For a Deeper Dive on this or other topics,
hit the DEEP DIVE button on the lower right corner of this page.
Best and Worst States for Very Young Children in the Child Protection System
New national data show a troubling pattern: in the United States, children under age five face the highest risk of serious harm once they are on the radar of child protective services (CPS). Using 2017 NCANDS (National Child Abuse and Neglect Data System) data for all 50 states and Washington, D.C., a recent research project compared states to see where very young children are most at risk – and where systems appear to be doing a better job of keeping them safe.(podcast)
The analysis focused on several key measures: the share of victims under age five, child fatalities per 1,000 confirmed victims, how often referrals are “screened in” for investigation, and how many high‑risk flags each state trips at once. What’s missing is the use of Family Assessment in place of investigation of the child and the abuse suffered. Only a few states consistently report using a “family assessment” (also called “alternative response” or “differential response”) track instead of a formal investigation for some portion of screened‑in reports, and even among those states, reporting practices are uneven and change over time. There is very little reporting around egregious incidents, near death, or torture show up for any of these children and zero reporting for children not reported to CPS (child abuse is invisible and likely many more cases are not reported than are reported).
This reporting is a simple predictive model of the probability that each state belongs to a “high‑severity” group and sorted states into four tiers: Immediate Review, Priority Review, Review, and Monitor.
The Worst States for Very Young Children
States in the Immediate Review tier are the most dangerous places for very young children in the CPS system. States in the highest‑risk tier are often those where CPS is already deeply troubled and overwhelmed by heavy caseloads, short on staff and funding, and struggling to provide consistent, meaningful support even to the families who do manage to get into the system. These states combine high fatality rates with multiple warning signs across other metrics. In this group, the model found:
-
An average predicted probability of “high severity” around 0.96
-
An average fatality rate of about 5.9 deaths per 1,000 confirmed victims
-
Multiple risk flags triggered at the same time
States in this top‑risk tier include:
-
Pennsylvania
-
Washington
-
Idaho
In these four states, very young children are both more likely to be seriously harmed and more likely to die once they are known to the system. High shares of victims under age five, elevated fatalities, and “stacked” risks (for example, high investigation‑to‑victim ratios and high neglect rates) combine into a profile that demands immediate review by lawmakers, agency leaders, and communities.
The Best‑Performing States for Young Children
At the other end of the spectrum are states in the Monitor tier. Here, the model sees a low probability of high‑severity problems, and core indicators look more favorable. These states tend to share several traits:
-
Lower fatalities per 1,000 victims
-
A smaller share of victims under age five
-
Fewer high‑risk flags triggered at once
-
More consistent, less extreme patterns across 2013–2017
“Monitor” does not mean “problem‑free.” Even the best‑performing states still experience child deaths, widespread neglect, and system gaps. But relative to the national picture, they appear to be safer environments for very young children and to operate with more balanced patterns in how cases are screened, investigated, and handled.
What the Data Say About Children Under Five
Taken together, the numbers for very young children are stark:
-
Children under age five account for a disproportionate share of child maltreatment victims in almost every state.
-
From 2013 to 2017, child fatalities increased by roughly 9%, while the total number of unique child victims rose only about 2.7% – suggesting that severity is getting worse even when overall case counts are not exploding.
-
Neglect is by far the dominant form of maltreatment nationwide, making up roughly two‑thirds of all cases, and it hits very young children hardest, especially in families already facing poverty, housing instability, and untreated mental health needs.
This means the “worst” states are not the only places struggling. They are simply the jurisdictions where these problems are most concentrated and most deadly for the youngest children.
Why CPS Transparency and Better Metrics Matter
One of the clearest lessons from this kind of analysis is the value of CPS transparency. When child protection data are fragmented, delayed, or reported in confusing ways, it becomes almost impossible for lawmakers, judges, and communities to see what very young children are actually facing.
Regular, clear reporting of key metrics can change that, for example:
-
Standardized dashboards showing fatalities per 1,000 victims, the share of victims under age five, screening‑in rates, investigation intensity, and repeat maltreatment.
-
Consistent, annual publication of comparable state‑level statistics, rather than buried tables or one‑off PDFs that are hard for non‑experts to interpret.
-
Disaggregated data by age, race, geography, and maltreatment type so patterns of risk for infants and toddlers are visible, not hidden in statewide averages.
When CPS agencies share this information in a timely, accessible way, lawmakers and policy makers can make better decisions:
-
They can see which states are losing very young children at alarming rates – and ask why.
-
They can identify whether high‑risk states are under‑resourced, applying different thresholds, or missing core prevention supports like home visiting and child‑care assistance.
-
They can study what lower‑risk states are doing differently in screening, investigation, and early‑childhood support, and adapt those practices.
Without this transparency, the lived conditions of at‑risk children and families remain largely invisible, and debates about child welfare policy are driven by anecdotes and headlines instead of evidence.
How Advocates and Communities Can Use This Information
A tiered, state‑by‑state view is not about shaming individual states. It is a tool for prioritizing questions and action:
-
Where should advocates and journalists focus first when investigating child fatalities among very young children?
-
Which states should be at the top of the list for hearings, audits, or independent reviews?
-
Where are there promising practices worth lifting up and replicating?
For organizations like KARA (Kids At Risk Action), analyses like this help:
-
Prioritize research and storytelling in the most dangerous states for very young children.
-
Design campus and community conversations around the places where infants and toddlers face the greatest risk.
-
Give legislators, judges, and the public a clear, data‑backed map of where child protection systems need immediate review, targeted improvement, or ongoing monitoring.
The data do not tell every story, and they cannot capture every family’s experience. But they do make one thing unmistakably clear: where a very young child lives still has far too much to do with whether that child will be hurt – or survive – once the child protection system gets involved. More transparent CPS reporting and regular, honest use of these metrics are essential if that is ever going to change.
Special thanks to Northeastern University student Tanay Hançerli for this powerful data analysis and state‑by‑state risk modeling, which made it possible to see more clearly where very young children are most at risk and where change is most urgently needed.
How are the at-risk children where you live?
Hit the deep dive button on the lower right of your screen to find out.
KIDS AT RISK ACTION / KARA / INVISIBLE CHILDREN
#ChildWelfare
#ChildProtection
#CPSReform
#KidsAtRisk
#ChildAbusePrevention
#NCANDS
#EarlyChildhood
#FamilySupport
#SocialWork
#ProtectChildren
#childdeath
#childviolence
