Housing Is the Treatment: DREDF Condemns the Federal “Treatment First” Toolkit

The Disability Rights Education and Defense Fund (DREDF) condemns the Best Practices Toolkit: Addressing Homelessness and Addiction through “Treatment First,”^[1] released this month by the Substance Abuse and Mental Health Services Administration (SAMHSA) with the Department of Housing and Urban Development and the White House Office of National Drug Control Policy.

The toolkit is a blueprint for conditioning shelter on compliance. It comes at a time when the Administration is simultaneously calling for expanded civil commitment, abandoning the integration mandate, and declining to invest in the housing and voluntary services that disabled people actually need.

Taken together, these moves point in one direction: more disabled people institutionalized, more disabled people on the street, and fewer legal tools to do anything about it.

The toolkit does not use the phrase “involuntary treatment.” It doesn’t need to. It builds a system in which disabled people have to earn housing rather than have a right to it, and characterizes the people it fails as noncompliant. The toolkit:

• Cites Executive Order 14321, Ending Crime and Disorder on America’s Streets^[2], as the authority and basis for its approach. That order states that shifting homeless^[3] people into long-term institutional settings through civil commitment will restore public order, and it directs federal officials to seek reversal of judicial precedents and termination of consent decrees that get in the way. The toolkit is the service-delivery arm of that project.
• Describes a SAMHSA grant program under which housing assistance requires participation in treatment.
• Instructs programs to mandate alcohol- and drug-free living environments, and states that when relapse occurs, the individual should be moved to a different level of care before returning to a similar housing environment. In plain terms: if you use, you lose your bed.
• Places non-subsidized, market-rate housing at the far end of a five-phase ladder, reached only after twelve to eighteen months of documented abstinence and full-time employment above minimum wage.
• Defines success as self-sufficiency, meaning housing “not reliant on government assistance,” measured alongside abstinence and employment.

Federal policy is supposed to follow evidence. Here it does the opposite.

Two decades of research, including randomized controlled trials in the United States and Canada, establish that Housing First produces substantially better housing outcomes than treatment-contingent models: faster exits from homelessness, higher retention, fewer returns to the street, and reduced use of emergency departments and hospitals.^[4] This holds for people with psychiatric disabilities, people who use drugs, people with chronic health conditions, and people with intellectual and cognitive disabilities.^[5] The Community Preventive Services Task Force — the independent panel appointed by the Director of the CDC — reviewed this precise comparison and recommended permanent supportive housing with Housing First over Treatment First models that require people to become “housing ready” first. It found that Housing First decreases homelessness, increases housing stability, improves quality of life, reduces hospitalization and emergency department use, and returns economic benefits exceeding the cost of the intervention.^[6] The Department of Veterans Affairs published its own review of the Housing First evidence base.^[7]

On substance use and mental health outcomes, the evidence is inconclusive rather than adverse.^[8] The research does not show the harm the toolkit asserts. The fear that low-barrier housing increases substance use has not been borne out.

The toolkit rests its central claim on a single 2026 study finding that overdose mortality risk was twice as high among veterans in HUD-VASH as among homeless veterans outside the program.^[9] But that study does not support the conclusion the toolkit draws from it. It is observational, not randomized, and propensity matching cannot eliminate the differences between people who enter supportive housing and people who do not.

More importantly, the study’s authors conclude that timely substance use treatment within supported housing could prevent overdose deaths. That is an argument for funding treatment inside housing, not for withholding housing until treatment is complete. The toolkit quotes the study accurately. What it gets wrong is the inference: its own source argues for putting treatment inside housing, and the toolkit converts that into an argument for putting housing behind treatment.

Meanwhile, other VA research finds that continued receipt of housing services is associated with reduced mortality, and that exiting those programs is a period of heightened risk.^[10] The toolkit’s discharge-on-relapse design would manufacture exactly those exits.

