What NCANDS Is and How It Counts Maltreatment

NCANDS stands for the National Child Abuse and Neglect Data System, the federal collection that gathers what state child protection agencies record each year and turns it into the national Child Maltreatment report series published by the Children's Bureau. If you have ever read a national rate of substantiated maltreatment and wondered where the number came from, it came from this system, built one state submission at a time.

The details cited in this entry were checked on Child Maltreatment reports.

What NCANDS Actually Collects

NCANDS does not run its own investigations or keep its own case files. It is a reporting pipeline. Each state child protection agency decides how it screens a call, how it investigates, and how it classifies the outcome, under its own state law. What NCANDS asks every state to do, once a year, is translate that local record into a common federal format so the Children's Bureau can combine the submissions into one national file. For federal fiscal year 2021, the 50 states, the District of Columbia and Puerto Rico all submitted data. Participation is voluntary: each state maps its own administrative system onto the NCANDS data structure before it submits.

That translation step is where most of the system's complexity lives. A state that uses an intermediate finding such as "indicated" and a state that only distinguishes substantiated from unsubstantiated are describing different legal categories, and NCANDS has to map both into shared fields before anyone can compare them. The system is built to absorb that variation, not erase it.

Two Files: Case-Level and State-Level

States send NCANDS two datasets. The Child File is the case-level component and the most detailed: one record per child on each report, the report-child pair, covering every investigation or assessment that received a disposition during the reporting year. Its data elements include the demographics of the children and of the alleged perpetrators, the types of maltreatment, the disposition, risk factors, and the services provided as a result.

The Agency File is the state-level component. It collects aggregate counts on items that cannot be gathered case by case, such as the number of referrals screened out, child protective services staffing, preventive and reunification services, and child fatalities. A state that submits the Child File also submits the Agency File, so the two are complementary rather than alternatives.

This distinction matters for anyone trying to compare states directly. Case-level data lets researchers ask detailed questions, such as how outcomes differ by the age of the child or the relationship of the alleged perpetrator. Aggregate items only answer the questions they were built to answer, usually totals. Reading a multi-state comparison without checking which file a figure comes from is a common way to overread the data.

Who Decides What Counts as Maltreatment

NCANDS inherits its categories from state law, not the other way around. Each state defines abuse and neglect in its own statutes, covered in detail on the Child Welfare Information Gateway's own pages on legal definitions, and that state definition is what drives whether a report is accepted, investigated, and ultimately coded for NCANDS. The federal system standardizes the format of the count. It does not standardize what gets counted in the first place.

That is why two neighboring states can report different rates for what looks, to an outside reader, like the same kind of harm. One state's statute may set a lower threshold for what counts as neglect, or require a different standard of evidence to substantiate a report, and that legal difference shows up in the national file as a rate difference, with no indication on the surface that the underlying law, not the underlying harm, moved the number.

How the Yearly Report Gets Built

Once state submissions are in, the Children's Bureau compiles them into the annual Child Maltreatment report, the document cited elsewhere on this site when discussing how a city or county traces its own numbers back to a national source. NCANDS counts on the federal fiscal year, from October 1 to September 30, and a case is counted in the year its investigation or assessment received a disposition, not the year the call came in.

The published edition always describes a period that ended well over a year earlier, because states have to extract, map and submit their files after the fiscal year closes, and the Bureau has to validate and compile them before publication. That lag is structural, and it is worth keeping in mind whenever the latest national edition is set beside a local figure from the current year.

What the System Cannot Tell You

NCANDS counts reports and their outcomes. It does not, on its own, measure how much maltreatment actually occurs and goes unreported, and it does not assign a cause to a change in the numbers from one year to the next. A rise in substantiated cases in a given state could mean more harm occurred, or it could mean a hotline campaign reached more people, or a change in state law lowered the threshold for substantiation. The file records the outcome of that year's reports. It does not record which of those explanations is true.

Program and data staff who use NCANDS figures to argue for funding or to evaluate a prevention effort need to pair the national count with local context: what changed in the state's law, its hotline reach, or its screening practice in the same period. Without that context, a single year's rate is a number without a reason attached to it.

Where This Fits in the Bigger Picture

NCANDS sits upstream of most of the community data work described elsewhere on this site, including risk mapping and the ethical questions that come with it. A model that tries to predict where maltreatment is likely to be reported usually draws, directly or indirectly, on the same state administrative records that feed NCANDS, with all the variation in law and practice those records carry. Understanding the collection system is a prerequisite for judging any tool built on top of it, because a model can only be as consistent as the reporting it learned from.

For a reader who wants the primary source rather than a summary, the Children's Bureau's Child Maltreatment reports page lists each year's edition in order, and reading one edition's technical notes directly is the most reliable way to see how a specific year's states reported.

About the source

The Children's Bureau, part of the US Department of Health and Human Services' Administration for Children and Families, administers NCANDS and publishes the annual Child Maltreatment report series built from state submissions to the system. NCANDS was designed in response to the Child Abuse Prevention, Adoption and Family Services Act of 1988, which amended the Child Abuse Prevention and Treatment Act (CAPTA) to require a national data collection program on child maltreatment. Researchers who need the case-level records rather than a published summary can obtain the Child File, with its user's guide and codebook, from the National Data Archive on Child Abuse and Neglect (NDACAN).