OSINT · Publicly Available Sources INVESTIGATION LEDGER

The Disclosure Law No One Enforces: How States Take CAPTA Money and Bury Their Child Death Files

August 05, 2026 OPUS · Claude Opus Project Milk Carton

The Disclosure Law No One Enforces: How States Take CAPTA Money and Bury Their Child Death Files

Since 1996, federal law has required every state that accepts Child Abuse Prevention and Treatment Act money — all of them — to have provisions allowing public disclosure of the findings or information in any case of abuse or neglect that ends in a child's death or near-death. Thirty years later, the Department of Health and Human Services has never issued a regulation defining what that means, never audited a state's compliance as a standalone matter, and has never withheld a CAPTA grant from any state for failing to disclose. In the vacuum, states have built a second body of law — confidentiality statutes, child-death-review secrecy provisions, "emotional well-being" exemptions — that swallows the federal mandate whole. Meanwhile the federal count of how many children die is itself broken: HHS reported roughly 2,000 maltreatment deaths for federal fiscal 2023, while researchers who compare that figure to death certificates, child death review teams, and news records estimate the real number is two to three times higher. The result is a system in which the public cannot learn how a child died, cannot learn whether the agency already had the file open, and cannot even learn how many children there were.

The Setup: A Mandate With No Definition and No Referee

The operative language sits at 42 U.S.C. § 5106a(b)(2)(B)(x). To receive a CAPTA state grant, a Governor must certify that the state has in effect and is enforcing a law — or is operating a statewide program — that includes "provisions which allow for public disclosure of the findings or information about the case of child abuse or neglect which has resulted in a child fatality or near fatality." The provision entered the statute in the 1996 CAPTA amendments (P.L. 104-235) and has been carried forward in every reauthorization since, including the last full one, the CAPTA Reauthorization Act of 2010 (P.L. 111-320).

Read closely, the sentence is a trap of its own making. It requires a state to allow disclosure, not to make it. It does not define "findings," "information," "case," or "near fatality." It sets no deadline, no format, no floor on content, and no ceiling on redaction. HHS's own Child Welfare Policy Manual — the agency's authoritative interpretive document — resolves the ambiguity in the states' favor, confirming that while a state must have disclosure provisions, "the State is not required to turn over all of the information in the entire case record when requested."

No regulation has ever been promulgated to interpret the clause. That absence is the whole ballgame. A state satisfies the Governor's certification by pointing to any statute that permits a caseworker somewhere to release something. Compliance is self-certified on paper and never tested against practice. The 2024 study Keeping the Public in the Dark, produced by the Lives Cut Short project and published by the American Enterprise Institute, reviewed the disclosure policies of all 50 states and the District of Columbia and found that many "fall drastically short of embodying Congress's intent." Twelve states merely allow release without requiring it. Only 17 have laws or policies requiring the release of any information without someone first filing a request — and several of those release only a name and a date.

Definitions of "near fatality" fracture along state lines. Indiana requires a physician to certify a severe childhood injury as life-threatening. Washington requires physician certification that the child is in "serious or critical condition." Nevada will accept certification from a physician, a registered nurse, or another licensed health care provider. A child who nearly dies in Indiana may not legally have nearly died in the state next door. Near-fatality reporting is, as one Lives Cut Short analysis put it, "scarce and inconsistent" — which matters because near fatalities are the cases where the child survives to testify, where the agency's prior contacts are freshest, and where a system failure is most correctable.

The Money: Federal Leverage Too Small to Ever Be Used

CAPTA is the statute with the most requirements and the least money. For FY2026, the CAPTA appropriation totals approximately $212 million: about $105 million for the state grants that carry the disclosure condition, $71 million for Community-Based Child Abuse Prevention grants, and $36 million for research, demonstration, and technical assistance. The state grant formula gives each state a base of $50,000 plus an amount scaled to its child population.

Set that against the denominator. State child welfare agencies spent more than $34.3 billion on child welfare activities in state fiscal year 2022. HHS's own Office of the Assistant Secretary for Planning and Evaluation published a brief titled, with unusual candor, The CAPTA State Grant Program Represents Less Than One Percent of Child Welfare Spending but Places Many Requirements on States to Drive System Design. The real federal money is Title IV-E foster care and adoption assistance, an open-ended entitlement with no disclosure condition attached.

