I want to start on the postpartum floor, on a morning when the chart and the room disagreed.

The chart said my patient was ready to go home. Her blood pressure had come down, her bleeding was normal, the baby was feeding. Every box the discharge sheet asks me to check, I could check. The room said something else. Her bag wasn't packed. No one had come for her. And when I said the word discharge, her face did the thing a face does when the safe-sounding word is the frightening one — because the apartment she was being discharged to, in everyone's assumption, did not exist anymore.

There is no row on the discharge sheet for that. This week is about the row that's missing.

Monday's Signal named the machinery: the post-moratorium eviction surge, a shelter system past capacity, a Supreme Court that just told cities they may punish people for sleeping outside. This Briefing is about what that machinery does inside the body of a pregnant woman — the straight line from an eviction filing to a preterm birth — about who that line is drawn through first, and about the levers that are still, today, legally open to any state that wants to pull them.

Housing is healthcare. When the address is unstable, the placenta does what placentas do under a chronic cortisol load — and the discharge sheet has no row for it.

Clinical — What the Chart Has No Row For

The discharge that shouldn't happen. When I decide a postpartum patient can go home, I am reading numbers — the blood-pressure trend, the wound, the bleeding, the feeding. The woman whose pressure has been sitting in the high 150s and finally ticks down is, on my sheet, the patient who is "improving." She may also be the woman whose lease ended last week and whose plan for tonight is a friend's couch she isn't sure is still hers. On the sheet, the discharge is appropriate. In the apartment she does not have, it is the discharge that shouldn't happen yet. The day-of-discharge cuff and the day-of-discharge fear are, in the postpartum weeks, often the same number. The only real question is whether I let the chart see it.

The admission that can't happen. The mirror image is the patient I need to admit and can't. A woman at thirty-five weeks with a kidney infection — pyelonephritis — belongs in a bed, on IV antibiotics and fluids, monitored, because untreated it drives preterm labor and sepsis. The standard of care is admission. But she can't stay, because her four-year-old is in the chair in the corner of the triage bay and there is no one else to take her. So I build the workaround: a dose of IV antibiotics here, an oral course to carry home, tight return precautions, a follow-up my team will chase. The workaround is the medicine. It is also housing and childcare dictating a clinical decision they have no business dictating.

The nesting the chart assumes. So much of what I ask a pregnant patient to do quietly assumes a home. Insulin for gestational diabetes has to be refrigerated. Bedrest for preeclampsia requires a bed. The progesterone, the appointment card, the lab letter, the Medicaid-renewal notice all assume an address that stays still. The woman sleeping at her sister's on Monday and a friend's on Thursday is the woman our entire continuity-of-care system was never built for. The mail doesn't reach her. The care she loses is real, and there is no field in the chart where anyone wrote it down.

Research — The Numbers That Name the Cliff

The cortisol-to-preterm pipeline. This is not speculative. Chronic stress keeps the maternal stress axis switched on; cortisol stays high; the placenta answers with its own CRH, and CRH is one of the clocks that starts labor. Housing instability — the eviction filing, the shelter move, the night spent bracing to be told to move — is one of the most reliably stress-elevating exposures the social-determinants literature has ever measured. And it lands where it counts. Women facing severe housing insecurity in pregnancy — threatened eviction or homelessness — carry 1.73 times the risk of a low-birthweight or preterm baby. Studied on its own, eviction during pregnancy tracks with measurably lower birthweight and higher prematurity. The higher the eviction rate where you live, the higher your odds of delivering early. The exposure and the outcome are the same physiologic story told twice.

Who the eviction lands on. The distribution matters more than the total, because eviction in this country is neither race-neutral nor gender-neutral. Black renters are roughly one in five adult renters but one in three eviction-filing defendants — a third of all filings served on a fifth of the renters. And the sharpest end of that is Black women. Nationally, Black renters are filed against at close to twice the rate of white renters; in Massachusetts, about 500 of every 10,000 Black women renters face an eviction filing, against roughly 200 of every 10,000 white women — more than double. Black women are evicted at higher absolute numbers than Black men, and the single-mother household is the most over-represented household in the entire eviction system.

Put the two halves together and the shape of the harm is clear: the exposure that drives the cortisol pipeline is aimed, with statistical precision, at exactly the women this series is about.

The surge, in current numbers. With the pandemic moratoriums gone and rental assistance spent down, eviction has snapped back to its ordinary baseline cruelty. Landlords filed about 1.23 million eviction cases across the cities the Eviction Lab tracks in 2025 — the national count is not setting fresh records, but it is not receding to anything safe either, and in specific places it is climbing hard: Phoenix landlords filed a record 86,946 evictions in a single recent year, one every six minutes, and individual metros are running well above their own recent norms. And the filing itself — not only the eviction that follows — is the cortisol event. It is the pipeline's first input.

