Why They Came for PRAMS

It seems harmless. A survey. Three thousand women a year, in each state, answering questions about their pregnancies. Founded in 1987 to fight infant mortality. Credited with halving the U.S. infant death rate. Bipartisan armor for thirty-eight years. The Trump administration laid off the entire team on April 1, 2025. Congress has voted, twice, to keep it funded. HHS is shutting it down anyway. And PRAMS did not fall alone. The entire division it lived inside was killed the same day. This is the briefing on why.

PRAMS did not fall alone. The whole division died with it.

The Pregnancy Risk Assessment Monitoring System lived inside CDC’s Division of Reproductive Health (DRH). On the same April 2025 day that the PRAMS team got their layoff notices, four other programs in the same division were killed:

• The Women’s Health and Fertility Branch — the structural home for non-pregnancy reproductive health surveillance. Eliminated.^[4]
• The team that maintains the U.S. Medical Eligibility Criteria for Contraceptive Use (MEC) — the clinical guidelines doctors use to decide which birth control is safe for patients with heart disease, kidney disease, sickle cell, lupus, or any of dozens of other complex conditions. Disbanded.^[10]
• The team that collects in vitro fertility (IVF/ART) data — the only national-level fertility-clinic outcomes surveillance. Eliminated.^[10]
• ERASE MM — the federal infrastructure supporting state Maternal Mortality Review Committees. Scaled back at the same time.^[5]
• PRAMS itself.

Together, these five programs constituted the federal infrastructure for surveillance, clinical guidance, and accountability across reproductive health. They were not killed individually. They were killed as a group, on the same day, in the same Reduction in Force action, by the same decision.

And the CDC’s stated reason — for once — was on the record. The CDC spokesperson told reporters in May 2025 that the cuts were made “in accordance with President Trump’s January 20 Executive Orders on ‘Defending Women from Gender Ideology Extremism’ and ‘Ending Radical And Wasteful Government DEI Programs.’”^[10]

The whole division died. PRAMS died loudest.

Inside the division, PRAMS was killed first and most visibly. Not by accident. There are four structural reasons:

One: It is the upstream of everything. Maternal Mortality Review Committees in all 50 states draw on PRAMS data to investigate deaths and make policy recommendations. The 12-month postpartum Medicaid extension that 49 states have now adopted was built on the evidence base PRAMS produced. Every state-level maternal health intervention study from 2010 onward depends on PRAMS as input. Kill PRAMS and you kill the upstream of an entire policy ecosystem — without ever having to argue against the ecosystem itself.^[2]

Two: It is the most efficient producer of DEI-coded data. Race. Ethnicity. Income. Education. Zip code. Insurance type. Rural vs. urban. PRAMS stratifies every question by every demographic variable. By design. That is what makes it useful for equity research. It is also, by the same design, what makes it the most efficient single producer of the kind of stratified-by-identity data the EO on “Radical and Wasteful Government DEI Programs” was written to eliminate.

Three: Pregnancy intention. PRAMS asks every respondent: “Thinking back to just before you got pregnant with your new baby, how did you feel about becoming pregnant?” The decades-long finding from that question — that roughly 40 percent of U.S. pregnancies are reported as unintended^[11] — is the empirical foundation of every argument for contraceptive access, every argument for postpartum support, every refutation of pro-natalist rhetoric that claims women want what is on offer. No other federal data system asks this question at population scale. PRAMS does. That is reason enough.

Four: The calendar made it free. PRAMS’ state implementation grants expired on April 30, 2026. If the team was gone and no one renewed the grants, the program died through inaction. No defunding vote. No floor speech. No appropriations fight. The grant calendar did the political work, so no one had to.

That is the full answer to why PRAMS, why first. Largest downstream footprint, plus most DEI-coded data, plus pregnancy intention, plus a convenient expiration date. The other four programs in the division died at the same time. PRAMS died loudest because PRAMS was the easiest to kill and produced the most damage to the project’s empirical foundations once killed.

The crisis being put on the chopping block has a race. So does the chopping block.