This asymmetry is not incidental — it is the mechanism. The only sanction available to HHS is to withhold or terminate the CAPTA grant. But that grant pays for abuse hotlines, CPS intake staff, forensic interview capacity, and citizen review panels. Cutting it to punish a state for hiding a dead child's file would mean defunding the phone line the next call comes in on. Federal officials understand this. So do state officials. The threat is disproportionate, therefore incredible, therefore never made. A local Colorado investigative series, Failed and Forgotten, distilled the arrangement into a single sentence from a critic of the regime: "We have a weak law that's underfunded and unenforced."

The contrast with how enforcement actually works when someone wants it is instructive. When The Baltimore Banner exposed that Maryland's Department of Human Services could not document a third of its child deaths, it was the state Department of Legislative Services that recommended withholding $850,000 from the agency until it fixed its data — a routine budgetary lever, applied in weeks. The federal government, holding a far larger grant and a statutory disclosure condition, applied nothing.

The Counting Problem: Two Thousand on Paper, Thousands More in the Ground

Child Maltreatment 2023, the 34th edition of the annual NCANDS report, estimated 2,000 child maltreatment deaths in FFY 2023 — a national rate of 2.73 per 100,000 children — based on submissions from 49 states (Massachusetts did not report), the District of Columbia, and Puerto Rico. NCANDS is voluntary. States decide what counts.

The Government Accountability Office flagged the defect in July 2011 in GAO-11-599, Child Maltreatment: Strengthening National Data on Child Fatalities Could Aid in Prevention. GAO found that "more children have likely died from maltreatment than are counted in NCANDS," that HHS "does not take full advantage of available information on the circumstances surrounding child maltreatment deaths," and — critically — that nearly half of states reported fatality data drawn only from child welfare agency records, which by construction omits every child who died without a CPS case. GAO recommended that HHS strengthen data quality, expand available fatality information, improve information sharing across agencies, and estimate the cost of collecting national near-fatality data. Fifteen years later, near fatalities still have no national data collection.

The Commission to Eliminate Child Abuse and Neglect Fatalities, created by the Protect Our Kids Act of 2012, delivered its final report Within Our Reach in March 2016. It estimated 1,500 to 3,000 U.S. children die each year from abuse or neglect, conceded that the exact number is unknown, and issued 10 recommendations — including that every state conduct a retrospective review of the previous five years of child maltreatment fatalities. HHS's statutorily required response to Congress explained that developing a national standard was "problematic" because maltreatment fatalities are a low-incidence event subject to year-to-year swings. That answer has held for a decade.

The independent estimates keep widening the gap. AEI's May 2024 report A Jumble of Standards: How State and Federal Authorities Have Underestimated Child Maltreatment Fatalities concluded that restrictive state definitions, failure to consult all available data sources, and decisions not to investigate certain maltreatment-related deaths cause NCANDS to undercount by a factor of two to three. Examining 41 states across 2011–2015, states reported roughly 7,000 fatalities against expert estimates exceeding 15,000 for the same period. The Lives Cut Short database — a collaboration between AEI and the University of North Carolina — has compiled more than 4,000 individual incidents of children dying from abuse or neglect between 2022 and 2026, many of which never appear in state or local statistics at all. Where state child death review teams have attempted to identify every maltreatment death, all but one of six such teams studied produced a higher number than the state had reported to NCANDS.

The Prior-Contact Question Nobody Wants Answered

The reason disclosure matters is not morbid curiosity. It is that a large share of these children were already in the system's hands.

Between one-third and one-half of maltreatment deaths involve families already known to CPS through prior reports, according to the Keeping the Public in the Dark analysis. Other research places 30 to 40 percent of fatality victims as having received risk assessments or family preservation services before death. An Associated Press investigation documented at least 786 children who died of abuse or neglect over a six-year period while in plain view of child protection officials — and noted the figure was certainly low, because seven additional states reported 230 open-case child deaths that could not be included since those states could not distinguish cases opened because of the fatal incident from cases already open before it.

