The Wards

There is a doctrine in American law that says when a parent cannot or will not act in a child's interest, the state may step into the parent's place. The doctrine has a name — parens patriae, the parent of the country — and a history older than the republic itself. It is the doctrine under which abused children are removed from violent homes. It is also the doctrine under which children have been radiated, infected, sedated, trafficked, and disappeared, by institutions operating with the full authority of the state and the full silence of the appropriations process. Part I of this series traced the architecture of post-mortem extraction in American prisons. Part II turns to the children, because the architecture is the same, and the children are easier.

SECTION ONEThe Doctrine

A ward of the state is not an orphan. The legal category is older than that, and broader. A ward is anyone whose custodial relationship to a parent or guardian has been displaced by an institution that the state has designated as the custodian. The displacement may be temporary or permanent. It may be voluntary or court-ordered. It may follow abuse, neglect, immigration enforcement, mental incapacity, or simply the inability of a family to provide what the state determines a child requires. The category covers foster children, institutionalized children, immigration detainees, unaccompanied minors in federal shelters, juveniles in correctional facilities, and children whose parents have been declared incompetent.

What the category does, legally, is transfer the consent function from the parent to the institution. The parent's standing to consent to medical procedures, research participation, placement decisions, photographic likeness, and most other questions of bodily autonomy is suspended. The institution — or the state employee designated to act in its name — signs in the parent's place.

The doctrine, properly applied, is protective. It is what allows a child to be removed from a violent home in the middle of the night. It is what allows an unaccompanied minor at the border to be sheltered rather than returned to a smuggler. It is what allows a foster child to receive medical care when a biological parent is incarcerated or absent. The doctrine, in other words, is necessary.

The doctrine, improperly applied, is a consent-laundering device. It permits institutions to authorize actions on a child's body that no parent in the country would authorize on their own child — and it permits those actions to be performed under color of law, with documentation, with federal funding, and with the protection of confidentiality statutes that prevent the public, and often the biological family, from ever learning what was done.

What follows is a record of the architecture's operation. It is not exhaustive. It is, in places, deeply incomplete — because the closing of the record is itself a function of the architecture, and the architecture has had eight decades to perfect that closing. What follows is what can be documented from court filings, federal audits, state audits, congressional hearings, peer-reviewed research, and the institutions' own admissions in writing. The reader will notice, as the record accumulates, that the architecture does not change. The populations change. The extraction modes change. The closing of the record changes only in its technical methods. The architecture is the same.

SECTION TWOThe Historical Track

The American architecture for the use of institutionalized children in medical research began in the late 1940s, in two state institutions for the disabled. The lessons learned in those institutions defined the framework that every subsequent program would inherit. The framework is worth describing precisely, because it has not been retired. It has been refined.

Fernald State School, Waltham, Massachusetts (1946–1953)

The Walter E. Fernald State School was a Massachusetts state institution for children classified as "feebleminded." Between 1946 and 1953, researchers from the Massachusetts Institute of Technology and Harvard University conducted experiments in which children at Fernald were fed oatmeal laced with radioactive iron and calcium isotopes. The experiments were funded by the Atomic Energy Commission and underwritten in part by Quaker Oats, which sought data on whether the iron and calcium in its products were absorbed by children's bodies. The children were told they had been selected for membership in a "Science Club." Parents were sent consent forms describing the activity as the provision of "a special diet rich in cereals." The forms did not mention radioactivity, isotopes, the AEC, or research.

The number of children involved was small — approximately seventy-three at Fernald, plus a smaller cohort at the Wrentham State School. The radiation doses were, by later assessment, low enough that no measurable harm was documented. The point is not the dose. The point is the structure: a state institution serving as the gatekeeper for access to a population that could not advocate for itself, federal research dollars flowing through a major university to the institution, parental consent obtained by misrepresentation, and a confidentiality regime that prevented the activity from being publicly known until the Clinton Administration's Advisory Committee on Human Radiation Experiments documented it in 1995, fifty years after the experiments began.

Willowbrook State School, Staten Island, New York (1956–1971)

Willowbrook was a New York State institution for children with intellectual disabilities. By the mid-1950s it was severely overcrowded, with hepatitis A endemic in the resident population. Between 1956 and 1971, Dr. Saul Krugman of New York University Medical School conducted a series of studies in which newly-admitted children were deliberately infected with hepatitis virus — initially by feeding them extracts prepared from the feces of infected residents, and later by direct injection. The stated research objective was to map the natural history of the disease and to develop a vaccine. The actual operating premise was that the children would contract hepatitis anyway, given conditions at the institution, and that controlled inoculation under research protocol was no worse than the ambient transmission.

