One in ten children with sepsis die in US hospitals, first national clinical estimate finds

More than 1,800 children die from sepsis in American hospitals every year. That number - drawn from the first national surveillance effort to use clinical data rather than billing codes - is published March 22 in JAMA, and it arrives with a blunt implication: we have been counting wrong.

Billing codes versus bedside reality

For years, pediatric sepsis estimates in the United States relied on administrative billing data. Hospitals code diagnoses for reimbursement purposes, and those codes became the default yardstick for tracking how often sepsis strikes children. The problem is that coding practices vary wildly from one hospital to the next, and they shift over time as guidelines and incentives change. The result is a moving target that makes it difficult to know whether sepsis is becoming more or less common, or whether any given intervention is working.

The new study, led by Chanu Rhee of Harvard Medical School and colleagues at the Harvard Pilgrim Health Care Institute, Children's Hospital of Philadelphia, and Nemours Children's Hospital in Delaware, took a different approach. The team built an electronic health record-based definition called the Pediatric Sepsis Event (PSE), modeled on the CDC's existing Adult Sepsis Event framework and adapted using the recently developed Phoenix pediatric sepsis clinical criteria.

Instead of relying on what a coder typed into a billing system, the PSE definition pulls objective clinical markers directly from the medical record: laboratory results, antibiotic orders, and measures of organ dysfunction. The intent is consistency - the same data elements measured the same way across hundreds of hospitals.

What 3.9 million hospitalizations revealed

The researchers analyzed 3.9 million pediatric hospitalizations spanning 2016 through 2023. When they applied the PSE definition, sepsis appeared in 1.3% of all pediatric hospital stays - roughly 1 in every 75 hospitalized children. Among those identified, more than 1 in 10 died before discharge. Scaled nationally, that translates to more than 18,000 sepsis cases and more than 1,800 deaths each year.

The PSE definition was validated against physician chart review and outperformed billing codes in accuracy. The rates of pediatric sepsis and associated mortality remained relatively stable between 2016 and 2022, a finding that contradicts some earlier billing-code-based studies suggesting dramatic changes over the same period.

Nearly 1 in 5 pediatric hospital deaths involved sepsis. Most cases were already present when the child was admitted, but a smaller fraction developed during hospitalization. That hospital-acquired subset carried higher mortality, pointing toward the importance of infection prevention alongside early recognition.

A surveillance framework, not a cure

The study is fundamentally an exercise in measurement. It does not test a treatment or propose a clinical intervention. What it offers is a standardized ruler - one that can be applied consistently across health systems to track pediatric sepsis over time and compare outcomes between hospitals.

That matters because you cannot improve what you cannot reliably measure. If Hospital A uses one coding convention and Hospital B uses another, comparing their sepsis rates is meaningless. A clinical-data-based definition sidesteps that problem, at least in theory.

There are important caveats. The PSE definition depends on electronic health records that capture the relevant data elements, which may exclude smaller or under-resourced facilities. The validation was performed against physician chart review at participating centers, and the extent to which it generalizes to all hospital types remains to be tested. The study also covers only hospitalized children; sepsis identified and treated in emergency departments without admission would not be captured.

Counting as a first step

The CDC funded this work, and the study was presented at the Society of Critical Care Medicine Congress. The underlying logic is straightforward: standardized national surveillance creates a foundation for targeted prevention, quality improvement, and policy decisions. Whether that foundation actually leads to fewer children dying from sepsis will depend on what comes next - the interventions, the funding, and the willingness of health systems to act on what the numbers show.

For now, the numbers themselves are sobering enough. Sepsis kills more than five children per day in US hospitals. We just did not have a reliable way to say so until now.