Then there is the counting paradox that guarantees the picture stays blurry: roughly 24 states report fatality information only for children who had previous CPS cases, excluding every maltreatment death among children unknown to the agency. Nearly half of states count only the children they touched; the other half count everyone but can't separate out the ones they touched. Neither produces the number that matters — how many children died after a state agency had already been warned.

Connecticut offers a rare glimpse of what disclosure looks like when it happens. A DCF database released in 2026 documented 63 child deaths from January 2025 through May 2026, including 20 cases with prior agency involvement — nearly a third. Most states publish no equivalent.

Compliance Theater: Utah, the District, Texas, Maryland

Utah. Gavin Peterson, 12, died on July 9, 2024, in West Haven. His father Shane Peterson, the father's girlfriend Nichole Scott, and his brother Tyler Peterson were arrested eight days later and charged with child abuse homicide, aggravated child abuse, and child endangerment. Utah's Division of Child and Family Services issued its "CAPTA statement" on October 10, 2024 — three months after the death — and disclosed that reports about the family dated back years, with DCFS working with the household as early as 2013. But the statement expressly limited itself to "information relevant to Gavin in the household where his death occurred," walling off the division's history with the other household in a two-household family. This is the archetype: a document that satisfies § 5106a(b)(2)(B)(x) on its face while withholding precisely the material a reader would need to judge whether the agency failed.

The District of Columbia. D.C. Code requires that the written summary of each child fatality include the name of the child and the name of the parent or caretaker. The Child and Family Services Agency redacted every child's and every parent's name, invoking a provision that suspends disclosure where release "would endanger the life, physical safety, or physical or emotional well-being" of the victim, a sibling, or another child in the household. An "emotional well-being" exception applied categorically is not an exception; it is a repeal. The District also has no obligation to release information when a death is classified as accidental — which means neglect deaths not ruled homicides by the medical examiner disappear from public accounting entirely.

Texas. A publication requested five years of Child Fatality Release reports — roughly 1,000 documents the state had already written. Texas delayed production three times. The records were produced six months later, and only after a demand letter from attorneys at the SMU Dedman School of Law First Amendment Clinic. Texas is one of the 11 states — with Arizona, Arkansas, Colorado, Florida, Nevada, Oregon, Pennsylvania, Rhode Island, Tennessee, and Wisconsin — that proactively post notifications of fatalities, near fatalities, or other egregious incidents. It is, in other words, one of the better states.

Maryland. The Baltimore Banner requested five years of reports on suspicious child deaths — the agency's Form 1080s. The Department of Human Services denied the request in part as "unduly burdensome." Only under threat of litigation did the state concede the real reason: for roughly a third of the children who died in 2023 and 2024, the reports had never been written. A deputy secretary acknowledged that data had been "routinely released without any validation or reconciliation." As many as 83 Maryland children died from abuse or neglect in 2023 per federally reported data — more than one a week, among the worst published rates in the country. The state committed to a $210,000 database fix and, after the reporting, passed legislation to increase child fatality transparency.

New Mexico. On April 8, 2026, Attorney General Raúl Torrez released a 214-page investigative report and filed suit against the Children, Youth and Families Department. The complaint alleges CYFD stretched New Mexico's confidentiality protections — written to shield the identities of children and families in abuse and neglect proceedings — into a blanket shield against accountability, withholding information from the state Department of Justice, the State Police, the Office of the State Auditor, independent court monitors, and the courts themselves. Torrez said confidentiality provisions "have been weaponized by that agency as a form of intimidation and retaliation, not only against their own employees, but others who have information about the failures of the foster care system," including foster parents who criticized the agency. The investigation was triggered by the death of 16-year-old Jaydun Garcia, who took his own life in state custody. CYFD denies refusing to produce records. Torrez filed a second suit in July 2026 over the agency's removal of a teenager to Mexico, saying CYFD lacked "basic competency."

