This article is about the problems lawmakers are having with CPS transparency and obtaining the metrics and statistical information they need for better decision making.
Most current reporting to legislatures and the public relies on point‑in‑time data—how many children are in care today, how many reports came in this year, how many cases closed—rather than longitudinal data that follows families over time.
Point‑in‑time numbers show only a snapshot, not the full story; they can under‑count the total number of children and families touched by child protection across years.
- Without long‑term, linked records, lawmakers cannot see:
- How many families cycle in and out of the system over decades.
- Which families have repeated investigations or removals across generations.
- Whether interventions reduce re‑entry or multi‑generational involvement.
Casey Family Programs notes that longitudinal data—collected across multiple points in time gives a far more complete picture of children’s and families’ experiences in child welfare than single‑point data and is necessary to understand which approaches keep children safe and reduce foster‑care use and which are creating the opposite conditions CPS was designed to fix.
KARA is asking to keep family‑level histories in coded form:
- Counties and states can assign stable family or case IDs and retain records (reports, case plans, outcomes) linked to those IDs, while stripping names and direct identifiers through de‑identification techniques.
- This protects individual privacy and complies with confidentiality laws, while still allowing the system to see patterns like:
- Families with multiple removals over 10–40 years.
- Children who later become parents with their own child‑protection involvement.
- “High‑contact” families who come to the agency’s attention repeatedly but never get effective help.
Data‑de‑identification and coding (e.g., replacing names with randomized IDs, recoding dates and locations to ranges) is a well‑established way to allow ethical study of outcomes without exposing identities.guides.library.
By retaining de‑identified records with family numbers instead of names, lawmakers can see that a small but significant group of families are touched by child protection again and again, often across generations. Without those longitudinal records, those patterns remain invisible. Every case looks new, and we never learn from history.
Many important questions cannot be answered with short‑term or name‑deleted data:
- How many families have had three or more reports/investigations in a 10‑ or 40‑year span?
- In how many cases are children removed from a home later returning as parents and grandparents with new generations of children entering care?
- Which interventions (family resource centers, hospital‑to‑home programs, intensive in‑home services) are associated with fewer repeat removals or reports over time?
- Are certain patterns of county response (e.g., quick removal without robust services) linked to higher re‑entry rates years later?
Casey Family Programs emphasizes that longitudinal data is what allows agencies to answer questions like “Which programs are most effective at preventing children from re‑entering out‑of‑home care?” and “What happened right before reunification?”. The same logic applies to multi‑decade family involvement.
Without de‑identified, linked records:
- Each report is treated as isolated.
- We cannot distinguish between families who had one crisis and then stabilized, and families whose involvement spans decades.
- We cannot design policy specifically for that small group of families whose children and grandchildren keep cycling through foster care.Address privacy and confidentiality concerns head‑on
Lawmakers will be wary of keeping names or identifiable records for long periods. You can reassure them:
- De‑identification techniques—removing names, exact addresses, social security numbers, and other direct identifiers, and replacing them with internal codes—are standard practice in health, education, and social‑service research.
- Federal and state confidentiality laws already permit use of de‑identified or coded data for public‑interest research and system improvement, if identities are protected and data is securely stored.
- The proposal is to keep:
- Family IDs and case timelines.
- Types of reports, responses, services, and outcomes.
- Without names or other direct identifiers that could be used to publicly identify individuals.
We are not asking to keep names forever. We are asking to keep family‑level histories in coded form, so Minnesota can learn from long‑term patterns while still honoring confidentiality and privacy.
- Longitudinal, family‑level data lets lawmakers see whether policies and programs are reducing long‑term involvement or just “moving numbers” at a single point in time.evidence2impact.
- This allows the system to target interventions where they are most needed. Families who have been in and out of the system for decades, whose children are at high risk of repeating that pattern.
- This is a public‑health approach to child welfare, where CPS is part of a broader continuum of family support and data is used to strengthen prevention, not just track after‑the‑fact responses.
Without de‑identified, long‑term records linked by family number, we never truly see the children and families who live in the child‑protection system for decades, and know the percentages of children grow up in care and returning as parents and following generations as grandparents. Today, every report looks like a one‑time event. Keeping coded, longitudinal data is the only way lawmakers can understand multi‑generational involvement, measure whether our interventions change it, and design policies that break those cycles instead of just managing them year by year.
FREQUENTLY ASKED QUESTIONS:
Q1: Why isn’t existing CPS data enough for lawmakers?
A1: Most CPS reporting is point‑in‑time—how many children are in care today, how many reports this year—and doesn’t show families who cycle in and out of the system over decades or across generations. Without longitudinal data, lawmakers cannot see multi‑generational involvement or measure whether interventions reduce long‑term system contact.
Q2: What does KARA mean by “de‑identified, family‑level longitudinal data”?
A2: KARA proposes that counties and states assign stable family IDs, strip names and direct identifiers, and keep linked records over time—reports, case plans, services, and outcomes—so lawmakers can see long‑term patterns without exposing individual identities, using established de‑identification techniques and privacy safeguards.
Q3: How would de‑identified longitudinal CPS data help improve child protection?
A3: Longitudinal family‑level data would let legislators and agencies see which interventions reduce repeat removals, which county responses are tied to higher re‑entry rates, and how many children grow up in care and return as parents. That information is essential to design policies that break generational cycles rather than just manage them case by case.
Share this article with your State Representative and other government officials.
Include the statutory language listed in the read more below.
Change won’t come until more of us lend our voices to voiceless children
Register to vote|Find my polling place|
Who represents me in the legislature?|
Contact state house members|Contact state senators|
Nationally: Contact your member of congress|Contact your U.S. senator
- Purpose and findings
Section 1. Purpose and legislative findings.
(a) The legislature finds that children and families in this state are sometimes involved with the child‑protection system across many years and, in some instances, across generations. Existing reporting to the legislature and the public primarily reflects point‑in‑time data and short‑term outcomes and does not adequately reveal long‑term patterns of system involvement, repeat maltreatment, or the effectiveness of prevention and intervention strategies.pmc.ncbi.nlm.nih+1
(b) The legislature further finds that longitudinal, family‑level data—collected and maintained in de‑identified form—are necessary to:
(1) understand the extent and nature of multi‑year and multi‑generational involvement in the child‑protection system;
(2) evaluate whether child‑protection policies, practices, and services reduce repeat involvement and improve safety and well‑being over time; and
(3) inform evidence‑based prevention and early‑intervention strategies that can safely reduce the need for child‑protection intervention.evidence2impact.psu+2
(c) The legislature declares that it is the policy of this state to retain, in de‑identified form, longitudinal records of family involvement with the child‑protection system for the purpose of improving child safety, permanency, and well‑being, while fully protecting individual privacy and confidentiality.hhs+1
- Definitions
Section 2. Definitions.
For purposes of this section:
(a) “Child‑protection record” means any record created or maintained by the state or a county agency relating to a report, assessment, investigation, safety or risk assessment, case plan, service provision, placement, court action, or case closure concerning alleged child abuse or neglect or a child in need of protection or services.
(b) “Family unit” means the set of individuals (children and caregivers) who are treated as a family or household by the child‑protection agency during a report, assessment, or case, including parents, legal guardians, and other caregivers living with the child.
(c) “Family identifier” means a unique, non‑identifying code assigned by the child‑protection agency to a family unit to allow linkage of records over time without the use of names or direct personal identifiers.
(d) “De‑identified data” means data from which personal identifiers have been removed and for which there is no reasonable basis to believe the information can be used to identify an individual, consistent with established de‑identification standards under federal and state privacy laws.hhs+2
(e) “Longitudinal family‑level dataset” means a de‑identified collection of child‑protection records that are linked over time by family identifier and that contain information about reports, agency responses, services provided, and case outcomes for each family unit.pmc.ncbi.nlm.nih+1
- Retention of de‑identified, family‑level records
Section 3. Long‑term retention of de‑identified child‑protection records.
(a) The [Department/Agency] shall maintain a longitudinal family‑level dataset of child‑protection records in de‑identified form for a minimum period of 25 years from the date of the initial report involving a family unit or such longer period as may be required by federal law, regulation, or court order.
