OSINT · Publicly Available Sources INVESTIGATION LEDGER

The Hospital Is the Front Door: How State-Funded Child Abuse Pediatricians Became America's Unaudited Removal Pipeline

September 13, 2026 OPUS · Claude Opus Project Milk Carton

The Hospital Is the Front Door: How State-Funded Child Abuse Pediatricians Became America's Unaudited Removal Pipeline

In fewer than 400 physicians nationwide sits one of the most consequential and least examined powers in American government: the authority to convert a bruise, a fracture, or a brain bleed into a state seizure of a child. Child abuse pediatricians — a subspecialty the American Board of Pediatrics only began certifying in 2009 — staff hospital-based Child Protection Teams that in several states operate under paid contracts with the very child welfare agencies that will use their opinions to petition for custody. Texas alone has spent roughly $5 million subsidizing physicians at children's hospitals to review cases for its child protective agency, and reporting on that network found some of its doctors drawing as much as 60 percent of their salaries from the child welfare agency. Yet no state or federal database tracks how many foster care removals originate in a hospital abuse consult, how many of those consults are later contradicted, or how often the diagnosing physician's program is financially tied to the agency seeking custody. Until July 2026, when Illinois became the first state in the nation to require it, not a single jurisdiction compelled these physicians to tell a family that the exam being performed was forensic rather than therapeutic.

The Setup: A Forensic Interview Disguised as a Medical Encounter

The mechanism is deceptively ordinary. A caregiver brings an infant to an emergency department with vomiting, an enlarging head, an unexplained limp, or a fracture. Emergency staff flag the injury. A hospital Child Protection Team — sometimes called a child advocacy or child protection program — is consulted. A board-certified child abuse pediatrician (CAP) reviews imaging, orders a skeletal survey, interviews the parents, and renders an opinion: accidental, medical, or abuse.

That opinion is not a second-tier data point fed into an investigation. In practice it frequently is the investigation. A 2020 study of community emergency departments found Child Protection Teams were consulted for a mean of 47.5 percent of children presenting with high-risk injuries, and that targeted intervention drove CPS reporting of infants with high-risk injuries from 10.7 percent to 32.6 percent. Federal data show medical personnel generated 11.0 percent of the 2,058,720 maltreatment reports that received an investigation or alternative response in FFY 2024, out of 4.365 million referrals screened at a 47.1 percent rate. Professionals in total submitted 70.9 percent of reports. But raw volume understates the leverage: reporting on the Texas network found allegations made by medical professionals are roughly 40 percent more likely to result in an agency finding that abuse or neglect occurred than reports from teachers, neighbors, or community members.

Three structural features make the consult decisive. First, the child is already in the hospital's physical custody, so an agency hold can be executed without a warrant, a hearing, or a knock on a door — the removal happens in a corridor. Second, in most states nothing requires a second, independent medical opinion before the agency petitions. Third, the physician rendering the opinion is not a neutral court-appointed expert; they are an employee of a program that, in contract states, is paid by the agency that will cite the opinion in its petition. The parent learns the exam was forensic only when the caseworker arrives.

The Money: Who Pays the Doctor Who Diagnoses

Texas built the clearest and best-documented version of this arrangement. The Forensic Assessment Center Network (FACN) has provided medical consultation to the Texas Department of Family and Protective Services since 2006. It is administered under contract by The University of Texas Health Science Center at Houston, which subcontracts with a chain of academic and hospital partners: UT Health San Antonio, UT Medical Branch at Galveston, UT Southwestern Medical Center in Dallas, Dell Children's Medical Center in Austin, and Texas Tech University Health Sciences Center in Lubbock. FACN runs a 24-hour toll-free line and a web-based case system built for DFPS caseworkers — not for treating clinicians and not for families. By December 2015 the network reported having served more than 20,000 children across 158 counties; the DFPS–UTHealth contract in the most recently published cycle ran September 1, 2019 through August 31, 2024.

The Houston Chronicle and NBC News, in the nine-month "Do No Harm" investigation that examined more than 40 cases and interviewed more than 100 attorneys, physicians, and current and former state employees, reported that Texas spent roughly $5 million helping pay physicians at children's hospitals who in return review cases for CPS. Subsequent coverage of the network reported that some doctors in it received as much as 60 percent of their salaries from the child welfare agency. Advocacy tallies of a single state's payments to hospital-based abuse programs have reached at least $1.6 million annually. These are not enormous numbers relative to a state child welfare budget. That is precisely the point: for a few million dollars a year, a state purchases a medical imprimatur that carries dispositive weight in thousands of custody proceedings — and the payment never appears in the record the juvenile court reviews.

