2004; 94:651–656
14.
Padgett DK, Stanhope V, Henwood BF, et al: Substance use outcomes among homeless clients with serious mental illness: comparing Housing First with Treatment First programs. Community Ment Health J 2011; 47:227–232
15.
Cheng AL, Lin H, Kasprow W, et al: Impact of supported housing on clinical outcomes: analysis of a randomized trial using multiple imputation technique. J Nerv Ment Dis 2007; 195:83–88
16.
Aubry T, Nelson G, Tsemberis S: Housing First for people with severe mental illness who are homeless: a review of the research and findings from the At Home–Chez soi demonstration project. Can J Psychiatry 2015; 60:467–474
17.
Tsai J, Rosenheck RA, Kasprow WJ, et al: Sobriety as an admission criterion for transitional housing: a multi-site comparison of programs with a sobriety requirement to programs with no sobriety requirement. Drug Alcohol Depend 2012; 125:223–229
18.
Rosenheck R, Kasprow W, Frisman L, et al: Cost-effectiveness of supported housing for homeless persons with mental illness. Arch Gen Psychiatry 2003; 60:940–951
19.
Conrad KJ, Hultman CI, Pope AR, et al: Case managed residential care for homeless addicted veterans: results of a true experiment. Med Care 1998; 36:40–53
20.
Rosenheck R, Leda C, Gallup P: Program design and clinical operation of two national VA programs for homeless mentally ill veterans. N Engl J Public Policy 1992; 8:315–337
21.
Combined Substance Use and Mental Health Treatment Episode Data Set (TEDS) State Instruction Manual, Version 4.3.1, With Data Submission System (DSS) Guide. Rockville, MD, Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality, 2019
22.
Treatment Episode Data Set Discharges (TEDS-D). Rockville, MD, Substance Abuse and Mental Health Services Administration, 2018. https://www.datafiles.samhsa.gov/dataset/teds-d-2018-ds0001-teds-d-2018-ds0001. Accessed Nov 23, 2021
23.
Ferguson CJ: An effect size primer: a guide for clinicians and researchers. Prof Psychol Res Pr 2009; 40:532–538
24.
Schaak G, Sloane L, Arienti F, et al: Priced Out: The Housing Crisis for People With Disabilities. Boston, Technical Assistance Collaborative, 2017
25.
Rosenheck R, Morrissey J, Lam J, et al: Service delivery and community: social capital, service systems integration, and outcomes among homeless persons with severe mental illness. Health Serv Res 2001; 36:691–710
26.
Anthony J: Economic prosperity and housing affordability in the United States: lessons from the booming 1990s. Hous Policy Debate 2018; 28:325–341
27.
O’Connell M, Kasprow W, Rosenheck RA: National dissemination of supported housing in the VA: model adherence versus model modification. Psychiatr Rehabil J 2010; 33:308–319
28.
Edens EL, Kasprow W, Tsai J, et al: Association of substance use and VA service–connected disability benefits with risk of homelessness among veterans. Am J Addict 2011; 20:412–419
29.
Caton CLM, Dominguez B, Schanzer B, et al: Risk factors for long-term homelessness: findings from a longitudinal study of first-time homeless single adults. Am J Public Health 2005; 95:1753–1759
30.
Rosenheck RA, Lam J, Morrissey JP, et al: Service systems integration and outcomes for mentally ill homeless persons in the ACCESS program. Psychiatr Serv 2002; 53:958–966
Information & Authors
Information
Published In
History
Received: 16 July 2021
Revision received: 17 September 2021
Accepted: 1 November 2021
Published online: 18 January 2022
Published in print: August 01, 2022
Keywords
Authors
Funding Information
Dr. Lo’s work was funded in part by the Department of Mental Health and Addiction Services, State of Connecticut.This article does not express the views of the Department of Mental Health and Addiction Services or the State of Connecticut. The views and opinions expressed are those of the authors.
Metrics & Citations
Metrics
Citations
Export Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
For more information or tips please see 'Downloading to a citation manager' in the Help menu.
View Options
View options
PDF/EPUB
View PDF/EPUBLogin options
Already a subscriber? Access your subscription through your login credentials or your institution for full access to this article.
Personal login Institutional Login Open Athens loginNot a subscriber?
PsychiatryOnline subscription options offer access to the DSM-5-TR® library, books, journals, CME, and patient resources. This all-in-one virtual library provides psychiatrists and mental health professionals with key resources for diagnosis, treatment, research, and professional development.
Need more help? PsychiatryOnline Customer Service may be reached by emailing [email protected] or by calling 800-368-5777 (in the U.S.) or 703-907-7322 (outside the U.S.).