Notified and Forgotten: How America Logs 44,000 Drug-Exposed Newborns a Year and Writes a Safety Plan for Almost None of Them

Federal law has required since 2016 that every infant born affected by prenatal substance exposure leave the hospital with a written "plan of safe care" addressing both the baby's needs and the caregiver's treatment needs — and that states build monitoring systems to verify the plan is actually followed. In federal fiscal year 2023, an estimated 500,000 American newborns were exposed in utero to alcohol, opioids, cannabis, or stimulants. Child protective services agencies in 49 states logged 44,453 of them. Just 22,319 infants — reported by only 35 of 52 jurisdictions — were recorded as having a plan of safe care at all. That is roughly one documented plan for every 22 exposed infants born in this country, in the single age group that dies of abuse and neglect at 24.11 per 100,000 — more than three times the rate of one-year-olds and eight times the rate of the general child population. The mandate has not failed quietly. It has failed inside a federal reporting system that was built to count it, funded at $105 million a year, and audited by an Inspector General whose completed CAPTA reviews do not contain the phrase "plan of safe care" even once.

The Setup: A Mandate With Three Moving Parts, Two of Which Nobody Checks

The governing text is 42 U.S.C. § 5106a(b)(2)(B), as amended by the Comprehensive Addiction and Recovery Act of 2016 and the SUPPORT Act of 2018. It requires three things of every state accepting a CAPTA grant.

Clause (ii) requires policies for "health care providers involved in the delivery or care of such infants [to] notify the child protective services system of the occurrence of such condition" — followed immediately by the escape hatch that this notification shall not "establish a definition under Federal law of what constitutes child abuse or neglect" or "require prosecution for any illegal action."

Clause (iii) requires "the development of a plan of safe care for the infant … to ensure the safety and well-being of such infant following release from the care of healthcare providers," including by "addressing the health and substance use disorder treatment needs of the infant and affected family or caregiver" — and, critically, "the development and implementation by the State of monitoring systems regarding the implementation of such plans to determine whether and in what manner local entities are providing … referrals to and delivery of appropriate services."

Section 106(d)(18) then requires states to report annually the number of infants identified, the number for whom a plan was developed, and the number for whom a referral was made — but only "to the maximum extent practicable," a phrase that has functioned as a nationwide opt-out.

The Children's Bureau's own binding guidance, Program Instruction ACYF-CB-PI-17-02, is unambiguous: "the development of a Plan of Safe Care is required whether or not the circumstances constitute child maltreatment under state law." The federal data system, by contrast, counts plans of safe care only for screened-in infants. Every infant a state screens out at the hotline — 12,223 of them in FFY 2023, 27.5 percent of the national total — is statistically invisible on the one question the statute actually asks. The mandate and the measurement do not overlap.

The Funnel: 500,000 to 44,453 to 22,319

The numbers come from the Children's Bureau's own Child Maltreatment 2023 report, Tables 3-11 through 3-13.

Nationally: 44,453 infants with prenatal substance exposure referred to CPS across 49 reporting states. 32,230 screened in (72.5%). Of the 35 states that reported at all, 22,319 of 30,956 screened-in infants (72.1%) had a documented plan of safe care. Of the 32 states reporting referrals, 20,468 (68.4%) had a documented service referral.

The state-level spread is where the mandate visibly collapses:

• Oklahoma: 2,209 screened-in exposed infants. 85 plans of safe care — 3.8 percent.
• Colorado: 18 screened in, 639 screened out. One plan of safe care in the entire state.
• Kansas: 43 screened in, 3 plans (7.0%).
• Kentucky: 323 screened in, 67 plans (20.7%).
• Nebraska: 131 screened in, 31 plans (23.7%).
• Utah: 26.7%. West Virginia: 37.8%. California: 39.7%. Washington: 40.1%.
• Minnesota: 998 plans (84.4%) but only 277 service referrals (23.4%) — a plan written for three times as many infants as were connected to anything.
• North Carolina: 1,382 plans (95.0%), 271 referrals (18.6%).

Seventeen jurisdictions — including Florida, Pennsylvania, Illinois, Missouri, Maryland, Mississippi, Montana, South Carolina, Wisconsin, Maine, Hawaii, Alaska, North Dakota, Rhode Island, Vermont, Connecticut and Wyoming — reported no plan-of-safe-care count at all.

Maryland reported 1,815 exposed infants, all screened out, none screened in, and no plans. Maine reported 456 exposed infants, all screened out, none screened in, no plans — in a state whose own data show 4.5 percent of babies born in 2024 were substance-exposed or affected. Pennsylvania, the state whose child-death record is examined below, reported zero exposed infants in either column.

Then there is Michigan, which reported 6,048 screened-in exposed infants and 5,874 plans (97.1%) — more than twice California's identified caseload in a state with roughly a quarter of California's births. The variance is not variance in prenatal substance exposure. It is variance in whether anyone writes anything down.

"Notification Is Not a Report": The Trapdoor Written Into State Law

The federal clause saying notification does not define child abuse was meant to protect mothers from prosecution and to keep them in prenatal care. In practice, states converted it into an administrative off-ramp.

Louisiana's Act 359 of 2017 requires physicians to notify the Department of Children and Family Services on a state form — and states that "such notification shall not constitute a report of child abuse or prenatal neglect." The form is titled Physician Notification of Substance Exposed Newborns, No Prenatal Neglect Suspected**. It is, in the state's own description, "for data-gathering purposes."

Missouri told the federal government flatly in its FFY 2023 commentary: "Newborn Crisis Assessments in Missouri are not considered reports of abuse or neglect and there are no plans, in Missouri, to change the way Newborn Crisis Referrals are categorized."

New York's March 2026 joint ACS/DOHMH clinical guidance is explicit that "Federal and state law does not require that child protective services be involved in the development of a POSC," that "the existence of a POSC does not warrant a report," and that hospitals satisfy the federal requirement by filing an aggregate, de-identified quarterly count through a health-department data system. New York City has roughly 100,000 births a year. New York State reported 215 screened-in exposed infants to NCANDS for the entire state — because, as the state itself explained, its plan-of-safe-care numbers "are limited to those cases in which the report source identified as medical personnel, under reporting the number of children in each category."

The design is not irrational. Every serious clinical body opposes routine drug testing of pregnant patients, and diverting low-risk families away from investigation is defensible policy. But the diversion was built without the second half of the statute. The notification arrives, the hotline logs it, the case is coded as not-a-report, and clause (iii) — the plan, the caregiver's treatment needs, the monitoring system — never attaches to anything. As Cathleen Palm of Pennsylvania's Center for Children's Justice put it: "Federal law and state law are very clear that a baby being born exposed to drugs is not child abuse. … The problem is we as a society have created an infrastructure that is only to deal with child abuse."

The Money: $105 Million a Year, $60 Million "Directed," No Consequence Attached

CAPTA State Grants (CFDA 93.669) are the funding vehicle. HHS-OIG documented $91.6 million awarded across 52 jurisdictions in FY 2022. The Children's Bureau's FY 2025 program instruction states that Congress appropriated $105.09 million for FY 2024 and that "since FY 2018, Congress has directed that $60 million of the annual appropriation for the program be prioritized to help states improve their response to infants affected by substance use disorders."

That direction lives in committee report language, not statute. The Children's Bureau's operative verb is "encourages": "CB encourages states to continue using a substantial portion of their annual grant to strengthen procedures for the development, implementation and monitoring of plans of safe care." Seven consecutive years of a $60 million annual set-aside is roughly $420 million routed toward plans of safe care with no statutory performance condition, no completion floor, and no clawback.

FY 2024 award records from USASpending show where it landed: California Department of Social Services, $11,771,863; Texas Department of Family and Protective Services, $10,325,234; Florida Department of Children and Families, $5,970,800; New York Office of Children and Family Services, $5,547,652; Oklahoma Department of Human Services, $1,363,441. Florida reported no plans of safe care. Oklahoma reported 85.

For scale: $105.09 million spread across 73.3 million American children is $1.43 per child per year — the entire federal investment in child-abuse prevention and CPS improvement. Title IV-E foster care maintenance, by contrast, moves billions. CAPTA itself points states toward Title IV-E to actually pay for plans of safe care, citing § 471(e) Family First prevention services and § 472(j) residential family-based substance abuse treatment. The financial gradient runs toward removal and placement, not toward a home-visit that verifies a mother got into treatment.

Congress also authorized a dedicated program — CAPTA § 106(b)(7), "Grants to States to Improve and Coordinate Their Response to Ensure the Safety, Permanency, and Well-Being of Infants Affected by Substance Use" — with a $500,000 base allotment per state plus a per-live-birth formula, expressly for "developing, updating, implementing, and monitoring plans of safe care." Authorization is not appropriation. CAPTA's own authorization expired at the end of FY 2015; Congress has funded the program without reauthorizing it for a decade, and successive reauthorization bills (H.R. 2480, S. 2971, S. 1927) have died.

The technical-assistance layer is real money too. The National Center on Substance Abuse and Child Welfare — operated by the Center for Children and Family Futures, Inc. — has fielded more than 3,000 technical assistance requests on plans of safe care since CARA passed, convened two Policy and Practice Academies, and visited eight state agencies. Federal award data shows the center receiving a $9.0 million HHS award under CFDA 93.670 (Child Abuse and Neglect Discretionary Activities) plus DOJ drug-court awards of $5.0 million (FY 2022), $4.5 million (FY 2019), $3.95 million (FY 2016) and $3.9 million (FY 2013) — roughly $28 million in traceable federal support. The Regional Partnership Grants program, targeted at exactly this population, has run about $40 million a year since FY 2018 and funded 127 projects across 40 states, capped at $600,000 per grantee per year. Every dollar of it is advisory. None of it carries the authority to compel a state to write a plan.

The Data Is Broken, and the States Said So in Writing

The most damning material in Child Maltreatment 2023 is not the tables. It is Appendix D, where states explain themselves.

Missouri: "During a prior review of reports, it was that noticed staff were not checking the box as they should. … Most staff members said they didn't know what it was for." After training, Missouri's count went from 273 to 933 — a 242 percent jump produced entirely by explaining a checkbox. Missouri still appears as a non-reporting state in the federal plan-of-safe-care table.

Louisiana: "a number of Plan of Safe Care and Referral cases have been dropped as a result of staff not accurately documenting the plan of safe care."

North Dakota: "There were 149 substance exposed infants identified during FFY 2023. Of the 149 identified substance exposed infants, 152 of them had a Plan of Safe Care developed (89%)." That sentence is arithmetically impossible and was published by HHS anyway. North Dakota reports a dash in the federal table.

Connecticut built a hospital portal and found 80.6 percent of notifications had a plan — then noted the fields "have not been incorporated into our legacy SACWIS system," so they never reached NCANDS. Connecticut reports a dash.

South Carolina, in FFY 2023 — seven years after CARA — was still "in the process of finalizing the Plan of Safe Care policy," with a goal "to have this finalized within the next 12 months."

Nevada reported that the relevant data element remains "below the goal of 95%."

The federal statute asks for this data "to the maximum extent practicable." Seventeen jurisdictions have concluded it is not practicable, and nothing happens.

What Happens to the Babies

Children under one are 44.0 percent of all child abuse and neglect fatalities and die at 24.11 per 100,000 — 3.3 times the rate of one-year-olds. Where states reported caregiver risk factors, 20.6 percent of child fatalities involved a caregiver with a documented drug abuse risk factor, versus 4.5 percent for alcohol.

Pennsylvania — which reported zero substance-exposed infants to NCANDS — supplies the ground truth. TribLive analyzed 845 state child death and near-death reports from 2018 to 2022 and found drugs were a factor in nearly one in four incidents: 70 deaths and 170 near deaths. Fifty-six of the 70 drug-related deaths involved infants and toddlers between six days and two years old. Drug-related child deaths outnumbered child gun deaths two to one. Child welfare had already been alerted in more than half the cases.

The individual files read like the plan-of-safe-care window closing on a specific baby: a Franklin County mother charged with involuntary manslaughter after using a heroin-fentanyl mixture and falling asleep on her six-day-old son; a three-month-old who died locked in a car while her mother was unconscious after injecting methamphetamine; four-month-old Naoki Hines, found unresponsive with fentanyl in his system; seven-month-old Zhuri Bogle, a stamp bag of fentanyl in her throat. In Allegheny County, drug ingestions — overwhelmingly opioids — became the leading cause of child fatalities and near-fatalities, accounting for roughly 40 percent of critical incidents in 2022 and nearly half in 2023, mostly in children under three.

West Virginia makes the arithmetic explicit. State surveillance found prenatal substance exposure at 124 per 1,000 births for 2020–2022 — about one in eight infants, with opioid and stimulant exposure rates near ten times national. On roughly 16,000 annual births, that is about 2,000 exposed newborns a year. West Virginia reported 286 screened-in exposed infants and 108 plans of safe care.

The Other Failure Mode: The System Polices What It Won't Plan For

The mirror image of non-planning is over-reporting, and it falls hardest on Black families. The Marshall Project's July 2026 analysis of National Data Archive on Child Abuse and Neglect records found that child welfare agencies in 20 states referred the parents of more than 25,000 Black newborns to police or prosecutors over alleged pregnancy substance use over seven years. In eight states with usable race data, more than 14,000 of those referrals involved cases where child welfare made no finding of abuse or neglect at all. Black babies were roughly two and a half times more likely than White babies to be flagged to law enforcement. In Oklahoma — the state with a 3.8 percent plan-of-safe-care rate — an estimated 1 in 11 Black babies was flagged to law enforcement over pregnancy substance-use claims.

The triggering evidence is frequently worthless. Hospital immunoassay screens produce false positives at rates reported as high as one in two, and as of 2024 not a single state required confirmatory testing before a report to child welfare. A poppy-seed salad produced a positive opiate screen and a newborn's removal. Melissa Robinson, an Alabama school librarian, was barred from breastfeeding and placed under spousal supervision after a false cocaine positive that a second hospital test disproved. At one South Carolina hospital, a Black mother who tested positive for THC in July 2024 was reported and arrested days later on a child abuse charge with no apparent inquiry into cause; two months later a White patient with the same positive result at the same hospital explained a store-bought CBD edible and was not charged. Following the reporting, New Jersey in 2026 began requiring confirmatory testing before results go to child welfare, informed consent for testing, and an anonymized hospital notification form that does not trigger an investigation.

So the system now runs at both extremes simultaneously: tens of thousands of families routed to police on unconfirmed screens, and hundreds of thousands of exposed infants going home with no plan, no referral, and no follow-up.

The Accountability Gap

In January 2018, GAO issued Substance-Affected Infants: Additional Guidance Would Help States Better Implement Protections for Children (GAO-18-196). It found 49 states claiming plan-of-safe-care policies while officials in visited states described not knowing what to put in a plan or what intervention level was appropriate. GAO recommended HHS issue additional guidance. HHS did not concur, arguing states needed "flexibility" to define both which infants are "affected by" substance use and what a "plan of safe care" is. Eight years later, that flexibility produced a range from 3.8 percent to 100 percent and seventeen jurisdictions reporting nothing.

HHS-OIG did eventually open a CAPTA audit series (SRS-A-25-005), announced May 2023, with seven projects and a target completion of FY 2028. Two are done. Maine: 94 percent of sampled child abuse and neglect reports out of compliance with one or more requirements. West Virginia: 91 percent of 100 screened-in family reports out of compliance; all four recommendations remain open and unimplemented. Both audits examined intake, screening, assessment and investigation. Neither examined plans of safe care. The Maine report — the full published audit — does not contain the phrase "plan of safe care" anywhere in it. The only federal audit of CAPTA compliance in the field does not audit the CAPTA provision that governs the highest-fatality-risk age group in America.

Which leaves this as the complete national accountability architecture for clause (iii): a governor's written assurance, a self-reported NCANDS count that seventeen jurisdictions decline to submit, and a peer-technical-assistance center with no enforcement authority. No state has been found out of compliance. No state has lost a CAPTA dollar. There is no federal measure anywhere of whether a plan was written, whether it addressed the caregiver's treatment needs as the statute demands, whether a service referral was accepted, or whether anyone ever looked at that baby again.

What Would Fix It

Five changes, in ascending order of difficulty.

Count every exposed infant, not just screened-in ones. NCANDS should require a plan-of-safe-care and referral disposition for every infant identified under clause (ii), including those screened out and those handled through non-CPS notification pathways. Anything less measures the wrong denominator by design.

Delete "to the maximum extent practicable" from § 106(d)(18) and attach a consequence. A state that cannot report the three counts should face a graduated withholding of its CAPTA State Grant, held in escrow for data-system remediation. Seventeen non-reporters is a policy choice Congress has ratified through inaction.

Convert the $60 million report-language set-aside into a statutory line with a completion standard. Fund CAPTA § 106(b)(7) directly — the $500,000-plus-per-birth formula already exists in law — and condition it on a verified plan-of-safe-care completion rate and an outcome measure at 6 and 12 months post-discharge.

Add plans of safe care to the OIG CAPTA audit protocol and to Children and Family Services Review case review. The audits are already running. Adding one instrument to the sampling frame is close to free, and it is the only mechanism in existence that has produced hard numbers about state CPS compliance.

Require confirmatory testing and informed consent before any hospital drug screen becomes a child welfare or law enforcement referral, and decouple the notification pathway from the police pathway entirely. New Jersey has already written the template. Seven of the eight states in the Marshall Project analysis auto-share pregnancy substance-use allegations with law enforcement even when unfounded — a policy that guarantees mothers will hide substance use from the exact clinicians who are supposed to build the plan.

The through-line is simple. Congress wrote a mandate with three parts — notify, plan, monitor — and built a federal enforcement apparatus for none of them. The result is a paper referral that satisfies a hospital's legal duty, a hotline log entry that satisfies a state's, an aggregate count that satisfies HHS, and a newborn who goes home to a caregiver whose treatment needs no one is required to address, in the age bracket that dies more often than any other. Twenty-two thousand plans against half a million exposed births is not a compliance shortfall. It is a system that measures its own paperwork and calls it safety.

---

Sources: HHS Children's Bureau, Child Maltreatment 2023 · 42 U.S.C. § 5106a / CAPTA as amended · ACYF-CB-PI-25-01, CAPTA State Grant Program Instruction · GAO-18-196, Substance-Affected Infants · HHS-OIG Work Plan SRS-A-25-005, CAPTA State Grant Audits · HHS-OIG, Maine Did Not Comply… (A-01-23-02500) · HHS-OIG, West Virginia Did Not Comply… (OAS-25-01-011) · NCSACW Brief 1: Identification and Notification · NCSACW Brief 2: Plans of Safe Care Data and Monitoring · NYC ACS/DOHMH Joint CAPTA-CARA Guidance, revised March 2026 · The Marshall Project, "Moms of Black Babies More Likely to Be Flagged to Police" · The Marshall Project, "New Jersey Adopts Reforms Around Drug Testing of Pregnant Patients" · TribLive, "'Frightening rate' of children dying due to parents' drug abuse" · Kids at Risk Action, "Parental Drug Use Killing Children in Pennsylvania at High Rates" · WV DHHR Office of Drug Control Policy, substance-exposed infants · Lloyd et al., "Planning for safe care or widening the net?" · Rebbe et al., "U.S. State rates of newborns reported to child protection at birth" · Casey Family Programs, Plans of Safe Care · USASpending.gov award records, CFDA 93.669