Disabled people are, by federal definition, the entirety of the chronically homeless population.^[11] Yet, this federal framework for serving us never once mentions the Americans with Disabilities Act (ADA) ^[12], Section 504 of the Rehabilitation Act (Section 504)^[13], the Fair Housing Act^[14], the Supreme Court’s decision in Olmstead v. Lois Curtis^[15], or the concept of reasonable accommodation. Nor does it acknowledge the profound inaccessibility of the nation’s existing housing stock.^[16] The omission of these foundational disability-rights protections, and its silence on the material barriers to accessible housing, is a statement about who this Administration considers worth serving.

• Integration. Title II of the ADA and Section 504 prohibit unnecessary segregation. Concentrating disabled people into congregate, rule-bound, staff-supervised settings — with market-rate independent housing available only to those who complete the program — is precisely the harm Olmstead addressed. The June 2026 Office of Legal Counsel (OLC) opinion^[17] asserting that these statutes impose no integration mandate does not change the law. Neither the ADA nor Section 504 nor Olmstead has been repealed, and courts are not bound by OLC opinions.
• Reasonable accommodation. Program rules on schedules, chores, meetings, work requirements, and abstinence must be modified when necessary to afford disabled people equal opportunity to use and enjoy a dwelling. A toolkit that mandates uniform structure and routine without a word about modification invites systematic violations.
• Accessibility. Dormitory-style living and repurposed buildings are recommended as low-cost options. Nothing in the toolkit addresses accessible units, accessible bathrooms, or effective communication.
• Due process and liberty. Confinement premised on disability, without individualized findings and adequate process, has been unconstitutional for fifty years.

Disability justice requires us to ask who absorbs harm first. Chronic homelessness is inseparable from disability. This toolkit is, in significant measure, a policy about disabled people. And disabled people experiencing homelessness are not a monolith. We are Black, Indigenous, Latino, Pacific Islander, transgender, and poor. Those intersections, and their bolder inequities, appear in every figure that follows.

Black people are 13.5 percent of the U.S. population and 32.7 percent of people experiencing homelessness.^[18] Native Hawaiian and Pacific Islander people experience homelessness at the highest rate of any racial or ethnic group — 99.9 per 10,000, against a national rate of roughly 22.^[19] Latino and Indigenous people are overrepresented as well.^[20] Transgender people experience homelessness at rates no other group in this analysis approaches: 8 percent experienced homelessness in the previous year, compared with 1 percent of cisgender straight adults, and nearly one in three have experienced homelessness at some point in their lives — roughly eight times the general population, with even higher rates among Indigenous and Black respondents.^[21] More current federal figures do not exist, because the Census Bureau stopped asking about gender identity in 2025.^[22] And roughly one in four working-age disabled adults already lived in poverty, 24.9 percent against 10.1 percent of nondisabled adults, before this Administration began dismantling the programs we rely on.^[23]

These disparities matter because the toolkit does not begin with a neutral system. It begins with institutions that already exclude and burden the people most affected by homelessness. For some of us, the first barrier is simply getting through the door. Emergency shelter is the first rung of the toolkit’s housing ladder: the toolkit directs programs to offer housing ranging from shelter to recovery residence to independent living, identifies “using shelter services” as the crisis-phase housing benchmark, measures progress by movement from the street or shelter into transitional or recovery housing, and lists shelter transport among its encampment-resolution activities.^[24] Yet among transgender respondents to the 2022 U.S. Transgender Survey who have experienced homelessness, 58 percent report having been denied access to emergency shelter because of their gender.^[25] That is a barrier at precisely the point where this framework tells us to begin.

That barrier is becoming more consequential. HUD has proposed rescinding the Equal Access Rule in a rule that would expressly permit single-sex emergency shelters to demand evidence of a person’s sex as a condition of admission.^[26] The toolkit acknowledges none of this. It provides no meaningful mechanism — no accommodation, grievance process, or alternative pathway — for people who are turned away at the door.

And exclusion from shelter is only one way this framework can deepen harm. We are also the people most likely to be swept into coercive systems when those systems expand. Every person in the research that follows has a psychiatric disability; the question is not whether disability is present, but which of us gets committed. Black patients are significantly more likely than white patients to be involuntarily admitted to psychiatric hospitals, more likely to have commitment petitions filed,^[27] more likely to be restrained,^[28] and far more likely to be diagnosed with a psychotic disorder in emergency settings. A policy that widens the door to civil commitment therefore does not widen it evenly. It expands state coercion along lines that are already deeply racialized.