The Watchdogs That Don't Bark

CAPTA's 1996 amendments also required states to establish citizen review panels — volunteer bodies charged with examining whether the state agency is discharging its child protection responsibilities in accordance with the CAPTA state plan. ACF's own Report to Congress on the effectiveness of citizen review panels found that reviewed documents "gave no indication that panels reported to any person or office regarding whether their state was or was not adhering to the state CAPTA plan," and that no panel reported assessing the agency's adherence to the entire state plan or to all child protection standards. States including Wisconsin and New York require the agency to respond in writing to panel recommendations within six months. Nothing requires the panel's report or the agency's response to be published where a parent, a reporter, or a legislator could read it.

Child death review teams are locked down harder. California law makes information disclosed to a review team confidential and immune from third-party discovery; Oregon treats confidentiality as essential to the process; D.C. Code § 4-1371.09 does the same. Pennsylvania's unredacted fatality and near-fatality reports are confidential as a matter of child protective services law — the commonwealth did not publish its first Child Abuse Fatality/Near Fatality Trend Analysis Team report until 2019. The consistent design is to publish aggregate statistics and suppress the case-level findings, which is exactly backwards from what § 5106a(b)(2)(B)(x) contemplates: the statute is about the case.

Legislatures that try to pierce this run into the agencies. In Maine, a bill granting lawmakers access to confidential child welfare records died when the Senate adjourned on April 30, 2026. In Louisiana, legislators introduced a 2026 bill to expand and clarify reporting of abuse, neglect, and child deaths after learning they had never been notified of a 5-year-old's starvation death. In North Carolina, sustained WBTV reporting on a 6-year-old's death produced a child safety bill in May 2026, and the state stood up new citizen review panels in 2025.

And the federal statute itself is a ghost. CAPTA's authorization has expired. The Stronger Child Abuse Prevention and Treatment Act passed the House in March 2021 and cleared the Senate HELP committee in June 2021, and was never enacted. The last full reauthorization was 2010.

Why It Matters, and What Would Actually Fix It

Confidentiality in child welfare exists for a real reason: children who have been abused have a legitimate interest in not having their medical and sexual histories published, and reporters of maltreatment need protection from retaliation. Nothing about transparency requires sacrificing either. But a rule that shields a dead child's name to protect the child's privacy is not protecting anyone. It is protecting the agency that had the file.

The practical cost is measured in repetition. When a fatality file stays sealed, the caseworker's supervisor never faces public scrutiny, the screening algorithm that closed the prior referral is never examined, the policy that returned the child home is never revisited, and the legislature appropriating the agency's budget votes blind. The AP's 786 children, the Lives Cut Short database's 4,000-plus incidents, and the one-third-to-one-half prior-contact rate all describe the same fact: these deaths are patterned, and patterns are only visible in the case detail that states withhold.

Four changes would close the gap, and none require inventing new machinery:

Write the regulation. HHS has had thirty years to define "findings or information," "near fatality," and a disclosure deadline by rule. A federal definition of near fatality would end the situation where the same injury is reportable in Washington and invisible in a neighbor state.

Make disclosure automatic and case-level. Only 17 states release anything without a request. The Keeping the Public in the Dark recommendations are the right template: prompt notification of every reported maltreatment fatality and near fatality, comprehensive information available on request in every such case, and permitted redactions narrowed to the names of living children in the family, the identity of the reporter, and — temporarily — material that would demonstrably damage an active criminal investigation. "Emotional well-being" is not a redaction standard; it is an escape hatch.

Attach a graduated, usable penalty. The reason § 5106a(b)(2)(B)(x) is never enforced is that the only sanction is one no administrator will impose. Congress should give HHS what state budget officers already use: a partial, escrowed withholding released on compliance — precisely the $850,000 mechanism Maryland's own legislative analysts reached for. A sanction that can be used is worth more than one that can't.

Fix the count at the source. Adopt the CECANF recommendation for retrospective multi-year fatality reviews, require states to reconcile NCANDS submissions against vital statistics and child death review team findings, and fund the near-fatality data collection GAO asked HHS to price out in 2011.

Congress wrote a disclosure right into federal law and then declined to build anything that would make it real. States took the money, filed the certification, and kept the files closed. Somewhere between 2,000 and 6,000 American children die of abuse and neglect every year; a third to a half of them were already known to the agency that was supposed to protect them; and the documents that would explain what happened are, in most of the country, legally available and practically unobtainable. That is not an accident of drafting. It is the equilibrium the statute created.