Parents who wished to admit their children to Willowbrook were, by the early 1960s, frequently told the institution was full — unless the child was enrolled in Dr. Krugman's hepatitis research program, in which case a bed could be made available in the dedicated research unit. The structure here is worth naming: institutional scarcity, manufactured or real, became the lever by which research participation was extracted from families. The consent form did not mention that admission to the research program was the only available admission pathway.

The Willowbrook studies generated significant scientific publication and contributed to the eventual development of hepatitis vaccines. They also generated, when journalist Geraldo Rivera's 1972 ABC News exposé brought the institution to national attention, the public consensus that something had gone very wrong inside it. Willowbrook was closed by court order in 1987. The research-program-as-admission-condition structure has never been formally repudiated by federal regulation. It has only been displaced into less visible institutional settings.

Holmesburg Prison, Philadelphia (1951–1974)

The adult parallel to Fernald and Willowbrook is essential to mention here because it demonstrates that the architecture is not unique to children. From 1951 to 1974, dermatologist Albert Kligman of the University of Pennsylvania conducted hundreds of pharmaceutical and chemical exposure experiments on prisoners at Holmesburg Prison in Philadelphia. Subjects were paid small sums to be exposed to topical drugs, dioxin, chemical warfare agents, and herpes virus inoculations. Dr. Kligman's notes from the period include a sentence frequently quoted in subsequent literature: "All I saw before me were acres of skin." The funding sources included Dow Chemical, Johnson & Johnson, the U.S. Army, the CIA, and dozens of pharmaceutical manufacturers. Holmesburg's experimentation program continued until 1974, the year after Congress passed the National Research Act and the year the federal Common Rule on human subjects research was being drafted.

The Advisory Committee on Human Radiation Experiments (1995)

In 1994 President Clinton appointed a fourteen-member Advisory Committee on Human Radiation Experiments, chaired by University of Chicago philosopher Ruth Faden, to investigate the full scope of government-sponsored radiation research on human subjects from 1944 to 1974. The Committee's final report, published in October 1995, ran more than 925 pages and documented thousands of experiments. The Committee identified specific institutional populations — prisoners, soldiers, mental patients, the terminally ill, and institutionalized children — as the recurring subject pools. Its conclusions on consent are worth reading in the original:

"The Committee finds that wherever there was insufficient evidence that the consent of subjects was both voluntary and informed, the failure to obtain such consent constitutes a moral wrong... For research that offered the prospect of benefiting the subject, the failure was less severe; for research that did not, the failure was substantial." — ACHRE Final Report, October 1995

The Committee's report produced a presidential apology, a small compensation program for a narrow subset of identified victims, and a series of regulatory clarifications that hardened the federal requirements for informed consent in subsequent research. What the report did not produce was any change in the institutional architecture that had made the experiments possible — the architecture in which custodial institutions served as the gatekeepers for access to populations that could not advocate for themselves. That architecture remained in place. The next program would simply have to be more careful about documentation.

SECTION THREEIncarnation Children's Center

The most consequential American institutional research program involving wards of the state since Willowbrook operated on West 142nd Street in Harlem from 1988 to 2002. Incarnation Children's Center was a nursing facility for HIV-positive children, the majority of whom were wards of the New York City Administration for Children's Services. The children had been removed from biological parents who, in many cases, had also been HIV-positive. The Administration for Children's Services placed the children at Incarnation under standing custodial authority. ACS, in its capacity as the legal guardian, was the consenting party for the children's medical treatment, including their participation in clinical drug trials.

Between 1988 and 2002 Incarnation served as a research site for at least thirty-six pharmaceutical clinical trials sponsored or funded by the National Institutes of Health, the National Institute of Allergy and Infectious Diseases, GlaxoSmithKline (and its corporate predecessors), Bristol-Myers Squibb, Pfizer, and other pharmaceutical manufacturers. The research was coordinated through Columbia University's Pediatric AIDS Clinical Trials Unit. The trials tested early-generation antiretroviral drug combinations — including AZT, nevirapine, ritonavir, nelfinavir, and others — for safety and efficacy in HIV-positive children. Many of the protocols involved Phase I and Phase II studies of drugs whose toxicity profiles were not yet established, particularly in pediatric populations.

The federal regulations governing such research, in force during the entire period, were unambiguous. Forty-five C.F.R. § 46.409, the Common Rule's provision on research involving wards of the state, restricts such research to two narrow categories: research related to the children's status as wards (which Incarnation's pharmaceutical trials were not), and research conducted in schools, camps, hospitals, institutions, or similar settings in which the majority of children involved are not wards. Twenty-one C.F.R. § 50.56, the FDA's parallel regulation, imposes the same restriction. The Incarnation trials, in which the substantial majority of subjects were wards, fell outside both regulatory exceptions.