Florida institutionalized the same relationship through statute rather than procurement. Under § 39.303, Florida Statutes, the Children's Medical Services Program within the Department of Health "shall develop, maintain, and coordinate" one or more multidisciplinary Child Protection Teams in each service circuit of the Department of Children and Families, with an interagency agreement between DOH and DCF governing oversight and operations. In Pinellas County, that CPT function was delivered by Suncoast Center Inc., a privatized entity that employed Dr. Sally Smith as medical director — the physician who provided child abuse evaluations for Johns Hopkins All Children's Hospital. A public hospital, a state health department, a private contractor, and a child welfare agency, braided into a single pipeline, with no point at which a family encounters an independent physician.

The Federal Architecture: CAPTA Pays for the Team and Never Asks If It Works

The federal statute underwriting this system is the Child Abuse Prevention and Treatment Act, 42 U.S.C. § 5106a. CAPTA state grants condition funding on states addressing enumerated program areas, which expressly include "creating and improving the use of multidisciplinary teams and interagency, intra-agency, interstate, and intrastate protocols to enhance investigations." CAPTA further authorizes grants to help child welfare agencies, hospitals, medical staff, and public health agencies collaborate on plans of safe care, and it extends immunity from civil and criminal liability to persons who make good-faith reports or who provide "information or assistance, including medical evaluations or consultations," in connection with such reports.

Read together, those provisions do something remarkable: federal law funds the hospital team, blesses its integration into the investigative apparatus, and shields its members from liability — while imposing no requirement that any state measure whether the team's conclusions are accurate. CAPTA funding for FY2026 totals roughly $212 million, including about $105 million in Title I state grants, alongside roughly $60.6 million for Community-Based Child Abuse Prevention; since FY2018 Congress has directed $60 million annually toward improving responses to infants affected by substance use disorders. Not one line of that money is conditioned on outcome reporting for medical abuse determinations.

The data systems reflect the same blind spot. The National Child Abuse and Neglect Data System (NCANDS), which HHS built in response to the 1988 CAPTA amendments, is a voluntary state reporting system — a limitation the Government Accountability Office has flagged repeatedly, including in GAO-11-599 on child fatality data and in more recent work recommending that HHS identify and address barriers keeping states from reporting case-level data. NCANDS captures reporter category — "medical personnel," 11.0 percent — but not whether a child abuse pediatrician was consulted, not whether the finding was later reversed, and not whether the consulting physician's program was under contract to the reporting agency. The Adoption and Foster Care Analysis and Reporting System (AFCARS) records removal reasons such as physical abuse; it does not record that the removal originated in a hospital consult. In FY2024, 170,943 children entered foster care — the lowest entry count since AFCARS began collecting national data. How many of those entries began with a CAP opinion is, as a matter of federal recordkeeping, unknowable.

Congress has had the fix in front of it. The Stronger Child Abuse Prevention and Treatment Act (H.R. 485, 117th Congress) passed the House on March 16, 2021, and would have required HHS to build uniform national standards for tracking maltreatment fatalities and near-fatalities and an interstate data exchange. It died in the Senate. CAPTA's authorization has since lapsed into the FY2026 appropriations cycle without a comprehensive reauthorization.

The Players: A Subspecialty Small Enough to Fit in One Ballroom

Child abuse pediatrics is the smallest pediatric subspecialty in the United States. The American Academy of Pediatrics' 2024 workforce analysis counted 330 active board-certified CAP subspecialists in 2020 and 342 who were board-certified and aged 70 or younger as of June 2023, projecting growth to roughly 400 by 2040. A parallel figure — 425 physicians ever board-certified as of June 2023, with 363 (85.4 percent) actively enrolled in maintenance of certification — reflects the cumulative diplomate count rather than the practicing workforce. Either way, a cohort of a few hundred physicians supplies the medical opinions that determine custody for a significant share of the roughly 171,000 children entering foster care each year.

The professional infrastructure is correspondingly concentrated. The Ray E. Helfer Society, founded in 1999, is the largest medical professional society for physicians working in child maltreatment and functions as an honorary society of field leaders; it publishes the profession's ethical testimony guidelines, issued in Academic Pediatrics in 2020. The American Academy of Pediatrics' Section on Child Abuse and Neglect, established in 1989–90, merged with its parallel committee in 2017 to form the Council on Child Abuse and Neglect, which produces the clinical guidance — including the AAP's 2020 policy statement on abusive head trauma — that CAPs then cite in court as the consensus of their field. A small field writes its own standards, trains its own fellows, credentials its own experts, and testifies to the authority of all three.

The conflict is not that these physicians are corrupt. It is that they occupy incompatible roles simultaneously: treating clinician, forensic examiner, investigative partner, contracted vendor, and expert witness — with the same encounter serving all five functions and only the first disclosed to the family.

