Paid by the Agency They Report To: Inside the Hospital "Child Abuse" Teams That Remove Children on One Doctor's Word — With No Second Opinion, No Disclosure, and No Error Count

A child arrives in an emergency department with a fracture, a bruise, or a subdural hematoma. Within hours, a hospital-based child abuse pediatrician — a member of a 330-person subspecialty, frequently working under a contract that pays the hospital with money from the very child protection agency that will receive the referral — renders an opinion. That opinion becomes an affidavit, the affidavit becomes an ex parte removal order, and the child is gone. Federal award records, state contracts, and a decade of litigation show that this pipeline operates with no default right to a second opinion in 48 states, no requirement that the physician's financial relationship to the referring agency be disclosed to the judge or the parents, and — most remarkably — no federal or state database that counts how many children are removed on a single physician's diagnosis, or how often that diagnosis is later reversed when the "abuse" turns out to be rickets, Ehlers-Danlos syndrome, metabolic bone disease of infancy, or a birth injury.

The Setup: A Subspecialty Built to Answer One Question

Child abuse pediatrics (CAP) was approved as a subspecialty by the American Board of Medical Specialties in 2006, and the American Board of Pediatrics issued its first certifying exams in 2009. It is the smallest pediatric subspecialty in America. Published workforce analysis in Pediatrics counted 330 active board-certified CAP subspecialists nationwide in 2020, projected to reach only about 400 by 2040, with roughly ten fellows matching per year and one of the oldest certified workforces in pediatrics.

That tiny cohort sits at a structural choke point. A national analysis of NCANDS data covering 2014–2023 identified 285,329 report-child pairs involving physical abuse concerns raised by a medical professional. Fifty-one percent involved infants under one year old. Physical abuse was substantiated in 31% of cases with an available determination — meaning roughly 69% were not substantiated — yet 18% of those children entered foster care as a result of the investigation.

The referral question posed to a CAP is functionally binary: abuse, or not abuse. But the underlying medicine is not binary, and the field's own literature says so. A study of CPS-initiated child abuse consultations published in Academic Pediatrics found child abuse experts identified definite maltreatment in 40.6% of cases, while the initial treating physicians had identified it in only 21.4% — and that agreement between child abuse experts and either the referring physician or CPS was "fair to poor." In the same consultation series, expert opinions broke down as 47.6% definite or probable maltreatment, 8.6% uncertain, and 43.9% definite or probable benign. Nearly half the cases sent up as suspected abuse came back the other way. That is a diagnostic domain with enormous interpretive latitude — and it is being used to generate court affidavits treated as near-dispositive.

The Money: Who Pays the Diagnosing Doctor

Federal child welfare funding across Title IV-E, Title IV-B, CAPTA, Adoption Opportunities, and the Victims of Child Abuse Act totaled $11.319 billion in FY2025. Title IV-E alone accounted for $9.5 billion in FY2023, including $5.1 billion reimbursing states for foster care. CAPTA funding for states runs about $212 million, of which roughly $105 million is state grants. The Children's Justice Act has distributed $17 million annually since FY2000 out of the federal Crime Victims Fund — $50,000 per state plus a child-population formula — explicitly to "improve the investigation, prosecution, and judicial handling" of abuse cases.

Layered on top is the Children's Advocacy Center architecture. Federal award records retrieved through PMC's CivicOps database show the National Children's Alliance — the accrediting body for 900-plus CACs serving roughly 370,000 children a year — has received approximately $279.5 million cumulatively from the U.S. Department of Justice, overwhelmingly under CFDA 16.758 (Children's Advocacy Centers), including $52.2 million obligated in FY2025, $63.6 million in FY2022, and $36.1 million in FY2019, plus separate awards under CFDA 16.834 (Victims of Child Abuse) of $6 million in FY2022 and $4 million in FY2025. Most of that is passed through to local centers. The point is not that the money is stolen; the point is that it builds a standing, co-located, multidisciplinary referral machine whose medical arm answers to the same funding stream as its investigative arm.

Then there are direct state-to-hospital contracts, which is where the conflict becomes explicit:

• Texas. The Department of Family and Protective Services contracts with UTHealth Houston (McGovern Medical School) to operate the Forensic Assessment Center Network (FACN) — the state's on-call panel of child abuse pediatricians. The 2019–2024 contract term governed the period covered by the NBC News/Houston Chronicle investigation, which reported that FACN network doctors receive as much as 60% of their salaries from the child welfare agency, consulting on abuse cases and testifying in court when children are taken into foster care.
• New Hampshire. In June 2024, the Governor and Executive Council approved a contract amendment increasing the price limitation for the state Division for Children, Youth and Families' medical-evaluation contract by $1,499,410 — from $2,998,820 to $4,498,230 — extending through June 30, 2026. The contractor is the hospital-based Child Advocacy and Protection Program at Dartmouth Health, which "provides DCYF with medical opinions" and also trains DCYF's own child protective service workers. The funding is 92% state general funds, 8% federal.
• Medicaid. Effective October 1, 2025 for fee-for-service and December 1, 2025 for managed care, New York State Medicaid began reimbursing physicians board-certified in Child Abuse Pediatrics a once-per-member-lifetime fee for diagnosis and treatment of suspected maltreatment — a dedicated payment channel keyed to the credential itself.