The federal complaint detailing this discrepancy was filed by the Alliance for Human Research Protection on March 22, 2004, addressed to the Office for Human Research Protections at HHS. The complaint, drafted by Vera Hassner Sharav and supported by independent reporting by Liam Scheff in the New York Press, alleged that thirty-six trials had been conducted on Incarnation children in violation of 45 C.F.R. § 46.409 and that the Administration for Children's Services, in signing as guardian for the children, had not satisfied the regulatory requirement that a ward's participation in non-therapeutic research be approved by a court-appointed advocate independent of the institution.

The investigation that followed, conducted by HHS's Office for Human Research Protections and the New York City ACS itself, produced a 2009 final report by the Vera Institute of Justice that confirmed several of the complaint's central findings while declining to characterize the conduct as a regulatory violation. Vera's report documented that approximately 532 children at Incarnation had been enrolled in clinical trials during the period in question. Vera's report also documented that approximately 25 of those children had died during or shortly after their trial participation.

The causation question is the one the architecture forecloses. The children were HIV-positive. Some of them would have died regardless. Some of them may have died from disease progression. Some of them may have died from drug toxicity. The available record — the trial sponsor's data, the institution's medical records, the city's death certificates — does not permit independent determination of cause for the twenty-five deaths. The HIV diagnosis, in the architecture's operation, forecloses scrutiny. The children were sick. The trial was experimental. The death is attributable to the underlying condition unless someone can prove otherwise, and the records required to prove otherwise are sealed under research confidentiality, medical privacy, and the city's standing custodial privilege.

Incarnation Children's Center closed in 2012. The medical records, the trial data, and the children's identities remain under the same custodial seals that governed them during the program's operation. The pattern — institutional custody, transfer of consent to the custodian, pharmaceutical-trial revenue, and the closing of the record through confidentiality — was complete.

SECTION FOURThe Medication Architecture

Most American children whose consent is held by the state are not enrolled in clinical trials. They are, instead, prescribed psychotropic medications. The scale of this prescribing is large, well-documented, and federally audited. The documentary record is thirteen years long and has not produced systemic change.

The GAO's December 2011 finding was unambiguous. In the five states examined — Florida, Massachusetts, Michigan, Oregon, and Texas — foster children were prescribed psychotropic medications at rates 2.7 to 4.5 times higher than non-foster children covered by Medicaid. The disparity is not, by itself, evidence of malpractice; foster children, as the GAO acknowledged, often have higher mental-health needs than the general population. The disparity is, however, the floor of the analysis, not the ceiling.

The ceiling is contained in two further GAO findings that have received less attention than the headline ratio. The first: hundreds of foster and non-foster children in the five states examined were prescribed regimens involving five or more psychotropic drugs concurrently. The GAO's contracted experts — two child psychiatrists with mental-health research backgrounds in foster care — were unambiguous on the supporting evidence for such regimens, in adults or in children:

"According to GAO's experts, no evidence supports the concomitant use of five or more psychotropic drugs in adults or children." — GAO-12-201, December 2011

The second: thousands of children in the same data were prescribed doses higher than the established maximum recommended for their age category. The numbers were not estimates. The numbers were drawn from state Medicaid prescription records that the prescribing physicians, the pharmacies, and the state Medicaid administrators had all signed off on.

The GAO's 2014 follow-up report (GAO-14-362) examined twenty-four specific cases sampled from the population in the 2011 analysis. Two contracted child psychiatrists were retained to evaluate whether the clinical documentation in each case supported the prescribing. The GAO's report describes the documentation as showing "varying quality" — an audit-report locution that translates, in plain language, into a finding that in some of the cases the documentation did not support the prescribing.

The 2017 follow-on (GAO-17-129) returned to the underlying ratios and confirmed that despite intervening federal guidance and state-level pilot programs, the foster-to-non-foster prescribing differential had not closed. The pattern was stable. The institutional consenting party — the state child welfare agency, acting through county and contracted-provider designees — signed off on prescribing decisions that fell outside the supporting clinical evidence. The pharmaceutical manufacturers received the Medicaid reimbursement. The state received the federal Title IV-E match. The children received the prescriptions. The closing of the record was effected through the routine medical-privacy regime that prevents independent audit of pediatric psychiatric prescribing at scale.

The architecture, as in Incarnation, did not require any of the parties to violate any federal regulation. The architecture required only that the consent function rest with an institution whose financial incentives ran parallel to the prescribing decision. The state agency was reimbursed for the foster placement at a higher rate when the child carried a diagnosis. The diagnosis often arrived with a prescription. The prescription was reimbursed by Medicaid. The child remained in placement. The consent had been transferred at the moment of removal from the biological parent. Everything that followed was within the operating envelope of the doctrine.

