OSINT · Publicly Available Sources INVESTIGATION ORACLE

The Paper Cure: How "Qualified Residential Treatment Programs" Let States Keep Billing Washington for the Same Group Homes

September 11, 2026 OPUS · Claude Opus Project Milk Carton

The Paper Cure: How "Qualified Residential Treatment Programs" Let States Keep Billing Washington for the Same Group Homes

Congress passed the Family First Prevention Services Act in 2018 to shut off the federal money that had made warehousing foster children profitable. Eight years later, the federal government still cannot say whether a single one of the roughly 40,000 children a year placed in American group facilities is actually receiving the trauma-informed treatment, 24/7 clinical staffing, family outreach, and six-month aftercare that the law made the price of continued federal payment. The Government Accountability Office reported in March 2026 that 26 of 49 surveyed states have seen congregate care hold steady or rise since the law took full effect, that three of four states it visited simply swapped in state dollars to keep the beds full, and that at least one state now classifies between 81 and 95 percent of the youth in its federally funded congregate placements as sex-trafficking victims or at risk of becoming them — a category Congress never defined and HHS has never audited. Meanwhile the prevention half of the same statute, the half that was supposed to keep children home, drew $173 million while foster care drew $5.1 billion. The reform did not fail. It was converted into paperwork.

The Setup: A Fourteen-Day Clock and Four Exits

The Family First Prevention Services Act was enacted February 9, 2018 as part of the Bipartisan Budget Act (P.L. 115-123). Its central lever is codified at 42 U.S.C. § 672(k): after October 1, 2019 — a deadline most states delayed by the permitted two years, to October 1, 2021 — a state may claim federal Title IV-E foster care maintenance payments for a child in a non-family setting for no more than 14 days.

Then come the exits. Section 672(k)(2) preserves federal payment past 14 days for four categories: supervised independent living for youth 18 and older; settings for pregnant and parenting youth; settings providing "high-quality residential care and supportive services" to children who are, or are at risk of becoming, sex-trafficking victims; and the marquee category, the Qualified Residential Treatment Program.

Section 672(k)(4) tells you what a QRTP is supposed to be. It must use a trauma-informed treatment model designed for children with serious emotional or behavioral disorders. It must have registered or licensed nursing staff and other licensed clinical staff who work within their scope of practice, are on-site according to that treatment model, and are available 24 hours a day, seven days a week. It must facilitate family participation in treatment, conduct documented outreach to biological family and fictive kin including siblings, and maintain contact information for them. It must provide discharge planning and family-based aftercare support for at least six months after discharge. And it must be licensed and accredited by CARF, the Joint Commission, the Council on Accreditation, or another nonprofit accreditor the Secretary approves.

Section 675a adds two procedural gates. Within 30 days of placement, a "qualified individual" must assess the child using an age-appropriate, evidence-based, validated functional assessment tool, determine whether the child's needs can be met in a family home, and if not, identify the least restrictive setting that can meet them. Within 60 days, a court must review that assessment and approve or disapprove the placement.

Read together, those provisions are a serious statute. Read as a compliance checklist — which is how they are administered — they are four documents: an assessment, a court order, a license, and an accreditation certificate. Nothing in the federal claiming process requires a state to demonstrate that the nurse was on shift, that anyone called the child's grandmother, or that a single day of the six-month aftercare was delivered.

The Money: A Small Federal Slice, Enormous Per-Child Stakes

The federal dollars directly at risk under the 14-day rule are modest. Federal congregate-care claiming has been reported at roughly $350 million annually, up about 25 percent between 2020 and 2023 even as the overall foster-care population shrank. Against a $5.1 billion federal Title IV-E foster care draw, that is small change.

The per-child economics are not. Approved QRTP per diems run from roughly $275 to more than $800 per child per day depending on state and facility. Minnesota approved a per diem of $433.82 for one out-of-state QRTP; Ohio caps certain Title IV-E residential reimbursement at roughly $200 per day; Colorado publishes a separate QRTP fee schedule through its Medicaid agency. At $433 a day, a single bed generates about $158,000 a year. A 40-bed campus generates $6.3 million. Losing QRTP designation does not cost a provider a rounding error; it costs the business.