Why the contracts exist at all is candidly documented in the field's own business literature: CAP divisions are chronic money-losers for hospitals. Lurie Children's states publicly that it relies on philanthropic funding to sustain its child abuse pediatrics program. Peer-reviewed work on child abuse team organization frames "a sustainable business model" as the central operational problem. The state contract is the sustainable business model. The agency that receives the referral is underwriting the department that generates it.

Named Players and the Conflict Nobody Has to Disclose

Dr. Sally Smith was medical director of the Pinellas County (Florida) Child Protection Team and held privileges at Johns Hopkins All Children's Hospital when she was asked in October 2016 to opine on whether 10-year-old Maya Kowalski's complex regional pain syndrome was actually Munchausen syndrome by proxy. Maya was held for three months. Her mother, Beata Kowalski, died by suicide in January 2017. Smith settled individually for $2.5 million. On November 9, 2023, a Sarasota County jury awarded the family $261 million, later reduced to roughly $213.5 million. On October 29, 2025, Florida's Second District Court of Appeal reversed the entire judgment and ordered a new trial, holding the trial court "misinterpreted and misapplied section 39.203(1)(a), Florida Statutes, which provides immunity for good-faith child-abuse reporting." The family has petitioned the Florida Supreme Court; retrial is tentatively set for 2027.

Dr. Barbara Knox led the child protection program at the University of Wisconsin's American Family Children's Hospital until she was placed on paid leave amid an investigation into allegations she intimidated colleagues who disagreed with her diagnoses. She became medical director of Alaska CARES in 2019. By fall 2021, the entire medical staff of that clinic had resigned; she resigned in January 2022 and moved to Florida. Wisconsin Watch documented a dozen instances in which her abuse diagnoses were rejected by courts, child welfare workers, or other physicians. In July 2022, two Alaska families filed a federal suit naming Knox, her supervisor, and Providence Alaska Medical Center; two Wisconsin couples sued the following month.

Dr. Elizabeth Woods served for more than three years as a go-to child abuse expert for Washington State authorities from Mary Bridge Children's Hospital in Tacoma. An NBC News/KING 5 investigation established that she never completed the three-year CAP fellowship and was never board certified — and that under oath in 2019 she claimed most child abuse pediatricians lacked board certification and that the fellowship program had existed for only three years. Both statements were false. She was removed from the state's expert roster after journalists, not regulators, surfaced the problem.

Dr. Debra Esernio-Jenssen of Lehigh Valley Health Network's John Van Brakle Child Advocacy Center is the lead defendant in a 233-page class action filed in Philadelphia County Court of Common Pleas in February 2024 by attorney Francis Malofiy, alleging 17 causes of action including corporate negligence, interference with parental rights, and fraud, on behalf of families alleging systematic over-diagnosis of Munchausen syndrome by proxy. Lehigh County Controller Mark Pinsley had already reported that Lehigh and Northampton counties accounted for one-third of all MSBP cases in Pennsylvania from 2017 to 2021. Esernio-Jenssen retired effective March 31, 2024.

Dr. Suzanne Dakil signed the affidavit that led armed officers to remove four-year-old Drake Pardo from his Kaufman County, Texas home in June 2019 — after his parents complained about a physician and sought a second opinion. Dakil later testified she had never examined Drake, never met the family, did not believe the case was an emergency, and did not recommend removal.

Dr. Sandeep Narang, division head of child abuse pediatrics at Lurie Children's, co-authored both the AAP's 2020 policy statement on abusive head trauma and its 2017 policy on expert witness participation — and has served as a paid expert witness and consultant in abusive head trauma litigation. That is the structural problem in miniature: a professional community of a few hundred people writes the diagnostic standard, testifies to its validity, and is compensated for applying it.

None of these arrangements are illegal. What is remarkable is what is not required: no state mandates that a removal affidavit, or the physician's courtroom qualification testimony, disclose the value of the hospital's contract with the referring agency, the share of the division's or physician's compensation derived from it, or cumulative expert-witness income. The judge signing the order and the parent losing the child typically do not know the contract exists.

What the Failures Look Like

Texas publishes the cleanest numbers. In fiscal year 2020, DFPS referred 5,368 of its 154,593 investigations (3%) to the FACN. Of those, the network graded 12% "substantial," 41% "concerning," and 22% "non-specific." Not one of those words is a legal standard. "Concerning" is not clear and convincing evidence — but it lands in an affidavit all the same.

The mimics are well documented in the medical literature: hypophosphatemic rickets associated with elemental formula use producing multiple fractures; metabolic bone disease of infancy; osteogenesis imperfecta; Ehlers-Danlos syndrome; vitamin D and vitamin C deficiency; and birth itself — studies cited by the Family Justice Resource Center report subdural hematoma and retinal hemorrhage in roughly 46% and 45% of vaginal births, respectively, the two findings most often invoked as evidence of shaking.

On the criminal side, the National Registry of Exonerations now counts roughly 40 exonerations in shaken baby syndrome / abusive head trauma cases — 15 of the 40 are women — with the underlying deaths and injuries traced instead to sickle cell disease, stroke, seizure disorder, sepsis, urinary tract infection, hypoxia, and SIDS. In Texas, Robert Roberson came within a week of becoming the first person in the United States executed on a shaken baby conviction; on October 9, 2025, the Texas Court of Criminal Appeals halted the execution and remanded, following its earlier grant of a new trial to Andrew Roark on changed-science grounds.

On the civil and dependency side: the Texas Supreme Court ordered Drake Pardo returned in October 2019 and the case was dismissed that December. A Wayne County, Michigan judge dismissed abuse allegations against the Parker family after eight months, saying from the bench, "I heard a lot deeper science from some of the other witnesses than I heard from Dr. Mohr." A Washington judge cleared an Aberdeen mother after her infant spent months in DCYF custody and specifically faulted the agency for delaying a second opinion. Two Children's Wisconsin physicians, Dr. John Cox and Dr. Sarah Dobrozsi, sued their own hospital and colleagues — including CAP Dr. Lynn Sheets — in federal court after their hospital's child abuse team classified their infant as an abuse victim (Cox v. Medical College of Wisconsin, 651 F. Supp. 3d 965 (E.D. Wis. 2023)).

The scale hint came in 2019: after NBC News and the Houston Chronicle published "Do No Harm" — nine months of reporting, more than 40 cases examined, more than 100 attorneys, doctors, and current and former state employees interviewed — more than 300 families from 38 states contacted the newsrooms with their own accounts. Reporters documented physicians describing conclusions that cannot be proven as "100 percent" and "certain."

The Accountability Gap: Who Is Supposed to Be Watching

Second opinions are the exception, not the rule. Texas enacted reform in 2021 (H.B. 2536, 87th Legislature; the companion S.B. 1578 ordered the state commission study of FACN), creating a right to additional medical opinions including in-person examination. Florida followed only this year: CS/CS/HB 47, "Patterson's Law," signed June 25, 2026 (Ch. 2026-165), effective July 1, 2026, gives a parent five days after an initial exam to request a second examination or a compilation of differential diagnoses, and explicitly names rickets, Ehlers-Danlos syndrome, osteogenesis imperfecta, and vitamin D deficiency. It was inspired by Michael and Tasha Patterson, who lost custody of eight-week-old twins in 2022 to an undiagnosed case of Ehlers-Danlos. Two states. Both remedies are opt-in windows that open after the child is already in state custody.

Nobody counts the errors. NCANDS records the category of reporter — "medical personnel" — and the case disposition. It does not record the identity of the diagnosing physician, and it contains no field for a diagnosis later reversed. AFCARS records entries and exits from foster care, not the evidentiary basis that produced them. There is no federal registry of removals initiated on a physician's opinion, no state that publishes a CAP-level reversal rate, and no licensing board that tracks reversal as a quality metric. That absence is not incidental — it is what allows a physician whose diagnoses were rejected a dozen times in Wisconsin to become the medical director of Alaska's statewide forensic child abuse center, and then to practice in Florida.

Immunity closes the loop. Every state grants good-faith immunity to abuse reporters; Florida's Second DCA used §39.203(1)(a) to erase a $213.5 million verdict. Federal courts are split on qualified immunity for CAPs: a Tennessee physician was granted it on the reasoning that no clearly established law makes it a Fourth Amendment violation for a physician reporting suspected abuse to omit medical conditions that could mimic abuse; an Indiana federal judge in 2025 extended it to Riley Hospital for Children physicians; a judge in the Central District of Illinois denied it to Dr. Petrak. Courts have found that a private physician can be a state actor for §1983 purposes — and still be immune.

The threshold itself is unevenly applied. The landmark randomized vignette study by Laskey and colleagues sent 5,000 pediatricians identical clinical facts — an 18-month-old with an oblique femur fracture — varying only the child's race and socioeconomic status; abuse was more likely to be diagnosed for lower-socioeconomic-status families. A Children's Hospital of Philadelphia study of 3,063 infants with non-crash traumatic brain injury across 39 pediatric hospitals found Black and publicly insured or uninsured infants significantly more likely to undergo skeletal surveys, with minority children aged one to three nearly nine times more likely to be evaluated for abuse after adjustment — while white children were more likely to actually receive an abuse diagnosis once surveyed. At one institution, a standardized clinical guideline for unwitnessed infant head injury eliminated the disparity entirely, which tells you the disparity was discretion, not disease.

Why It Matters, and What Would Actually Fix It

Infants are killed by caregivers. The subspecialty exists because for decades physicians missed abuse, and children died of injuries someone should have recognized. Nothing here argues for dismantling child abuse pediatrics. What the record argues is that a diagnosis capable of dissolving a family in an afternoon is currently produced under conditions no other high-stakes medical determination would tolerate: single-reader, unblinded, financially entangled with the party that acts on it, and completely unaudited.

Six changes would close the gap without weakening genuine child protection:

1. Mandatory financial disclosure on the face of every affidavit and in every expert qualification: contract value between the hospital and the referring agency, the percentage of the division's and the physician's compensation derived from it, and cumulative expert-witness income.
2. An automatic, agency-funded independent second opinion before any non-emergency removal predicated on a medical opinion — from a physician with no financial relationship to the referring agency, with a documented differential that must address metabolic bone disease, rickets, EDS, osteogenesis imperfecta, coagulopathy, genetic disorders, and birth trauma.
3. Federal error tracking. Add NCANDS/AFCARS fields for physician-initiated removals, the diagnosing physician's NPI, and final case outcome including reversal. CAPTA already conditions roughly $105 million in state grants on state plan requirements; make CAP-level concordance and reversal reporting one of them.
4. Blinded review. The consulting physician should receive imaging, labs, and history — not the caseworker's narrative of what the family is suspected of doing.
5. Immunity reform. Good-faith immunity should not shield a physician who omits a documented differential from a sworn affidavit, or who opines on a child they have never examined.
6. A national, public registry of adverse judicial and administrative findings against child abuse pediatricians, so a record follows a physician across state lines instead of evaporating at the border.

The most damning fact in this entire investigation is the simplest one. After more than 300 families in 38 states came forward, after 40 exonerations, after a $261 million verdict and a class action and a stayed execution, no agency in the United States can tell you how many children were removed last year on the word of a single child abuse pediatrician, or how many of those diagnoses were wrong. Not because the answer is defensible — because nobody is required to look.

Sources: - NBC News — "Do No Harm" investigation · Hundreds of parents say kids wrongly taken · New Texas law · Doctor removed from expert role - The Imprint — New Texas Law Curtails Power of Pediatricians Contracted by CPS - Texas Children's Commission — SB 1578 FACN Report · DFPS FACN Evaluation, April 2022 - Johns Hopkins All Children's Hospital v. Kowalski, Fla. 2d DCA (2025) · WUSF — Appeals court reverses judgment - Wisconsin Watch — Two couples sue former UW child abuse doctor · Knox resigns Alaska position · KTOO — Alaska families file federal lawsuit - WFMZ — 12 new families file class action against LVHN · LehighValleyNews — Pinsley report on systemic misdiagnosis - Florida CS/CS/HB 47 (2026) · Florida Phoenix — DeSantis signs measure - NH DHHS Governor & Council contract item, June 12, 2024 - NYS Medicaid Update, August 2025 — CAP reimbursement - Pediatrics — Child Health Needs and the Child Abuse Pediatrics Workforce: 2020–2040 - Physical abuse of young children reported by medical professionals, 2014–2023 (NCANDS analysis) · Academic Pediatrics — Child Abuse Consultations Initiated by CPS - National Registry of Exonerations · Texas Tribune — Roberson execution blocked - Laskey et al., Influence of Race and Socioeconomic Status on the Diagnosis of Child Abuse (PubMed) · Disparities in Detection of Suspected Child Abuse - Bloomberg Law — No Liability for Tennessee Doctor Who Diagnosed Child Abuse · FamJustice — CAP ruled not to have qualified immunity - Bipartisan Policy Center — Government Spending to Prevent and Respond to Child Abuse · ACF — CAPTA State Grants · ACF — Children's Justice Act - DOJ award records for National Children's Alliance (CFDA 16.758, 16.834) retrieved via PMC CivicOps / USASpending.gov