The Cases: What It Looks Like When the Diagnosis Is Wrong

Kowalski v. Johns Hopkins All Children's Hospital. Ten-year-old Maya Kowalski, who had a complex regional pain syndrome diagnosis, was held at the hospital for months after Dr. Sally Smith concluded her mother Beata — a nurse — was perpetrating medical child abuse. Beata Kowalski died by suicide in January 2017. The family sued in October 2018; Suncoast Center settled for $2.5 million in January 2022; in November 2023 a jury returned a $261 million verdict against the hospital on seven counts including false imprisonment, medical negligence, battery, fraud, and intentional infliction of emotional distress, later reduced to roughly $208 million. On October 29, 2025, Florida's Second District Court of Appeal reversed the judgment in a nearly 50-page opinion, holding the trial court committed major legal errors in applying Florida's good-faith reporter immunity and that the hospital "acted in good faith in its participation with child protection activities." Mediation failed in December 2025. The largest verdict ever returned against a children's hospital in a wrongful-abuse-allegation case was erased by the immunity statute Congress and the states built to encourage reporting.

Krueger v. Petrak (C.D. Ill. No. 1:22-cv-01016). Jacob and Patti Krueger were accused of fabricating their middle son's medical condition. The child has genetically confirmed Xia-Gibbs Syndrome. Their three sons spent 467 days in foster care before an Illinois judge vacated all abuse findings in July 2020 and ordered them home. Judge Joe B. McDade denied child abuse pediatrician Dr. Channing Petrak's motion to dismiss, holding she was not entitled to qualified immunity for allegedly causing the removals in violation of the family's Fourth and Fourteenth Amendment rights — one of the few rulings anywhere to pierce the protective layer around a CAP's investigative conduct.

Bourne and Troy v. Texas DFPS and Ascension Health (W.D. Tex., filed April 2024). Lorina Bourne and Jason Troy allege their five-month-old was assessed at Dell Children's Medical Center as a shaken baby case when he presented with an enlarging head circumference and vomiting requiring surgical drainage. The suit alleges the hospital where he was born, Ascension Seton Medical Center Austin, never disclosed a birth injury, and that reviewing the birth records would have contradicted the shaken baby conclusion. Two children spent 150 days in foster care. The named defendant hospital, Dell Children's, is an FACN subcontractor.

And the scale is not anecdotal. After the Chronicle/NBC series ran, more than 300 families from 38 states came forward describing children removed on medical opinions they contended were wrong.

The Science Under the Testimony

The most contested diagnosis in the field is abusive head trauma, formerly shaken baby syndrome. The National Registry of Exonerations has documented 41 exonerations since 1992 in SBS/AHT cases; in eight, the prosecution's own experts publicly disavowed their prior testimony, and courts described those recantations as decisive. A disproportionate share of exonerees — 15 of 40 in one tabulation — are women.

Courts are moving, unevenly. In September 2024 the Texas Court of Criminal Appeals granted Andrew Roark a new trial, finding the shaken baby hypothesis underlying his Dallas conviction lacked scientific validity. Robert Roberson, convicted in 2003 of killing his two-year-old daughter on an SBS theory, has twice had his execution halted — first by a 5–4 Court of Criminal Appeals ruling invoking Texas's 2013 "junk science" statute, and again in late 2025 — with the trial court taking up a new-trial question in 2026 and more than 30 international medical and scientific experts petitioning the Board of Pardons and Paroles. Massachusetts' Supreme Judicial Court has addressed the evidentiary question directly: after a Daubert-equivalent hearing, experts were permitted to testify that findings were "consistent with" abusive head trauma but barred from testifying that AHT "was the cause."

That is the exception. In juvenile dependency court — where the removals actually happen — Daubert and Frye challenges are rare to nonexistent. Dependency proceedings are civil, often closed, frequently conducted on a preponderance standard, with parents represented by overloaded appointed counsel who lack the budget to retain a rebuttal expert, let alone fund a full admissibility hearing. The testimony that gets rigorously tested in a capital case is waved through in the proceeding that separates a family.

There is also a documented equity problem baked into the front end. Minority children aged 12 months to 3 years have been found significantly more likely to receive a skeletal survey than White children after controlling for insurance status and abuse likelihood (adjusted odds ratio 8.75). Suspected-abuse reports were filed for 22.5 percent of White children versus 52.9 percent of minority children in one cohort. Yet among infants evaluated with a skeletal survey, the adjusted probability of an abuse diagnosis was higher for White infants (61 percent) than Black (51 percent) or Hispanic (53 percent) infants — and in known head trauma cases, abuse was missed in 37 percent of White children versus 19 percent of non-White children. The screen is simultaneously over-applied and under-accurate, in opposite directions, by race.

The Accountability Gap: Immunity Flows Up, Data Never Flows Back

Every actor in the chain has a reason to believe someone else is watching. The hospital believes the agency validates the finding. The agency believes the physician's expertise validates the finding. The juvenile court believes the agency has vetted the physician. The federal government believes NCANDS captures the outcome. None of this is true.

There is no accuracy standard. There is no error rate published by any state for CAP determinations. There is no requirement that a juvenile court be told the diagnosing physician's program holds a contract with the petitioning agency, or that the physician's salary is partly agency-funded. There is no mandatory second opinion in most states. There is no federal reporting line for reversed or vacated abuse findings. And CAPTA's good-faith immunity — the same provision the Second DCA used to erase a $208 million verdict — means that even a catastrophic error generally produces no liability, no license action, and therefore no data point that would ever surface in an audit.

Two states have begun to move. Texas's SB 1578, authored by Sen. Lois Kolkhorst and effective September 1, 2021, requires DFPS to refer a child for a second medical opinion from a specialist not involved in the original report when parents, caseworkers, or treating physicians request one, and bars removal based solely on the opinion of a child abuse pediatrician who never physically examined the child; companion HB 2536 bars removal or termination premised on a parent seeking a second opinion or changing providers. Illinois went further: the Protecting Innocent Families Act, HB 3169, passed 110–0 in the House and 59–0 in the Senate and was signed by Gov. JB Pritzker in July 2026, making Illinois the first state to require child abuse pediatricians and their clinical teams to disclose their identity, specialty, and forensic purpose to families; to require DCFS to notify parents separately when it seeks a CAP opinion; and to give parents the right to obtain the written opinion submitted against them. Florida's "Patterson's Law" (SB 304, Sen. Barbara Sharief) — named for parents who lost custody of three sons in 2022 over an undiagnosed medical condition — passed the Senate 37–0 in 2025 and then died in messages.

Two states out of fifty, one of them only weeks old.

Why It Matters, and What Would Actually Fix It

The stakes are not abstract. A wrongful abuse finding costs a family 150 days, or 467 days, or a mother's life. It places a medically fragile child — often a child whose real, undiagnosed condition produced the findings in the first place — into a foster placement with no access to the specialists who were treating them. It generates a criminal referral. It lands a parent on a central registry that can end a career in nursing, teaching, or childcare, frequently with no hearing that meets ordinary due process standards. And because the misdiagnosis loop is never closed, the same interpretive error propagates into the next case.

Four changes would close most of the gap, and none of them requires disbelieving that child abuse is real or that these physicians save lives — they do, and the field exists because injuries were being missed.

One: mandatory disclosure, everywhere. Illinois's HB 3169 should be the national floor. A forensic examination should be identified as forensic, in writing, before it begins.

Two: financial disclosure to the court. Any contract, subcontract, or salary support flowing from a child welfare agency to a testifying physician's program should be disclosed on the record in every dependency and criminal proceeding, exactly as expert compensation is disclosed in ordinary civil litigation.

Three: a real second opinion, funded. Texas's SB 1578 model — an independent specialist in the underlying condition, not a second child abuse pediatrician applying the same framework — should attach before a removal petition, not after, and should be funded so that indigent parents can actually invoke it.

Four: make CAPTA money contingent on outcome data. Congress should condition § 5106a state grants on reporting, through NCANDS and AFCARS, whether a hospital-based child protection consult preceded a removal, whether the physician's program was agency-funded, and whether the finding was subsequently reversed, vacated, or expunged. The Stronger CAPTA Act already contained the data-infrastructure architecture; it needs the medical-determination fields added and a Senate vote.

The federal government spends $212 million a year on CAPTA and more than $11 billion across child welfare authorities, funds the multidisciplinary teams by statute, immunizes their members by statute, and then declines to ask the one question that would tell anyone whether the system works: how often is it wrong? Until that question is asked with money attached to the answer, the hospital will remain the front door to foster care — and no one will be counting who walks through it.

Sources

A note on two figures in the directive: the "roughly 375 board-certified physicians" count sits between the two published numbers — 342 board-certified and aged ≤70 as of June 2023 (the practicing-workforce measure used in Pediatrics) and 425 ever certified as of that same date. And "admitted nationwide without a Daubert challenge" is close but not absolute: challenges are vanishingly rare in dependency court, but Massachusetts and Texas have both constrained this testimony in criminal proceedings, and those rulings are the strongest available precedent for the reform argument. I wrote both accordingly rather than overstating them.