The toolkit’s chapter on drug-endangered children compounds these risks by bringing the child welfare system into homelessness services. Universal screening of children at intake, mandatory reporting training across homeless service providers, and testing children for exposure would operate within a child welfare system that already separates Black and Indigenous families at grossly disproportionate rates.^[29] The predictable consequence is greater fear of seeking help. Families do not walk willingly into a program that may take their children.

These are not isolated implementation concerns. Taken together, they describe who is most likely to be excluded at the front door, subjected to coercion once inside, or deterred from seeking services altogether. The people who bear those risks are the same people already overrepresented among those experiencing homelessness. A framework that treats shelter access, civil commitment, and child welfare surveillance as tools for managing homelessness without accounting for these disparities does not merely fail to protect the most vulnerable. It places the greatest burdens on the people who already carry the most.

1. Withdraw the toolkit and rescind Executive Order 14321.
2. Fund housing.No U.S. housing market allows a person living on SSI — $994 a month in 2026 — to afford a modest apartment without rental assistance.^[30] Until that changes, “self-sufficiency” is not a goal. It is a disqualification.
3. Fund treatment inside housing, voluntarily, on demand, including full access to medications for opioid use disorder. Nothing about Housing First prohibits treatment. It prohibits using treatment as a condition of shelter.
4. Keep disability civil rights in the frame. Any federal homelessness policy must be measured against the ADA, Section 504, the Fair Housing Act, and the integration mandate.
5. Measure what matters. Days housed. Returns to homelessness. Deaths. Not compliance.

Housing is the treatment. Disabled people should not have to earn a place to sleep. The SAMHSA toolkit is guidance, not law. The ADA, Section 504, and the Fair Housing Act are still law, whatever an executive order or OLC opinion asserts. Conditioning housing on compliance is not a best practice. It is unlawful, and we will fight it.

[1] Substance Abuse & Mental Health Servs. Admin., Best Practices Toolkit: Addressing Homelessness and Addiction Through “Treatment First”, Pub. No. PEP26-04-009 (2026).

[2] Exec. Order No. 14,321, Ending Crime and Disorder on America’s Streets, 90 Fed. Reg. 35,817 (July 29, 2025).

[3] This statement uses “homeless” rather than “unhoused,” and uses federal terms of art such as “chronically homeless” throughout. The definitions and datasets it engages — HUD’s Point-in-Time count, the definition of chronic homelessness at 24 C.F.R. § 578.3, and the toolkit itself — are built on that language, and departing from it would obscure what is being measured and disputed. These are federal eligibility categories with legal consequences, not descriptions we would ordinarily choose. Nothing here should be read as a judgment about which term people should use for themselves.

[4] Sam Tsemberis et al., Housing First, Consumer Choice, and Harm Reduction for Homeless Individuals with a Dual Diagnosis, 94 Am. J. Pub. Health 651 (2004); Vicky Stergiopoulos et al., Effect of Scattered-Site Housing Using Rent Supplements and Intensive Case Management on Housing Stability Among Homeless Adults with Mental Illness: A Randomized Trial, 313 JAMA 905 (2015); Tim Aubry et al., A Randomized Controlled Trial in Five Canadian Cities of the Effectiveness of Housing First with Assertive Community Treatment for Persons with Serious Mental Illness and a History of Homelessness, 67 Psychiatric Servs. 275 (2016); Paula Goering et al., National At Home/Chez Soi Final Report (Mental Health Comm’n of Can. 2014).

[5] Tim Aubry et al., Effectiveness of Permanent Supportive Housing and Income Assistance Interventions for Homeless Individuals in High-Income Countries: A Systematic Review, 5 Lancet Pub. Health e342 (2020); Andrew J. Baxter et al., Effects of Housing First Approaches on Health and Well-Being of Adults Who Are Homeless or at Risk of Homelessness: A Systematic Review and Meta-Analysis of Randomised Controlled Trials, 73 J. Epidemiology & Cmty. Health 379 (2019); Nick Kerman et al., Harm Reduction Outcomes and Practices in Housing First: A Mixed-Methods Systematic Review, 228 Drug & Alcohol Dependence 109052 (2021) (collecting studies).

[6] Cmty. Preventive Servs. Task Force, Social Determinants of Health: Permanent Supportive Housing with Housing First (Housing First Programs), Guide to Cmty. Preventive Servs. (June 2019), https://stacks.cdc.gov/view/cdc/168591.

[7] Jack Tsai, The Evidence Behind the Housing First Model, Nat’l Ctr. on Homelessness Among Veterans, U.S. Dep’t of Veterans Affairs (May 2023).

[8] Susan E. Collins, Housing First: Overview of the Evidence Base, Harm Reduction Rsch. & Treatment Lab, Univ. of Wash. (Aug. 30, 2025).

[9] Jack Tsai et al., Changes in All-Cause, Overdose, and Suicide Mortality Risk in the First 2 Years of Supported Housing, United States, 2017–2021, 116 Am. J. Pub. Health 665 (2026).

[10] Ann Elizabeth Montgomery et al., Association Between Receipt of a Continuum of Supportive Housing Services and Mortality Among Veterans with Experience of Housing Instability, 68 Am. J. Preventive Med. 497 (2025). The outcomes measured are suicide and all-cause mortality.

[11] 24 C.F.R. § 578.3 (defining “chronically homeless” to require a homeless individual with a disability).

[12] 42 U.S.C. § 12191 et seq.

[13] 29 U.S.C. § 794.

[14] 42 U.S.C. § 3602 et seq.

[15] Olmstead v. L.C., 527 U.S. 581 (1999).

[16] Luke Bo’sher et al., Accessibility of America’s Housing Stock: Analysis of the 2011 American Housing Survey 1 (U.S. Dep’t of Hous. & Urban Dev., Off. of Pol’y Dev. & Rsch. 2015) (finding that roughly five percent of U.S. housing units are accessible to people with moderate mobility difficulties and less than one percent is wheelchair accessible).

[17] Application of the Rehabilitation Act and Americans with Disabilities Act to State Institutionalization of Patients with Severe Mental Illness or Disabilities, Op. O.L.C. (June 18, 2026) (slip op.), https://www.justice.gov/olc/media/1446701/dl.

[18] U.S. Dep’t of Hous. & Urban Dev., The 2025 Annual Homelessness Assessment Report (AHAR) to Congress, Part 1: Point-in-Time Estimates of Homelessness (2026).

[19] Id.; USAFacts, How Many Homeless People Are in the US? What Does the Data Miss? (July 14, 2026), https://usafacts.org/articles/how-many-homeless-people-are-in-the-us-what-does-the-data-miss/. Rates per 10,000 by race are derived from AHAR and Census Bureau population data.

[20] U.S. Dep’t of Hous. & Urban Dev., supra note 18 (reporting that Black, Latino, American Indian, Alaska Native, Pacific Islander, and Native Hawaiian people continue to be overrepresented among people experiencing homelessness relative to their share of the U.S. population). HUD reports these categories as “Hispanic/Latina/e/o” and “American Indian, Alaska Native, or Indigenous.”

[21] Bianca D.M. Wilson et al., Homelessness Among LGBT Adults in the US (Williams Inst., UCLA Sch. of L. 2020) (nationally representative past-year estimates); JSandy E. James et al., Early Insights: A Report of the 2022 U.S. Transgender Survey (Nat’l Ctr. for Transgender Equal. 2024) (lifetime and subgroup estimates; large non-probability sample). On poverty, see M.V. Lee Badgett et al., LGBT Poverty in the United States (Williams Inst., UCLA Sch. of L. 2019).

[22] Williams Inst., UCLA Sch. of L., Household Trends and Outlook Pulse Survey: Public Comment (Mar. 2025). The Household Pulse Survey was the first Census Bureau survey to ask directly about sexual orientation and gender identity; upon its January 2025 relaunch as the Household Trends and Outlook Pulse Survey, the Census Bureau announced that it would not include gender identity questions in future collections.

[23] Annual Disability Statistics Compendium tbl.6.1 (Inst. on Disability, Univ. of N.H. 2024) (American Community Survey 2022 data). The measure covers people living in the community and therefore excludes institutionalized disabled people.

[24] SAMHSA, Best Practices Toolkit: Addressing Homelessness and Addiction Through “Treatment First,” Pub. No. PEP26-04-009 (Aug. 2026), at 6 (Core Elements, “Phased housing”: programs should offer a range of recovery housing options from shelter to recovery residence to independent living), 10 (ch. 2 phase table, Self-Sufficiency: Housing; crisis-phase benchmark is “using shelter services”), 11 (ch. 2 supplemental measures: participants transitioning from the street or shelter to transitional or recovery housing), 23 (ch. 5: encampment-resolution activities include on-site housing placements, shelter transport, and service linkage).

[25] Advocs. for Trans Equal. & Nat’l All. to End Homelessness, Nationwide Survey Shows Widespread Discrimination Against Gender-Expansive People, Including in Emergency Shelters (May 2026) (previously unreleased data from the 2022 U.S. Transgender Survey, a non-probability sample of more than 92,000 respondents; the figure reflects respondents who have experienced homelessness).

[26] Equal Access to Housing in HUD Programs Revisions, 91 Fed. Reg. 22,779, 22,784 (proposed Apr. 28, 2026) (to be codified in scattered parts of 24 C.F.R.) (Docket No. FR-6518-P-01; RIN 2501-AE12) (proposed 24 C.F.R. § 5.106(c) would require that placement in temporary, emergency shelters and other facilities with shared sleeping quarters or shared bathing facilities be made in accordance with the individual’s sex, and would provide that “[a] facility provider may require reasonable assurances or evidence to establish a person’s sex”). The proposed rule implements Exec. Order No. 14,168 § 4(b), 90 Fed. Reg. 8615 (Jan. 20, 2025), which directs HUD to rescind Equal Access in Accordance with an Individual’s Gender Identity in Community Planning and Development Programs, 81 Fed. Reg. 64,763 (Sept. 21, 2016). On June 17, 2026, DREDF submitted comments opposing rescission.

[27] Timothy Shea et al., Racial and Ethnic Inequities in Inpatient Psychiatric Civil Commitment, 73 Psychiatric Servs. 1322 (2022) (prospective cohort of all patients admitted to an inpatient psychiatric unit, 2012–2018; 28 percent were admitted involuntarily; after adjustment, Black patients remained more likely than white patients to be involuntarily admitted (aOR 1.57, 95% CI 1.26–1.95), as did patients identifying as other race or multiracial (aOR 2.12, 95% CI 1.44–3.11); Black and Asian patients were more likely to have court commitment petitions filed; 25 percent of white patients were diagnosed with a psychotic disorder, compared with 58 percent of Black patients).

[28] Ambrose H. Wong et al., Association of Race/Ethnicity and Other Demographic Characteristics with Use of Physical Restraints in the Emergency Department, 4 JAMA Network Open e2035241 (2021); Kristina Schnitzer et al., Disparities in Care: The Role of Race on the Utilization of Physical Restraints in the Emergency Setting, 27 Academic Emergency Med. 943 (2020).

[29] Child Trends, Use of Multiracial Category Underestimates Disproportionate Representation of Black and Indigenous Children in Foster Care (Nov. 2025) (counting multiracial children who are Black or Indigenous, the share of Black children in foster care is 2.2 times their share of the general child population, and the share of Indigenous children is 2.5 times); Legis. Analyst’s Off., Update: Racial and Ethnic Disproportionalities and Disparities in California’s Child Welfare System (Apr. 3, 2024) (the proportions of Black and Native American youth in California foster care are roughly four times their proportions of the state’s child population; research on California’s 1999 birth cohort found that nearly one in two Black and Native American children experienced some level of child welfare involvement by age 18, compared with approximately 22 percent of white children).

[30] Tech. Assistance Collaborative, Priced Out: The Housing Crisis for People with Disabilities; Soc. Sec. Admin., How Much You Could Get from SSI (2026), https://www.ssa.gov/ssi/amount.