SECTION FIVENorth Dakota — The Closing of the Record

The architecture's clearest contemporary documentation comes from a state most readers will not have associated with the foster-care or institutional-licensing failure record. North Dakota is small — approximately 760,000 residents in 2015. Its state government is concentrated, with relatively few institutional layers between policy and implementation. Its political culture is conservative and ostensibly accountable. The state's institutional posture, in short, is the posture least likely to produce systemic concealment. And yet the record produced by the state's own auditor is the most complete contemporary documentation in this series of how the architecture operates and how it closes the record on its own operation.

What follows is drawn from the primary-source state audit report transmitted on July 13, 2016 by State Auditor Robert R. Peterson to Governor Jack Dalrymple, the North Dakota Legislative Assembly, and Department of Human Services Executive Director Maggie Anderson. The report covers the biennium ended June 30, 2015. The auditor-in-charge was Allison Bader; the audit manager was Paul Welk, CPA. The report contains twenty numbered findings. Findings 15-9 through 15-19 concern Early Childhood Services Licensing — the regulatory infrastructure under which approximately 39,000 North Dakota children were placed daily in the care of 1,624 licensed providers.

Finding 15-9 — Failure to Suspend Providers and Notify Parents

Childcare providers in North Dakota were permitted to continue operating under Memoranda of Understanding while the Department had confirmed knowledge of specific activities inside the licensed facilities. The activities, as enumerated in the audit's own findings: illegal drug use by the provider; restricted persons present at the facility; inappropriate touching from adults; inappropriate sexual play between children; and other concerns of supervision and discipline. Thirteen MOU agreements were identified during the audit review — two within the 58-license sample, eleven additional flagged during the broader review. The audit notes that under NDCC 50-25.1-11, the confidentiality statute, the auditor could not disclose specific details of the underlying investigations. The auditor could, however, document the structural failure:

"Upon confirmation of activities jeopardizing the health and safety of children, providers are allowed to continue operating under memorandum of understanding (MOU) agreements... No action was taken by the Department to perform further oversight of providers operating under MOU agreements and the Department did not notify parents with enrolled children." — ND State Auditor Finding 15-9, July 2016

The Department's own explanation, recorded in the audit's Cause section, is the sentence that should be read twice:

"While state law allows for parents to be notified at any time after the onset of a child abuse and neglect investigation, the Department has indicated that procedures are to notify parents after the provider's appeal process." — ND State Auditor Finding 15-9, July 2016

The North Dakota statute — the law as passed by the legislature — permits the Department to notify parents at any time after an abuse or neglect investigation begins. The Department, by internal policy, had inverted the statutory permission into a procedural prohibition. Parents would not be told their child was in a facility where an adult was committing inappropriate touching until the provider had exhausted the administrative appeals process. The provider's procedural rights had been elevated above the parents' right to know. The audit identified this. The Department acknowledged it. The provider operated. The children remained in care.

Finding 15-13 — License Effective Before Background Checks Performed

The Department's stated policy permitted childcare provider licenses to be effective before the regional supervisor reviewed the background check materials. The justification, as recorded in the audit's Cause section, was administrative convenience — "delay in review of application documents due to employee absence at the county and regional offices."

Of 58 licenses tested: 25 were dated effective before the regional supervisor's review. 51 of 58 lacked documentation that proper background check procedures had been performed at all. An 88 percent documentation failure rate. The audit further notes that the fingerprint background checks performed by the Department's Criminal Background Check Unit — the post-2014 enhanced check required by the Child Care and Development Block Grant Act reauthorization — were significantly backlogged due to the new federal requirements. The Department's response: it knew. It had advocated for additional staffing. We will return to that point.

Finding 15-14 — Improperly Backdating License Effective Date

Nine of 58 tested licenses had been backdated by the Department. The audit lists the specific cases in language drawn directly from the case files:

• Five providers had licenses effective prior to the submission of applications, completion of fire or health and sanitation inspections, performance of the licensing study, or authorization for completion of the child abuse and neglect background searches.
• One provider was issued a license after testing positive for illegal drugs and a subsequent negative test. The license was backdated as effective prior to and during the time of confirmed drug use.
• One provider was issued a correction letter six days after the effective date of the license stating that the application fee was not collected, training hours were incomplete, and all authorization forms for child abuse and neglect background searches were not completed.
• One provider had health and sanitation deficiencies that, per the Department's own correspondence, had not been corrected for two years.
• One provider did not have a date identified by the county licensor of when all license application documents were received, so a proper effective date for the license could not be determined.

The Department's policy on file at the time of the audit — DHS Policy 620-01-35-25 — stated explicitly: "A license will not be backdated if a provider fails to submit all required licensing documentation prior to the expiration date." The policy was on the books. The backdating was happening anyway. The auditor's documented finding, in plain language: the Department had a written policy against backdating, and was backdating.

Finding 15-17 — Child Care Providers Not Properly Licensed

Of 58 providers tested: 4 with no record of application fee payment; 10 with no indoor floor plans showing room square footage; 24 with no outside play space plans with square footage; 4 with no health and sanitation reports; 3 with no fire inspection reports; 2 with no pet vaccination records; 15 with no programming schedules; 13 with no policies and procedures documentation; 14 with no childcare evacuation plan; 5 with no CPR or First Aid certificates; 13 with no certifications or verification of training hours; 14 with no evidence of degrees, certifications, or experience requirements; 5 with no liability insurance.

The specific cases listed: one provider licensed for nine children when space restrictions allowed only eight; one provider operating with a health and sanitation report issued for a facility under a previous owner (the audit notes that licenses are statutorily non-transferable, a point the Department's own staff would have known); one provider whose insurance was an unaccepted quote rather than evidence of policy; one provider whose insurance policy began two months after the license effective date.

Finding 15-18 — Lack of Monitoring of Corrective Orders

Corrective orders are the mechanism by which the Department, having identified a deficiency at a licensed facility, requires the deficiency to be cured. 193 of 444 corrective orders (43 percent) issued during 2014 and 2015 did not identify a date on which the orders were determined to have been corrected. The audit could not determine whether the cited deficiencies were ever fixed. The Department, the audit notes, did not have a tracking method for ensuring sanctions were properly imposed when corrective orders went unresolved.

Finding 15-19 — Ineffective Monitoring Procedures

This is the auditor's structural conclusion. It is worth quoting in full, because it constitutes the auditor's own characterization of what the Department's monitoring infrastructure had become:

"The significant errors identified in our testing of child care provider licensing indicate these reviews are clearly ineffective at identifying material weaknesses and inconsistencies. These situations include failure to identify incomplete application requirements, improperly licensed providers, nonperformance of unannounced inspections, backdated licenses, lack of monitoring of corrective orders, non-imposed sanctions, and lack of documented notification to parents for serious safety concerns." — ND State Auditor Finding 15-19, July 2016

That paragraph is the closing of the record articulated as audit finding. The audit was telling the legislature: the state's licensing system is not functioning, the monitoring infrastructure that is supposed to detect that failure is itself ineffective, and the failures are occurring across every element of the licensing process.

The Defunding

The sentence that ties this audit to the architecture of Part I — the sentence that demonstrates how the closing of the record operates at the appropriations level — appears three times in the Department's formal responses, attached to Findings 15-13, 15-17, and 15-19. It is the same paragraph, repeated verbatim:

"The Governor's Budget for the 15-17 biennium included 3 additional FTEs for regional supervisors due to their increased workload and also 1.5 FTEs to ensure background checks were completed within the timelines required by the Child Care and Development Block Grant of 2014. During the 2015 legislative session the 4.5 FTEs and associated funding were removed from the Department's 15-17 appropriation." — ND Department of Human Services Response, July 2016

The Governor had proposed the remediation. Three additional regional supervisors and one and a half full-time-equivalent positions to handle the federally-mandated background check timelines — 4.5 FTEs in total. The 2015 North Dakota Legislative Assembly stripped the funding. The audit then documented the consequences of the defunded position. The Department, in its formal written response to the audit, identified the legislative action as the cause — three separate times, in case anyone reading the audit missed it the first two.

This is the architecture made visible. The harm is identified. The fix is proposed. The legislature kills the fix. The harm continues. The next audit cycle returns the same finding. The audit cycle after that returns it again. The cumulative record — eventually — produces a State Auditor's statement that 577 cases of severe child abuse for extended amounts of time may have been avoided if the original 2018 findings had been addressed. But the original 2018 findings were a continuation of the 2015 findings. And the 2015 findings were the consequence of a legislative vote.

The Anonymous Child

The audit cannot name her. NDCC 50-25.1-11 forecloses disclosure. The reporter, Chris Berg of Point of View, did not name her in the August 29, 2016 broadcast that brought the audit to wider attention. What is on the documentary record is the structure of her death. She was five years old. She drowned in a pool at a childcare facility licensed by the Department of Human Services. The facility's license had been backdated. The reporter's observation on air, which he properly framed as speculation: if the Department had followed its own rules, she might still be alive.

That is the architecture's logic in one sentence. The Department had a written policy against backdating. The Department backdated the license. A facility was permitted to operate. A child went into the facility. A child died. The audit identified the policy violation. The Department promised reform. The legislature defunded the staffing required to perform the reform. The audit cycle returned three years later with the same finding. The 577-cases-preventable statement came six years after that.

The closing of the record was complete before her parents knew what had happened. The five-year-old's death was, in the architecture's operation, the cost of doing business.

SECTION SIXThe National Audit

If North Dakota were an outlier, the case would be one of state-level failure rather than architectural pattern. The HHS Office of Inspector General's 2024 national audit establishes that North Dakota was not an outlier. North Dakota was a case study of a national operating condition.

The OIG report — State Agencies Can Improve Their Reporting of Children Missing From Foster Care to Law Enforcement for Entry Into the National Crime Information Center Database As Required by Federal Statute — examined the period from July 1, 2018 to December 31, 2020. The scope is worth reading slowly. The audit examined 74,353 episodes in which a child was missing from foster care for two calendar days or longer during the thirty-month review window. Each episode represents a child whose state custodian could not locate that child for at least forty-eight hours.

Federal statute requires state child welfare agencies, upon learning that a foster child has gone missing, to report the child to law enforcement for entry into the FBI's National Crime Information Center database within 24 hours. The OIG sampled 100 missing-child episodes from the 74,353-episode population. Of those 100, only 86 had been reported to law enforcement within the 24-hour federal requirement. Fourteen percent of sampled episodes were federal reporting failures. Extrapolated against the full population of episodes, the projection — which the OIG declined to quantify in a single sentence, presumably because the implications were considered too inflammatory for an audit narrative — would represent thousands of children whose disappearance from state foster care was never entered into the federal database that exists, in significant part, to locate missing children.

The architecture, again. State custody. Federal reporting requirement. Federal reporting failure. The accountability vacuum at scale.

SECTION SEVENThe Federal System

For roughly two decades, the federal government has operated its own custodial system for children. The Office of Refugee Resettlement, an agency within the Administration for Children and Families inside the Department of Health and Human Services, has primary federal custody of unaccompanied alien children — minors apprehended at the U.S. southern border without a parent or legal guardian. The William Wilberforce Trafficking Victims Protection Reauthorization Act of 2008 created the current statutory framework. The TVPRA requires unaccompanied minors to be transferred from Department of Homeland Security custody to ORR within 72 hours. ORR then becomes the custodian, with statutory responsibility for safe placement and welfare.

The scale of this custody is substantial. ORR's network includes approximately 300 facilities in 27 states. In Fiscal Year 2022, DHS referred 128,904 unaccompanied minors to ORR. The custodial pipeline operates by referral to "sponsors" — adults in the United States, in the majority of cases relatives, to whose care the child is released following an ORR vetting process.

The documentary record on ORR's operation is unusually full. Hannah Dreier's Pulitzer-winning New York Times investigative series, which began publication in February 2023, documented at scale that minors released from ORR custody were ending up in dangerous full-time labor across the United States — meatpacking plants, factory floors, construction sites, roofing crews. Two-thirds of unaccompanied minors after release were, per Dreier's reporting, working full-time jobs that were either illegal under federal child-labor law or in hazardous conditions. Documented injuries included crushed limbs and seared lungs. Major American brands' supply chains were named in the reporting.

Internal whistleblowers at ORR documented the same pattern from inside the agency. A July 2021 memo from eleven ORR managers, obtained by the New York Times, stated that the office "rewards individuals for making quick releases, and not one that rewards individuals for preventing unsafe releases." Then-HHS Secretary Xavier Becerra was recorded, per the Times, using assembly-line language with staff: "If Henry Ford had seen this in his plants, he would have never become famous and rich. This is not the way you do an assembly line."

The named deaths in ORR custody have been documented but not aggregated in a single federal disclosure. Ángel Eduardo Maradiaga Espinoza, age 17, from Honduras, died on May 10, 2023 at the Gulf Coast Jewish Family and Community Services facility in Safety Harbor, Florida. The Pinellas County Medical Examiner conducted the autopsy; HHS has declined to release further information citing "privacy and safety." Darlyn Valle, age 10, died in September 2018; the death was not publicly disclosed for nearly eight months. Juan de León Gutiérrez, age 16, died on April 30, 2019 of health complications following days in ORR care. In each case, the federal custodian's standing as the consenting party for the child's medical care, the institutional confidentiality regime governing disclosure to families and to the public, and the absence of an independent oversight authority capable of compelling release of clinical records, produced the same closing of the record observed in the Incarnation cases two decades earlier.

In February 2025, the Department of Homeland Security and the Department of Health and Human Services rewrote the Memorandum of Agreement governing information-sharing between ICE and ORR. ICE officers — not the previously-authorized juvenile coordinators alone — were granted access to the ORR unaccompanied minors database. The agreement's full operational consequences are still emerging. What is clear is that the database originally constructed as a child-welfare tool is now also an immigration-enforcement tool. The institutional posture going forward will inherit both functions in a single set of records.

SECTION EIGHTThe Transit Route

The most contested testimony associated with this series concerns what happens to some unaccompanied minors before they reach the U.S. border — during their transit through Mexico, in the hands of the smuggler and cartel networks that operate the routes. The testimony is unusually direct, was delivered under oath before a congressional committee, and has not been independently corroborated. The evidentiary status of the testimony — what it is, and what it is not — must be stated precisely, because the structural argument of this section does not depend on the testimony being adjudicated. The structural argument depends only on the inability of the federal custodial system to certify that the conditions described in the testimony are not occurring.

On November 19, 2024, the U.S. House Committee on Homeland Security — the Subcommittee on Border Security and Enforcement and the Subcommittee on Oversight, Investigations, and Accountability — held a joint hearing titled Trafficked, Exploited, and Missing: Migrant Children Victims of the Biden-Harris Administration. The witnesses included Tara Lee Rodas, a former HHS Deputy to the Director of the Federal Case Management Team and identified federal whistleblower; J.J. Carrell, a retired U.S. Border Patrol Deputy Patrol Agent in Charge; Alicia Hopper, a consultant on human trafficking affiliated with Sadulski Enterprises LLC; and Kathy Larin, Director of the Government Accountability Office's Education, Workforce, and Income Security division.

Vice Chairman Michael Guest, Republican of Mississippi, questioned the witnesses directly on the question of organ harvesting and unaccompanied children. The testimony, drawn from the official hearing record:

"Organ harvesting is real. To think that it's not, that... this federal government's bringing children over, that are being lost, sexually mutilated, raped, murdered, and then we're shocked or appalled that there could be organ harvesting. The people I've interviewed, multiple people — and I asked them, is this an anomaly? Is this a one-off? No. This is pervasive and is growing daily." — J.J. Carrell, oral testimony, November 19, 2024

"During the interview with the former Sinaloa Cartel affiliate... He expanded on the fact that these migrant children, especially if they're traveling without their parent, there is no way to communicate whether they've made it through their journey or not. And if a cartel member or smuggler gets a call that there is a specific organ that they're looking for, they will scout out those organs and especially focus on migrant children, namely for the fact that they are vulnerable, they're unable to fight back. And, as horrid as it sounds, their organs are more intact and in better shape. So, there is no way to communicate with these children once they've crossed into the border, so they kind of disappear on the way." — Alicia Hopper, oral testimony, November 19, 2024

Hopper's written testimony added that in her conversation with the former sex trafficker and ex-Sinaloa cartel member, "he revealed the lengths family members will go to for a dying relative, illustrating how organ traffickers exploit this vulnerability." Her written statement referenced a 12-year-old boy "whose eye was harvested in Mexico for $15,000."

The ranking Democrat, Representative Glenn Ivey of Maryland, posed the obvious follow-up question: were there documented court cases, in Mexico or in the United States, supporting these accounts? Hopper testified that her information came from her prison interview with the cartel affiliate and that she was keeping her source confidential to protect him from cartel retaliation. Asked the same question, Carrell stated that he had interviewed people who said the practice was occurring inside U.S. borders. Pressed on whether anyone had been charged with organ harvesting, Carrell acknowledged he did not know of charges.

This is the evidentiary line, and it must be drawn clearly. The testimony is on the record. The witnesses have documented credentials. The testimony has not been independently corroborated through public arrests, indictments, forensic findings, named victims with autopsy reports, or transplant-network traces in U.S. courts. The testimony is consistent with the documented international record on organ trafficking targeting migrants — a record maintained by the United Nations Office on Drugs and Crime, the Council of Europe, and peer-reviewed medical literature, which has documented for two decades that international organ-trafficking networks operate on migrant and refugee corridors worldwide. The Mediterranean route, the Southeast Asian route, and the African overland route have all been independently documented. The Mexican transit route is the route through which the children referenced in the November 2024 hearing pass.

The structural point does not require the testimony to be adjudicated:

The American government's documented inability to maintain contact with tens of thousands of unaccompanied minors after release from federal custody does not, by itself, prove those children entered any specific exploitation pipeline. It does, however, demonstrate that the federal system cannot certify that they did not.

A custodial architecture that loses 85,000 children to its own follow-up call system, releases 31,000 of them to incomplete addresses, fails to serve Notices to Appear on 233,000 of them, and rewrites its information-sharing memoranda across administrations without re-establishing custody-of-the-record, is not a system capable of falsifying the testimony just quoted. The architecture is incapable of the falsification. The closing of the record forecloses it.

SECTION NINEThe Architecture, Applied

What follows is the four-step pattern that connects every section of this article to every section of Part I, and that will appear again in subsequent installments. The pattern does not depend on the institution. It does not depend on the era. It does not depend on the consent regime that nominally governs the institution. The pattern depends only on the legal transfer of consent from a parent or close relative to an institution whose financial or operational incentives run parallel to a specific extraction mode.

The institutions change. The funding flows change. The extraction modes change. The pattern does not.

What the pattern produces, in every operating mode, is a population whose bodies and labor and biological material are accessible to extraction by an institutional custodian whose financial incentives align with the extraction. What the pattern requires, in every operating mode, is a confidentiality regime that prevents the family of the affected individual, the journalistic press, and the legal system from independently verifying the extraction's occurrence or scale.

The pattern is the architecture. The architecture is not new. The architecture has not been disturbed.

SECTION TENThe Constitutional Throughline

The Fourteenth Amendment's Due Process Clause is the document of last resort here, as it was in Part I. The clause has, since its 1868 ratification, been read by the Supreme Court to protect a substantive zone of liberty interests that includes — among others — the right of parents to direct the upbringing of their children. The modern statement of the doctrine is Troxel v. Granville, 530 U.S. 57 (2000), in which Justice O'Connor's plurality opinion observed that "the interest of parents in the care, custody, and control of their children is perhaps the oldest of the fundamental liberty interests recognized by this Court."

The doctrine does not, by itself, foreclose state intervention in cases of abuse, neglect, or genuine custodial necessity. It does, however, establish that the displacement of parental authority by the state is a constitutional event that requires justification and that does not extinguish all parental interests in the child's well-being. The parent retains an interest. The state's role as substitute custodian is bounded by that retained interest.

The architecture documented in this article operates as if the constitutional throughline does not apply. The institutional custodian, having received transfer of consent through a court order, an administrative placement, a federal apprehension, or in some cases nothing more than the parent's poverty, treats the consent transfer as final. The parent's standing to know what is happening to the child is, in the operating logic of the architecture, extinguished by the placement.

This is not what the doctrine permits. This is what the doctrine has permitted in practice, because the alternative — an enforceable parental right to ongoing information about a child in state custody, including information about medical procedures, research enrollment, prescribing patterns, and disclosure of harm — would foreclose the architecture's operation. The institutional custodians, the research sponsors, the pharmaceutical manufacturers, and the agency budgets all depend on the architecture remaining as it is. The doctrine that would constrain the architecture has been left, in practice, to academic citation rather than litigation.

That can change. Pro se petitioners have, in the cases documented in Part I, begun to articulate the Fourteenth Amendment claim in pleadings that the agency defendants have not yet found language to dismiss. The same claim, articulated by the biological family of a child whose state-custody death produced no notification, no autopsy, no return of organs, and no opportunity to contest the medical decisions made during the institutional placement, would land in a constitutional doctrine that the Supreme Court has not yet had occasion to extend into the custodial-extraction context. The architecture's vulnerability is in that extension.

CLOSINGWhat Remains

The closing of the record is the operation. The audits will continue to be written. The findings will continue to be repeated. The legislative budget cycles will continue to defund the remediation, because the defunding is itself the architecture functioning as designed. The deaths will continue. The disappearances will continue. The pharmaceutical trials, the foster placements, the unaccompanied-minor pipelines, and the transit-route exploitation will continue.

What changes is whether the public record permits the architecture to remain undocumented. The Faraino consolidated cases changed it for Alabama prisoners. The North Dakota State Auditor's biennium report changed it for North Dakota childcare licensing. Hannah Dreier's Times series changed it for ORR labor pipelines. The 2024 House hearing changed the transit-route question from one that could be dismissed as speculation to one that sits on the official record of the United States Congress.

None of these documents, individually, has produced systemic change. The institutional incentives have not shifted. The agencies remain in place. The contracts continue to flow. The biennium audits return.

What the documents have done, collectively, is to produce a public record from which a constitutional doctrine could be extended. That extension is the work that remains. It is not investigative work. It is litigation work. It is appropriations-process work. It is electoral work. It is the work of a body politic that has begun to read its own audit reports and to ask why the architecture has not been disturbed.

The harvest does not stop because it is named. The harvest stops because the consent transfer is closed.

Part III — The Unclaimed — will examine the institutional disposition of adult bodies for whom no family was ever notified.