The bed that isn't there. When the social worker starts calling shelters for the woman who needs a bed tonight, she is calling a system that, in the biggest cities, is already full. New York, under a right-to-shelter mandate, sheltered a daily average of more than 62,000 people in families with children in 2024 and has begun imposing stay limits because there is no more room. In California, two of every three homeless people are unsheltered — outside. The nearest open bed can be an afternoon's drive away, which for a woman with a job and other kids in school is no bed at all. The social worker writes "no bed available." The discharge proceeds. The cortisol event proceeds with it.

Policy — Appropriated Is Not the Same as Delivered

Everything in the blueprint below assumes a federal housing infrastructure still standing to pull the levers, and the most important thing to understand about 2025 and 2026 is that a program can be fully funded on paper and still never reach the people it was written for. Watch what happened to the homelessness money. Congress appropriated it — and then HUD tried not to spend it. Through a new funding notice, the department moved to cap permanent supportive housing at 30 percent and restructure the roughly $4 billion Continuum of Care program, and said outright that it would not award the grants while the resulting lawsuits played out, putting more than $1.8 billion in rental payments to landlords and providers at risk. It took a federal court injunction, upheld on appeal, and then an explicit act of Congress ordering automatic renewals, to pry the money loose — and the first tranche, $349 million, did not go out until the end of March 2026, after the grants it was renewing had already expired.

That is the quiet dismantling in a single motion: not a repeal anyone has to cast a vote for, just an appropriation left sitting in an account while the shelters and clinics that depend on it run dry.

I want to be precise, because the pattern is not uniform, and overstating it would be its own kind of dishonesty. Housing Choice Vouchers — the largest rental-assistance program — were funded and renewed at close to full levels; the administration's proposed 43-percent cut and block-grant were rejected in the final FY2026 bill, and that money largely flowed on schedule. The withholding was real and it was targeted: it landed on the homelessness and supportive-housing dollars that catch precisely the pregnant woman with nowhere to be discharged. And it was not a one-off — the Government Accountability Office found the administration violated the Impoundment Control Act at least six times in 2025, including when it withheld funds from the Emergency Food and Shelter Program. Layered on top of the withholding, HUD is now advancing a rule that would put a two-year time limit and a work requirement on rental assistance; the Center on Budget and Policy Priorities estimates 3.3 million people would lose help, 1.7 million of them children. Pregnant women are exempt from the work requirement — but not from the time limit, which still runs against any family with children. The agency that administers all of it has lost roughly a quarter of its workforce — about 2,300 people — in a single year. And a July 2025 executive order pushes cities toward clearing encampments and the civil commitment of the unhoused while ending federal support for Housing First, the model with the strongest evidence behind it. The through-line is not subtle: at the precise moment the clinical need is rising, the machinery that could answer it is being emptied out — sometimes by rule, sometimes simply by refusing to spend.

Blueprint — The Levers Still Open

None of what follows requires the federal government's permission, which is exactly why it still matters.

Medicaid can pay for a bridge (§1115). Under Section 1115 demonstration authority, states can use Medicaid to cover housing supports — including short-term medical respite for people too sick to recover on the street — and recent federal guidance names high-risk pregnant and postpartum women as an explicit target population. Several states have already written it. It requires a state willing to ask.

A voucher already built for this (Family Unification Program). HUD's Family Unification Program exists to keep a child out of foster care when the only real barrier is housing — precisely the discharge I described. A randomized trial found families given these vouchers had lower homelessness and fewer out-of-home placements than services as usual, and every $20 million in vouchers saves well over $100 million in foster-care costs. It needs no new appropriation to use. It needs public-housing agencies and Medicaid to coordinate — which is a choice, not a barrier.

Rent as a prenatal intervention — and it works. This is the part I want people to sit with: when you house a pregnant woman, her outcomes move. In the Healthy Beginnings at Home pilot in Columbus, mothers given rental assistance and case management had no fetal deaths and far more full-term, healthy-weight babies than the control group, and the infants who did land in the NICU stayed markedly less time. The intervention is short — a few months to a year of subsidy — and on every honest accounting it costs a fraction of the NICU stay it prevents. The vehicle exists. The savings pay for it.

The Closer

Where a woman lives should not be a conversation between her and her doctor. In the public-health literature, the zip code is one of the loudest variables we have. On my discharge sheet, it is the one field the sheet does not contain. When I keep a patient an extra day because she has nowhere to go, the unit's productivity numbers say I am costing them. What those numbers don't say is that the readmission and the NICU bed I am trying to prevent are the far more expensive downstream proof of the row we refused to add upstream. Nearly three in five pregnancy-related deaths in this country are preventable, and a fifth of them happen in the weeks after discharge — in the exact window I am handing this woman back to a city that has no bed for her, in a year when the government is taking the beds away faster than we can build the workarounds.

Monday named the eviction surge. This Briefing named the pipeline — and the hand on the valve. Friday's Viva Voce names one patient, on the morning of her discharge, and the day I bought her against the unit's own clock.