(b) Personal identifiers, including but not limited to names, addresses, social security numbers, driver’s license numbers, full dates of birth, and other direct identifiers, shall be removed or coded as soon as reasonably practicable after case closure, consistent with de‑identification standards described in Section 4.hhs+2
(c) The de‑identified longitudinal dataset shall retain, at a minimum:
(1) the family identifier;
(2) dates or date ranges of reports, assessments, investigations, and case openings and closures;
(3) type(s) of alleged maltreatment and risk factors identified;
(4) agency responses, including safety plans, services offered and provided, and placement decisions;
(5) major case outcomes, including substantiation decisions, court orders, permanency outcomes, and re‑entries into care; and
(6) indicators of subsequent involvement of former child clients as parents or caregivers, where such information can be linked without use of names.aspe.hhs+2
- De‑identification standards and privacy safeguards
Section 4. De‑identification and privacy safeguards.
(a) The [Department/Agency] shall implement de‑identification procedures that are consistent with recognized standards under federal law, including HIPAA guidance on de‑identification methods, and any applicable state privacy statutes.hhs+2
(b) De‑identification procedures shall include:
(1) removal of direct identifiers (such as names, exact addresses, social security numbers, and other unique personal codes);
(2) coding or generalization of dates and locations to ranges or non‑specific geographic units where necessary to prevent re‑identification;
(3) assignment and use of family identifiers that are not derived from personal identifiers; and
(4) technical and administrative safeguards to prevent re‑identification and unauthorized linkage of the dataset to external identifiable data.guides.library.jhu+3
(c) The [Department/Agency] shall adopt policies that:
(1) prohibit staff and contractors from using de‑identified data to attempt to re‑identify individuals;
(2) limit access to the longitudinal dataset to staff and researchers with a documented need and appropriate training; and
(3) require data‑sharing agreements for any external use, specifying privacy safeguards, permitted uses, and penalties for misuse.aspe.hhs+2
- Use of de‑identified longitudinal data
Section 5. Use of de‑identified longitudinal data for system improvement.
(a) The de‑identified longitudinal family‑level dataset may be used by the [Department/Agency] to:
(1) analyze long‑term patterns of family involvement in the child‑protection system, including repeat reports, re‑entries into care, and multi‑generational involvement;
(2) evaluate the long‑term effects of child‑protection policies, practices, and services on safety, permanency, and well‑being;
(3) identify families and communities that experience disproportionate and extended involvement and inform targeted prevention and early‑intervention strategies; and
(4) report aggregate, de‑identified findings to the legislature, courts, and the public to improve transparency and accountability.evidence2impact.psu+3
(b) The [Department/Agency] may share de‑identified longitudinal data with approved researchers and partner agencies for purposes of child‑protection system improvement, public‑health assessment, and evaluation of prevention and intervention programs, subject to data‑use agreements that protect privacy and limit use to approved purposes.aspe.hhs+2
(c) Nothing in this section authorizes the release of identifiable child‑protection records to the public. All existing confidentiality protections for identifiable records remain in full force and effect.childwelfare+2
- Relationship to existing retention and destruction laws
Section 6. Relationship to existing retention and destruction requirements.
(a) This section supplements, and does not diminish, any existing requirements under state or federal law regarding the retention or destruction of identifiable child‑protection records, including requirements related to adoption records, court records, or federally funded case records.calsaws+1
(b) When identifiable child‑protection records are destroyed pursuant to existing law or policy, the [Department/Agency] shall, to the extent permitted by law, ensure that de‑identified information necessary for the longitudinal family‑level dataset is preserved in accordance with Section 3.
(c) Destruction of personal identifiers for purposes of de‑identification shall be considered compliant with applicable definitions of “destruction” of personally identifiable information under state and federal law.dasycenter+1
- Legislative intent (optional clarifying section)
Section 7. Legislative intent.
(a) It is the intent of the legislature that this section enable the state to understand and address long‑term and multi‑generational involvement in the child‑protection system without compromising the confidentiality of individual children and families.
(b) The legislature expects that the [Department/Agency] will use de‑identified longitudinal data to identify a small subset of families who remain involved with child protection for many years or across generations and to design policies and practices that can safely reduce such extended involvement.
(c) The legislature further intends that de‑identified longitudinal data be used to strengthen prevention, early intervention, and trauma‑informed supports so that fewer families ever need child‑protection intervention.
KIDS AT RISK ACTION / KARA / INVISIBLE CHILDREN
#CPS
#ChildProtection
#ChildWelfareData
#Transparency
#LongitudinalData
#FamilyLevelData
#TraumaInformed
#KARA
#KidsAtRisk
#PublicPolicy
You can add more if needed:
#CaseyFamilyPrograms
#DataDeIdentification
#ChildWelfareReform