And the statute quietly prices the alternative lower. Foster care maintenance payments are matched at each state's Medicaid FMAP — for FY2026, from 50 percent in California and New York to roughly 77 percent in Mississippi. Title IV-E prevention services are matched at a flat 50 percent. For a low-income state, the federal government pays substantially more to place a child than to prevent the placement. That is not a loophole. That is the arithmetic Congress wrote.

The Assessor Problem: Independence, Contracted Out

Section 675a(c)(1)(D) requires the qualified individual to be "a trained professional or licensed clinician who is not an employee of the State agency and who is not connected to, or affiliated with, any placement setting in which children are placed by the State." Congress understood the conflict perfectly. It then attached a waiver: HHS may relieve a state of the independence requirement if the state demonstrates its process maintains objectivity. Minnesota, for example, obtained two Children's Bureau waivers permitting designated county and tribal agency employees to serve as qualified individuals.

Where the function is genuinely outsourced, it is outsourced to a small set of vendors. Maximus Inc. (NYSE: MMS), the publicly traded government-services contractor, announced in July 2021 that it had won QRTP assessment contracts in Indiana, Michigan, North Dakota, and New Hampshire. Maximus Clinical Services runs Indiana DCS's 30-day assessment program and Michigan MDHHS's independent assessments, deploying assessors and clinical supervisors who use the CANS (Child and Adolescent Needs and Strengths) tool.

This is the structural defect. The "independent" assessor is a vendor whose contract is written, funded, scored, and renewed by the same agency that placed the child and that will pay the per diem the assessment authorizes. The vendor is independent of the facility, which is what the statute literally requires, and dependent on the agency, which is what actually matters. Chapin Hall's July 2024 policy brief examining the QI role across Colorado, Illinois, Kentucky, Minnesota, New York City, Utah, and Virginia documents how differently seven jurisdictions have staffed and monitored this role — which is another way of saying there is no national standard for what independence means in practice, and no federal collection of how often a qualified individual actually recommends against the placement the agency already made.

The Court Gate: Sixty Days, No Scoreboard

The 60-day judicial review was Congress's backstop. It has never been measured. AFCARS — the Adoption and Foster Care Analysis and Reporting System, the federal child-level data collection — does not capture whether a court approved or disapproved a QRTP placement, or on what record. No federal report publishes disapproval rates. Chapin Hall's implementation research found that all five studied jurisdictions relied on "early engagement and communication with their judicial partners" to hit the deadline — the operative verb being hit the deadline, not test the placement.

The practical posture at a 60-day hearing is a child already living at the facility, an agency that has no alternative bed, a qualified individual's written assessment endorsing the placement, and a judge whose disapproval creates an immediate placement crisis with no remedy attached. Nothing in the record establishes that judges are rubber-stamping; nothing in the federal data would reveal it if they were. That absence is itself the finding.

What GAO Actually Found

GAO-26-107592, Child Welfare: HHS Should Clarify Guidance on State Spending for Congregate Care, published March 3, 2026, surveyed 49 states in the first half of 2025. Its findings:

26 of 49 states reported the share of foster youth in congregate care had increased or stayed the same compared to October 2021. The same number reported average length of stay in congregate care flat or rising.
Three of four states GAO visited reported using state, county, or local funds to sustain congregate placements after federal funds were cut off. The placement did not end. The funding source rotated.
The sex-trafficking exemption is undefined in federal law or regulation. Some states read it to cover any youth in a facility that serves trafficking-affected youth; others require individualized findings. At least one state classifies 81 to 95 percent of youth in its federally funded congregate placements as possible or confirmed trafficking victims. In other states the figure is under 40 percent.
ACF's own reviews caught a state claiming Title IV-E past 14 days in five non-exempt congregate cases, and another claiming funds for a child in a QRTP for more than a year without the state child welfare director's required approval letter to HHS.
25 state child welfare agencies and 26 state juvenile justice agencies did not know whether the share of dually involved youth in juvenile justice placements had risen since Family First. Of the 20 states that did know, 10 reported an increase in dually involved youth in detention since October 1, 2021.

