HHS Built Its Case Against Housing First on a Study Whose Authors Say the Opposite

Carla Vanegas stood at a podium inside the San Diego Rescue Mission on August 12 and told a room full of cameras and cabinet officials that nineteen years ago she was nine months pregnant and living in a drug house. Robert F. Kennedy Jr. stood a few feet away. He was there to announce that the federal government would no longer fund the housing model that, in a different decade, might have been the thing that got her off the street before she got clean — not after.

That’s the trapdoor at the center of the toolkit HHS, the Department of Housing and Urban Development, and the White House drug policy office rolled out this week: it uses stories like Vanegas’s to sell a policy that would have made her story less likely to happen. Read the document’s own citation and you’ll find a study whose authors argue for putting treatment inside housing — not for putting housing behind treatment.

The federal government just used a study to justify killing the very thing that study’s authors said would save lives.

The toolkit is called Best Practices Toolkit: Addressing Homelessness and Addiction Through Treatment First, and it is the clearest signal yet of where national homelessness and addiction policy is headed. It tells states and cities to stop building “Housing First” programs — the two-decade-old model that gets someone off the street and into an apartment without requiring sobriety, then wraps services around them once they’re stable — and start funding programs that require treatment participation before or alongside housing access, with sobriety maintenance as a condition of staying housed. HHS Secretary Kennedy called it accountability. “Treatment must come first,” he said at the announcement. “Recovery begins when we connect people with effective care, not leave them trapped in addiction and homelessness.”

If you’re reading this in early recovery, or you’ve cycled through a program that kicked you out over a single positive test, you already know what “accountability-driven” means when it’s written into a federal funding rule. It means the door locks behind you if you use again. Keep that in mind, because it’s the exact mechanism this toolkit writes into policy — and it’s worth understanding exactly what evidence the government is standing on when it does that, because the evidence doesn’t say what they’re using it to say.

The Study Says Fund Treatment Inside Housing. The Toolkit Reads It Backward.

The centerpiece of the administration’s case is a study published in the American Journal of Public Health in May, examining nearly 61,000 veterans who entered the HUD–VA Supportive Housing program, the federal government’s flagship Housing First initiative for homeless veterans, between 2017 and 2021. Researchers then built a comparison group: for every veteran in supported housing, they paired someone outside the program who looked alike on paper — same age range, same health history, same regional housing market — a technique statisticians call “matching,” since nobody ran a true randomized experiment assigning people to homelessness or housing. What they found: veterans in supported housing died of overdose at roughly twice the rate of the comparison group, consistently, for at least two years after moving in.

That’s a real and troubling number inside an opioid overdose crisis that is still killing tens of thousands of Americans a year, and it deserves a real answer. But the researchers’ own answer is not “stop housing people before they’re sober.” Their conclusion — the one the toolkit’s authors had in front of them when they wrote the document — is that supported housing needs faster, more consistent delivery of methadone and buprenorphine to the people living in it, not a gate that keeps people out until they’ve already stopped using on their own. That distinction is the whole argument. As the Disability Rights Education & Defense Fund put it in a statement condemning the toolkit two days after its release: “Housing First produces substantially better housing outcomes than treatment-contingent models: faster exits from homelessness, higher retention.” DREDF attorney Michelle Uzeta’s statement goes further, noting the toolkit never once cites the Americans with Disabilities Act, Section 504, the Fair Housing Act, or the Supreme Court’s Olmstead decision — the body of civil rights law that exists specifically to stop the government from conditioning basic services on someone’s disability status, and addiction is a disability under federal law.

That’s a real and troubling number inside an opioid overdose crisis that is still killing tens of thousands of Americans a year, and it deserves a real answer.

Here is the layman version of what actually kills someone in that gap: your body forgets how much fentanyl or heroin it can handle faster than your mind forgets the craving. Someone who’s forced into abstinence, whether by a program’s rules or a jail cell, loses tolerance within days. If they return to use — and most people do, more than once, on the way to lasting recovery — the same amount that used to get them high can now stop their breathing. That’s not a moral failure. It’s pharmacology. It’s also precisely why decades of NIH-funded research has found that people treated with methadone or buprenorphine after a nonfatal overdose cut their risk of dying from the next one by more than a third to nearly 60%. A toolkit built to reduce overdose deaths that barely mentions the two medications proven to do exactly that isn’t a technical oversight. It’s a choice about which tool gets used.

Treatment-First Isn’t New. It’s the Idea Housing First Was Invented to Bury.

None of this is new. Before 1992, the dominant model for homeless people with mental illness or substance use disorders was exactly what this toolkit revives: prove you’re “housing ready” — stay clean, follow the rules — and you might earn a bed. A psychologist named Sam Tsemberis, working the streets of New York, watched people he called his clients die waiting to qualify. In 1992 he started Pathways to Housing and flipped the sequence: apartment first, services wrapped around it after, no sobriety test at the door. Within five years, 88% of his participants were still housed, against 47% in the city’s old treatment-first system — a gap so wide that George W. Bush’s administration adopted Housing First as the federal standard in 2004, where it stayed for twenty years.

The reason that history matters this week is not nostalgia. It’s that the toolkit doesn’t call itself a return to the pre-1992 model, and most of the officials championing it may not even know that’s what it is. But the mechanism is identical: housing as the reward for sobriety, instead of housing as the floor sobriety gets built on. The country already ran this experiment once, at scale, and got worse retention and more people cycling back onto the street. If you’ve spent any time inside the addiction treatment system, you already know how this rhymes — it’s the same logic that kept methadone clinics zoned to the edge of town for decades because neighbors didn’t want “those people” nearby, the same logic that blocked federal funding for syringe exchanges for eleven years starting in 1988 while HIV spread through shared needles, on the theory that making drug use safer would be read as endorsing it. Each time, the policy optimized for how the intervention looked rather than what it did. Each time, people who could have been reached were not.

The Fine Print Is an Eviction Clause

The toolkit doesn’t just discourage Housing First. It builds an alternative structure where housing assistance requires ongoing participation in treatment, mandates alcohol- and drug-free living environments, and results in loss of housing if someone returns to use. That last part is the one to sit with. A “relapse” clause in a housing contract does not mean a person gets a warning and a referral. It means the thing keeping them off the street becomes the thing they lose at the exact moment they’re most at risk — freshly destabilized, freshly ashamed, and, per the tolerance math above, freshly vulnerable to a dose that used to be manageable.

This is where the toolkit’s own evidence works against it a second time. DREDF’s statement notes that Black people make up 32.7% of the U.S. homeless population against 13.5% of the general population, and that transgender adults experience homelessness at eight times the rate of cisgender adults — populations already locked out of emergency shelters at higher rates for reasons that have nothing to do with sobriety. A federal SSI check runs $994 a month in 2026, nowhere close to market rent in most cities the toolkit is aimed at. Layering a treatment-compliance requirement on top of that arithmetic doesn’t produce more people in recovery. It produces more people who fail a program’s terms and go back outside — where, per the very study the government is citing, the mortality risk is worse, not better.

A federal SSI check runs $994 a month in 2026, nowhere close to market rent in most cities the toolkit is aimed at.

Even the toolkit’s own supporters are quietly flagging the capacity problem underneath the philosophy. At the San Diego announcement, Joanne Standlee of Housing 4 the Homeless told KPBS that “thirty days does not heal anyone. Twelve months, eighteen months residential — that’s what people really need,” and that San Diego, like most of the country, doesn’t have nearly enough long-term residential beds to make “treatment first” anything other than a waiting list with a moral judgment attached to it.

What’s Real Tonight, and What Isn’t Yet Decided

Here’s what didn’t change this week: methadone and buprenorphine are still federally legal, still covered by Medicaid in every state, and — as STAT reported just two days after the toolkit dropped — opioid treatment programs nationwide are actually loosening up, not tightening. Take-home methadone flexibilities that started as pandemic emergency measures are now permanent, more than 70% of clinics have adopted at least half of the federal government’s own 2024 reforms, and early data shows patient retention up 17% where clinics made the changes. That reform came from the same federal government issuing this week’s housing toolkit — proof that “accountability” and “access” aren’t opposites, even inside one administration, even in the same month. If you’re on methadone or buprenorphine right now, or thinking about starting, that door is open wider than it was two years ago, regardless of what happens to the housing debate.

What is still being decided is whether “treatment first” becomes the operating principle for every HUD grant competition and every state homelessness plan going forward, or whether the pushback from disability law groups, homelessness researchers, and the study’s own authors slows it down before shelters start writing sobriety clauses into leases. Congress hasn’t voted on this. It’s a toolkit, not a statute — guidance that shapes how HUD scores grant applications and which model cities feel pressure to adopt, not a law that rewrites what any single shelter has to do tomorrow.

Carla Vanegas is, by every account, a genuine success story — nineteen years clean, running outreach for the same mission that took her in. Nobody should begrudge her that, or the administration for putting her on a stage. But the toolkit built around her story doesn’t say whether the outreach that found her in a drug house nineteen years ago demanded she get clean before anyone would help her, or simply showed up and stayed. What the government’s own study says is that the second kind of door — the one that opens before someone is ready, not after — is the one more likely to still have someone standing in it two years later. That’s not a footnote. That’s the finding they built the toolkit on and then wrote around.