You can read everything above and still ask: but isn’t this just generic anti-DEI policy? Why call it racial animus? Because the documentation, when you line it up, does not describe a colorblind data restructuring. It describes a sustained, multi-front campaign to erase a specific racial category from the federal apparatus for understanding maternal harm. Eight independent actions, same direction, same year.

One. HHS told grant applicants to delete the word “Black.” At the April 17, 2026 House Education and Workforce hearing during Black Maternal Health Week, Rep. Summer Lee confronted HHS Secretary Robert F. Kennedy Jr. with the existence of a list of nearly 200 words and phrases that federal funding programs had been instructed to remove from applications. The word Black was on the list.^[12] The instruction is operational, not rhetorical: a grant application that uses the word will be flagged. The applicant must rewrite the Black maternal mortality crisis out of their own proposal in order to be eligible to study it.

Two. The Momnibus Act dropped “Black” from its title. The Black Maternal Health Momnibus Act, introduced in 2021 with explicit racial framing and bipartisan support, was reintroduced in March 2026 as the Momnibus Act. The word Black appears exactly once in the entire latest text of the package — a legislative shift that advocates described as Black women being “erased from the policy that was written for them” in order to keep the policy alive at all.^[13]

Three. Morehouse School of Medicine lost the grant. Morehouse, the historically Black medical school in Atlanta, had a $2.9 million federal grant (with $1.6 million remaining when it was cut) for a research center dedicated to improving the health of Black pregnant and postpartum women. The grant was terminated in 2025.^[14] The most institutionally credible body in the country at studying Black maternal mortality — defunded. Specifically.

Four. The Office of Health Equity and the Office of Minority Health were eliminated.^[15] These were the two federal offices most directly responsible for producing race-stratified health data and policy. Not reduced. Closed.

Five. NIMHD took the deepest proportional cut at NIH. The National Institute on Minority Health and Health Disparities — the NIH institute whose entire reason for existing is studying racial disparities in health — lost 29.6 percent of its previously active funding, $223.5 million across 77 grants, per Liu et al. in JAMA, May 2025. NIMHD had the highest proportional cut of any NIH institute.^[15] Across NIH overall, JAMA documented 694 grants worth $1.81 billion terminated between February 28 and April 8, 2025; the Congressional Research Service updated the cumulative figure to 1,392 awards by April 2026, including intimate partner violence research — a leading cause of pregnancy-associated death and one that disproportionately affects Black women.^[14][16]

Six. The CDC “Racism and Health” page was removed from CDC.gov. The 2021 declaration by CDC Director Dr. Rochelle Walensky that “racism is a serious public health threat” — and the entire web infrastructure built around it — has been taken down from the CDC’s own website. A preserved version now exists only at an independent archival project, RestoredCDC.org.^[17] The federal government has formally retracted its position that racism is a public health issue.

Seven. Race and ethnicity variables were removed from federal datasets. Per the Center for American Progress and the Center for Reproductive Rights documentation, the administration has “removed variables relating to race and ethnicity” from federal datasets in coordination with the EO framework.^[15] You cannot analyze a disparity that the dataset no longer measures.

Eight. The data was getting worse for Black women in real time — and that is the year the surveillance got killed. The most recent CDC maternal mortality data, for 2023, showed that Black women were the only racial or ethnic group whose maternal mortality rate did not decline. While every other group’s rate fell, the Black maternal mortality rate moved from 49.5 to 50.3 per 100,000 live births — a change CDC noted was not statistically significant, but one that left Black women alone in a deteriorating direction.^[18] In other words: the disparity was widening, the trend was the only one moving the wrong way, and the data system documenting all of this was killed in the same calendar year.

Secretary Kennedy was given the opportunity to deny this framing at the April 17, 2026 House Education and Workforce hearing. He did not deny that the cuts had targeted race-based data. His framing was: “President Trump is trying to end division in this country. That’s what DEI did; it divided people. It polarized people.”^[12] In other words: documenting that Black women die at three times the rate of white women is now classified by the federal government as a divisive act. The data is what divides. Not the deaths.

This is the move. The animus is not displayed as overt language. The animus is displayed as the systematic removal of the federal capacity to see, name, fund, study, or organize against the specific harm Black women are experiencing. The harm continues. The visibility of the harm is what stopped.

How a division with bipartisan armor was removed in fourteen months.

The shutdown of PRAMS is not the result of one decision. It is the cumulative result of a sequence of actions, each one defensible in isolation, all of them together producing the obvious outcome.

Look at this sequence carefully. Congress kept the money. The administration killed the program anyway. This is the same mechanism — exactly the same — as the Title X freeze covered in this week’s signal. No bill. No floor vote. No public debate. Just the executive order, the RIF notice, the unanswered email, the mission redefinition, and the calendar.

The paperwork keeps moving. The team that does the work has not existed for a year.

The clearest single piece of evidence that PRAMS is being killed without being killed is what happened in the Federal Register seven months after the team was laid off.

On November 21, 2025 — 234 days after the entire PRAMS team got their RIF notices — the CDC published a Federal Register notice (FR Doc. 2025-20583, Docket CDC-2025-0750) asking the Office of Management and Budget for a three-year extension of PRAMS data collection authority. The notice opened a 60-day public comment window that closed January 20, 2026. The agency received 419 comments. The Association of Maternal & Child Health Programs (AMCHP) filed formal supportive comments on January 7, 2026, noting that PRAMS “covers approximately 81 percent of U.S. live births. For many states and jurisdictions, PRAMS is the only source of data on critical maternal and infant health indicators.”^[9][21]

A federal agency was asking permission to operate, for three more years, a data collection program no one inside the agency remained to operate. The team that would clean the data, weight the sample, manage state grants, supervise the callback protocol, and release the results had not been on payroll since April. The paperwork moved through the system without interruption. The work the paperwork authorized had stopped seven months earlier.

Then it got more pointed. On March 9, 2026, CDC published the 30-day follow-on notice (FR Doc. 2026-04565, Agency Docket 30Day-26-1273). The category quietly changed. What had been requested as an “Extension” in November was reclassified as a “Revision.” The agency did not narratively explain the reclassification. It did not need to. Under federal information-collection rules, a “Revision” reserves administrative authority to alter the instrument — to drop modules, to remove race and ethnicity questions, to eliminate the pregnancy intention or intimate partner violence items — without triggering a fresh 60-day public comment period. The category change is not cosmetic. It is the procedural lever that allows the surviving rump of CDC, at any later date, to redefine PRAMS into something that asks the questions the executive orders permit and stops asking the questions the executive orders forbid.^[21]

The same 30-day notice reduced the annual data-collection burden by 1,495 hours. 1,395 of those hours were the telephone callback surveys — the methodological mechanism by which PRAMS reaches low-income, less-educated, and non-white respondents who do not complete the mail or web instruments on the first pass. The agency framed the callback survey as a “completed component,” language that misrepresents its function. The callback is not a one-time validation exercise. It is the standing protocol that protects every annual PRAMS sample against differential non-response bias. Removing it, whether by line item or by attrition, produces a dataset that systematically over-represents the women whose outcomes were already best understood — and systematically under-represents the women PRAMS was built to surface.

Kristin Rankin at the University of Illinois Chicago, who has worked with PRAMS data for two decades, submitted formal comments to the docket. Her central finding: 76 percent of PRAMS site administrators reported they did not have the time or resources to navigate the PRAMS weighting process for the 2024 data without CDC support, and 97 percent reported that future data files would require CDC technical assistance for cleaning and weighting.^[22] Translation: the federal agency was asking permission to maintain a national surveillance system that the state implementers could no longer technically operate without federal staff who no longer existed.

This is what the bureaucratic ghost looks like. The paperwork track and the operational track have been deliberately decoupled. The paperwork continues because keeping it in motion satisfies the appearance of program continuity and creates a procedural lever for future instrument alteration. The operational track does not continue because the people who would operate it were eliminated thirteen months ago and have not been rehired. A future federal court, asked whether PRAMS was lawfully dismantled, would find that no formal dismantling occurred. The OMB filings are in order. The program is technically authorized. The team simply does not exist.

That, named cleanly, is the architecture. Action through omission is the method. The Bureaucratic Ghost is its signature.

Project 2025 does not mention PRAMS by name. It does not need to.

The 900-page Heritage Foundation document Mandate for Leadership: The Conservative Promise — the policy blueprint widely understood as Project 2025 — contains a specific architectural proposal for CDC. The proposal does not name PRAMS, and it does not need to. It describes the structural conditions that require PRAMS’ elimination.^[7]

Project 2025 proposes three things for the CDC, each of which independently spells the end of population-level reproductive health surveillance:

• Split the CDC into two agencies. One for data collection. One for policy recommendations. The justification is that “CDC is not equipped to make policy decisions.” The effect is that the data-collection agency loses the institutional clients that justify its survey programs, and the policy agency loses the data it would otherwise use.^[7]
• Move CDC’s data infrastructure to a public-private partnership. The Project 2025 text describes moving “CDC’s data infrastructure and management to a public-private partnership.” The structural effect, stated openly in the document: the federal government loses ownership and control of the data.^[7]
• Eliminate DEI data architectures. Project 2025 calls for the federal government to stop collecting data on gender identity, to remove the terms “sexual orientation,” “gender identity,” “gender equity,” and “gender awareness” from federal rules, and to eliminate DEI programs across agencies.^[8] PRAMS, which produces stratified data by race, ethnicity, income, and pregnancy intention, sits squarely inside the architecture this directive targets.

You cannot read this proposal honestly and conclude that PRAMS survives it. PRAMS exists because the federal government collects, owns, and stratifies maternal-experience data through a public agency. Project 2025 proposes to end all three of those preconditions. The CDC layoffs and the PRAMS shutdown are not departures from the plan. They are the plan, executed.

It is not a death certificate. It is the experience.

To understand why the architecture above specifically requires PRAMS’ removal, you have to understand what PRAMS uniquely produces — and what it produces that no remaining federal system can.

PRAMS asks roughly three thousand women in each participating state, every year, about three hundred questions covering the months before, during, and after each pregnancy. The list of what it captures, in plain English:

• Pregnancy intention. Did you mean to be pregnant? Were you using contraception? Did you want to be pregnant later, or not at all?
• Access to prenatal care. When did you start care? Were you on insurance? On Medicaid? Did you have to travel? Wait?
• Lived experience during pregnancy. Intimate partner violence. Housing instability. Depression. Anxiety. Substance use — and why.
• The structural variables. Race. Ethnicity. Income. Education. Zip code. Insurance type. State. Rural vs urban.
• Postpartum. Postpartum depression. Contraception access after birth. Breastfeeding. Whether Medicaid expired at 60 days. Whether postpartum care was received.
• Birth outcomes linked to all of the above — preterm birth, low birthweight, NICU admission, Sudden Infant Death Syndrome risk factors, infant mortality.

Each question, on its own, is mundane — the kind a competent OB asks any patient on intake. The product of asking all of them, across hundreds of thousands of women, year after year, in every state — is not mundane. The product is a full-resolution map of what it costs to be pregnant in America, sliced any way you ask it to be sliced.

Death certificates count corpses. Hospital discharge records count what was billed. Medicaid claims count what was reimbursed. PRAMS counts experience. That is the entire difference. And it is the entire reason we are having this conversation.

What you can build only with this data — and what you cannot.

Public health data has political weight in proportion to what it allows you to argue. PRAMS, structurally, makes three categories of argument possible that no other federal data source supports at this scale. These are exactly the three arguments the current administration’s policy framework cannot survive.

Argument One: Black maternal mortality is not biological. It is structural. PRAMS is the empirical foundation for the claim that Black women die in childbirth at three to four times the rate of white women, controlling for income, education, and clinical risk. Without PRAMS’ race-stratified, experience-stratified data, the disparity collapses into anecdote. With PRAMS, it is a measurable, replicable, defensible fact — and the explanations that point to structural racism in the system itself become the only ones the data supports. Take the data away, and every maternal mortality review committee, every implicit bias intervention study, every Medicaid extension argument, every state-level disparity report loses its load-bearing wall. The administration’s stated position is that “race-based” data analysis is illegitimate DEI activity. PRAMS is the most efficient race-based data producer in maternal health.

Argument Two: Abortion bans kill the women who voted for them. PRAMS is how you compare states by policy regime. It is the input behind the now-well-known finding that maternal mortality rose 56 percent in Texas in the year after SB 8 — and that the rise was concentrated in white women, whose mortality rose 95 percent. It is the input behind the data showing sepsis rose in restricted states, that the OB workforce collapsed in Idaho, that hospital transfer-out rates climbed across abortion-ban geographies. None of that analysis is possible without state-level, longitudinal pregnancy-experience data. With PRAMS gone, the question “did the abortion bans make women safer or less safe?” becomes empirically unanswerable. Which is, in policy terms, the same as answered in their favor.

Argument Three: Pro-natalist rhetoric is empirically false. The administration is publicly committed to a pro-natalist worldview — declining birth rates as crisis, marriage and motherhood as the solution, family-formation incentives as policy. PRAMS data, year after year, has shown the opposite of what this rhetoric requires. Roughly forty percent of U.S. pregnancies are reported as unintended. Significant fractions of pregnant women report intimate partner violence, lack of contraceptive access, depression, financial insecurity, and absence of postpartum support. The data does not show women who want what the project claims they want. PRAMS is the falsification mechanism for the pro-natalist story the political project tells about itself.

What they will say the reason is. And what is wrong with each.

No defunding of a politically inconvenient data system is ever framed as a defunding of a politically inconvenient data system. The cover stories are predictable. They are also, individually, false.

Cover story one: Core mission. RFK Jr.’s public framing — that the CDC should focus on infectious diseases, on “epidemics and outbreaks.” This is the most operationally honest of the cover stories, because it tells you the architecture without telling you the architecture’s purpose. The 1946 founding mission of the CDC was infectious disease, yes. The 1987 expansion to chronic disease, injury, and maternal health surveillance was a deliberate policy decision made by both parties over four decades. The redefinition is not a “return” to core mission. It is a contraction of mission, and the contraction’s casualties are surveillance systems that produce politically inconvenient data. Maternal health is the first one out the door. It is not an accident.

Cover story two: Cost. PRAMS’ federal cost runs in the low tens of millions of dollars per year. NIH’s women’s health budget for FY 2026 is over forty billion dollars. PRAMS is roughly one one-thousandth of that. No serious budget conversation requires defunding PRAMS. Both the House and Senate Appropriations Committees reinstated the funding.^[6] Cost is not the issue.

Cover story three: Federalism. “Let states handle their own surveillance.” This is operationally honest in a different way — it tells you the outcome without telling you the outcome is the goal. States will not, in aggregate, replace PRAMS. The states with high maternal mortality have neither the will nor the budget. The states with low mortality will continue to track their own outcomes but cannot compare to anyone else. The Commonwealth Fund’s January 2026 analysis put it plainly: “If this becomes a broader trend, cross-state comparisons and national-level analysis could be seriously compromised.”^[2] National comparability is the threat. Federalism is how you eliminate it while sounding reasonable.

Cover story four: Duplication. The claim that “the data is already collected by other systems.” It is not. PRAMS is the only national pregnancy-experience surveillance system. State systems exist in roughly six states and capture a fraction of what PRAMS captures. Hospital discharge records and Medicaid claims are billing data — they capture what was billed, not what was experienced. There is no redundancy. Nothing else collects what PRAMS collected.

Cover story five: Efficiency. “CDC needs to streamline.” The CDC lost roughly 25 percent of its workforce in 2025. The Division of Reproductive Health was reduced to a skeleton; the Women’s Health and Fertility Branch was eliminated; ERASE MM scaled back at the same time. This is not streamlining. This is the surgical removal of a specific organ — the one that produces population-level data on women.

What is happening, in one sentence.

The political project currently running the federal government has staked itself, publicly, to three claims that PRAMS data either falsifies or makes vulnerable to falsification:

• That racial disparities in maternal mortality are exaggerated or imagined. PRAMS made the disparities measurable.
• That abortion restrictions protect women. PRAMS made the policy effects visible.
• That women want the pro-natalist project on offer. PRAMS showed what women actually report wanting, and the report is incompatible.

If the data continues to be produced, the claims do not survive. If the data stops being produced, the claims become unfalsifiable. Unfalsifiable claims are the only kind that stay politically useful indefinitely.

This is not a deduction about motive. Motive is unknowable. It is a deduction about function: the function of killing PRAMS is to make the three claims unfalsifiable, regardless of what anyone in the room intends. The action is the policy, regardless of how it is described.

Once you see this, the question “why PRAMS, why now” stops being mysterious. PRAMS is being killed because PRAMS works. The same logic that kept it alive for thirty-eight years under both parties — that it produces good data cheaply, and that the data drove the largest reduction in U.S. infant mortality in modern history — is what makes it intolerable to a project whose claims do not survive good data.

The field is on the record. The administration is not.

Health policy researchers who built their work on PRAMS data are not silent. Katy Kozhimannil, a University of Minnesota maternal health researcher whose work has shaped Medicaid postpartum extension policy in multiple states, told STAT News in April 2025: “I am truly concerned about the potential impacts. With the loss of staff that support PRAMS, we risk a longer delay or loss of this important data resource.”^[3]

The Commonwealth Fund’s January 2026 analysis was sharper:

“If PRAMS were to be discontinued, federal and state public health officials, policymakers, researchers, and advocates will lose a representative, multistate data source that has long been essential for efforts to improve maternal and infant health. Without this information, the U.S. will be less able to direct public resources toward the most effective interventions and hold public officials accountable for meeting maternal and infant health outcome goals.” Commonwealth Fund, “What Is PRAMS, and Why Is It at Risk?” — January 2026^[2]

The Association of Maternal & Child Health Programs (AMCHP), the professional organization representing state maternal and child health officials, submitted formal comments to the CDC in January 2026 urging the agency to preserve PRAMS.^[9]

Both the House and Senate Appropriations Committees reinstated PRAMS funding in the FY 2026 appropriations bills, with bipartisan concern noted explicitly about the pause in data collection and the staffing reductions. P.L. 119-75 was signed into law on February 3, 2026.^[6]

Against all of that, HHS has said nothing publicly defending the specific decision to end PRAMS. The mission redefinition is the defense. That is the whole story.

The work shifts. The data does not return.

It took thirty-eight years to build the evidence base PRAMS produced. The CDC’s official position is that infant mortality dropped from 10 per 1,000 to 5.5 per 1,000 partly because of PRAMS-informed policy.^[1] All of that infrastructure can be dismantled in a single budget cycle. The asymmetry is the entire point.

Restoring PRAMS, if a future administration chooses to, will require rebuilding the workforce, restoring state participation, re-validating the survey instrument, and re-establishing data linkages. The most optimistic estimate is three to five years before national pregnancy-experience surveillance is back online at scale. The pessimistic estimate is that state participation, once broken, will remain bifurcated by political alignment.

What the field can do in the meantime is enumerated. None of it replaces PRAMS:

• Document at the chart level what PRAMS used to document at the population level. Diagnoses, severity, interventions, timing, barriers, outcomes, through one year postpartum. Your records may be the only remaining data source.
• Support state-level surveillance where it exists. California, Massachusetts, New York, and Washington have some independent infrastructure. Encourage your state to preserve or build its own.
• Preserve historical PRAMS data. Download and archive what is publicly available. CDC’s data hosting cannot be assumed to remain stable.
• Use what was already published. Decades of PRAMS-based research is in the literature. It is the last clean baseline against which future patterns will be compared. Cite it. Teach it. Build curriculum around it.
• Call your representatives. Congressional appropriations have reinstated the funding. The mechanism failing is the executive branch ignoring the appropriation. That is a different fight, with a different lever — and it is one Congress has, if it wants to use it.