GAO made one recommendation: that HHS clarify what "found to be, or at risk of becoming, a sex trafficking victim" means. Not that HHS verify QRTP designations. Not that HHS audit aftercare. One definitional fix, aimed at the exit ramp that doesn't even require a QRTP.

The Designation Nobody Checks

Independent research reaches the same place from the other direction. A study published in Pediatrics in July 2024 by researchers working with the American Academy of Pediatrics and Chapin Hall surveyed state child welfare directors and got responses from 47 states (90 percent). Thirty-four states had at least some QRTPs — and half of those 34 reported that 50 percent or less of their congregate facilities had achieved QRTP status. Thirty states called the QRTP requirements challenging to meet. The single requirement they identified as hardest was family-based aftercare for six months post-discharge — the one element that costs money after the child has stopped generating a per diem. The AAP's companion commentary concluded that "the use of QRTP as an individualized and quality treatment intervention, as opposed to a standardized placement, has yet to be realized."

So: the hardest requirement to meet is the one with no revenue attached, and no federal process verifies it. Children's Bureau Title IV-E eligibility reviews examine child and provider case records and payment documentation to validate the accuracy of reimbursement claims — eligibility and paperwork. They are not designed to establish whether the trauma-informed model described in the licensing binder is the model the child experienced, whether licensed nursing was genuinely available around the clock, whether anyone made the documented family outreach, or whether aftercare occurred. GAO's own examples of ACF findings are documentary: a missing director's letter, a miscounted 14 days.

The Safety Record Underneath

The oversight vacuum is not theoretical. HHS's Office of Inspector General reported in 2024, in Many States Lack Information To Monitor Maltreatment in Residential Facilities for Children in Foster Care, that nearly one-third of states could not identify patterns of maltreatment within facilities in their own state; that states had limited visibility into abuse across multi-state chains owned by the same company; and that 13 states did not consistently report to the national maltreatment database whether a maltreated child was living in a residential facility. GAO documented parallel failures in GAO-24-107625 on abuse of youth in residential facilities.

The Senate Finance Committee's June 12, 2024 report, Warehouses of Neglect: How Taxpayers are Funding Systemic Abuse in Youth Residential Treatment Facilities, produced by a two-year investigation with the HELP Committee under then-Chairman Ron Wyden, examined four providers: Universal Health Services, Acadia Healthcare, Devereux Advanced Behavioral Health, and Vivant Behavioral Healthcare (the successor to Sequel Youth & Family Services). It found physical, sexual, and verbal abuse; improper and overused restraint and seclusion; unsafe conditions; and inadequate treatment — harms the committee described as inherent to a model that maximizes census while minimizing cost. At the time, Acadia and UHS carried valuations of roughly $6.5 billion and $11.8 billion; Medicaid supplied more than half of Acadia's 2023 revenue and more than a quarter of UHS's. Wyden released follow-up findings on LGBTQIA+ youth in these settings on July 31, 2025, announced a reform package in December 2025, referred potential Medicaid fraud and civil rights violations to the Department of Justice, and in May 2026 expanded the investigation to federally contracted facilities for unaccompanied children.

The nonprofit side of this industry is equally large. The Devereux Foundation (EIN 23-1390618, Villanova, Pennsylvania) reported FY2023 revenue of $509.3 million, expenses of $521.4 million, and $252.2 million in assets, with roughly 9,300 employees and aggregate officer, director, and trustee compensation of about $5.96 million. Federal award records traced through Project Milk Carton's database show Devereux receiving more than $360 million in federal money, including $48.2 million under CFDA 93.566 in 2022 and a decade of ACF Unaccompanied Children Program grants running $8.1 million to $13.8 million per year. Devereux remains in active litigation over child safety, including D.L. v. Community Based Care of Brevard, Inc. d/b/a Brevard Family Partnership, The Devereux Foundation, Inc., decided by Florida's Fifth District Court of Appeal on February 13, 2026 (Nos. 5D2024-2975, 5D2024-3492).

Litigation is where these placements surface when oversight doesn't. In D.S. v. Washington State Department of Children, Youth, and Families, No. 2:21-cv-00113 (W.D. Wash.), the certified class expressly included children who had been in a QRTP for a year or more — a fact that should be impossible under a statute built around time-limited, clinically necessary stays. A New Mexico jury returned a $485 million verdict in July 2023 for a child sexually assaulted after a placement agency put her with a foster parent already accused of assault. Oregon settled foster abuse claims for $40 million in December 2023, with the Ninth Circuit expanding the settlement class in August 2025.

The Other Half of the Bargain

Family First's prevention title was the trade Congress offered: restrict congregate care, and in exchange draw open-ended federal match for evidence-based prevention services for candidates for foster care. The trade was never consummated.

ASPE's brief by Sarah Oberlander, Katie Allen, and Amanda Benton found that in FY2023, prevention services amounted to less than two percent of all Title IV-E reimbursement claims, and that 60 percent of states, territories, and tribes had never submitted a single prevention claim. As of 2025, 36 jurisdictions were claiming prevention funds at all; the most recent expenditure data shows $173 million in prevention draw against $5.1 billion for foster care.

The Prevention Services Clearinghouse had reviewed 210 programs as of July 2025, rating 95 as promising, supported, or well-supported. As of December 2025, only 28 percent of those 95 appeared in any approved state, territorial, or tribal prevention plan. And the evaluation requirement — the mechanism that would prove any of this works — is routinely waived. The Children's Bureau may waive rigorous evaluation for well-supported practices if a state meets continuous quality improvement conditions. California requested and received waivers for all ten of the evidence-based practices in its five-year plan; Nebraska requested waivers for seven of nine. What remains is fidelity monitoring by the same agencies claiming the match.

Child-level prevention data does exist on paper: ACF's Technical Bulletin #1 requires jurisdictions to submit records for each child receiving Title IV-E prevention services, using AFCARS record numbers where assigned and newly created identifiers for candidates who have none. But a submitted service record is not a delivered service, and the waiver structure means no independent outcome study is required to keep the money flowing.

Why It Matters, and What Would Actually Fix It

Congregate care's share of the foster population fell from 13 percent in 2015 to 9 percent in 2021 — then climbed back to 11 percent by 2024. The count of 13-to-17-year-olds in congregate care declined only modestly, from 27,500 in 2023 to 26,100 in 2024. Roughly 40,000 children a year still live in group facilities, a number GAO describes as essentially steady since 2021. These are the children with the worst measured outcomes in the system: lower educational attainment, higher rates of running, higher rates of justice-system contact, and — per GAO's dually involved findings — increasingly likely to be absorbed into juvenile detention when the child welfare bed disappears.

Four changes would close the gap, and none requires reopening the underlying bargain:

Audit the designation, not the paperwork. Title IV-E eligibility reviews should pull a random sample of QRTP placements and verify the four substantive elements — treatment model in practice, nursing availability, documented family outreach, and delivered aftercare — against payroll, shift logs, contact records, and post-discharge service records. A license is a prediction. A shift log is evidence.
Make the extended claim contingent on delivered aftercare. Condition the final tranche of a QRTP episode's federal match on documentation that six months of family-based aftercare was actually provided. States call this requirement the hardest to meet precisely because nothing currently depends on meeting it.
Put the gates in AFCARS. Report, per child: qualified-individual identity and independence status (including whether a § 675a waiver applies), assessment date, whether the assessment recommended the QRTP, and the court's approval or disapproval. Until disapproval rates are public, no one can tell oversight from ceremony.
Define the trafficking exemption and close the FMAP gap. GAO's single recommendation is correct and should be implemented immediately — a state classifying 81 to 95 percent of its congregate population as trafficking-affected is either facing an emergency that warrants federal intervention or is coding for reimbursement. And Congress should match prevention at each state's FMAP. As long as Washington pays Mississippi 77 cents to place a child and 50 cents to keep her home, no amount of clarifying guidance will change what states choose.

Family First was written on the premise that money drives placement. That premise was correct. The mistake was assuming that restricting the money would restrict the placement, rather than restricting only the vocabulary used to describe it.

Sources: