A new study reveals that the sepsis alert trusted by hospitals across America misses most children who have sepsis.
The major new study, published in the prestigious JAMA Pediatrics this week, has found that a widely used automated sepsis alert system for children — built into the Epic electronic health record, which is used in roughly four out of ten acute care hospitals and controls over half (54.9%) of all acute care hospital beds in the US — misses most of the children who have sepsis when using the Phoenix Sepsis Score.
Not some. Most.
The researchers looked at more than 166,000 pediatric emergency visits across eight emergency departments in New York State from January 2024 through June 2025. Of the children who actually had sepsis, the alert identified only 36%. That means nearly two out of every three children with sepsis were not flagged by the system that doctors and nurses are relying on to catch them.
It gets worse. The study also asked how often the alert fired before sepsis had fully taken hold — the window when early treatment can change everything. The answer: 30% of the time. Seven in ten children with sepsis were not warned about in advance of the critical threshold. The early detection alert threshold had low sensitivity, which was poorly suited for pediatric sepsis screening.
64% of children with sepsis were NOT detected by the alert
70% of cases were not identified early, before sepsis took full hold
42% of US hospitals use Epic — the system running this alert
The Children Who Were Failed the Most
The study didn’t just reveal that the alert fails broadly. It revealed that it fails some children far more than others — and the pattern is not random.
Among Black children, the alert’s sensitivity dropped to just 22%. That means nearly four out of five Black children with sepsis were missed entirely. For early identification — catching sepsis before it reached its most dangerous stage — the sensitivity in Black children fell to just 12%.
For children under five years old, one of the groups most vulnerable to sepsis and the age group that accounts for the greatest share of sepsis deaths worldwide, the identification rate was 20%. Early identification was just 14%.
And for children treated in hospitals that were not dedicated pediatric centers — the community hospitals, the mixed adult-and-children emergency departments — the alert sensitivity fell to 25%.
I want you to sit with that for a moment. If you are a Black mother, or the parent of a child under five, or you live somewhere without a children’s hospital nearby, the computer system your emergency department is using to catch sepsis in your child could be failing at a rate that should terrify every one of us.
Nearly four out of five Black children with sepsis were missed entirely by the alert.
“A False Sense of Security”
What troubles me as much as the numbers is what the researchers describe as the consequence of a failing system that nobody is told is failing.
They write that reliance on this alert has the potential to contribute to what they call “a false sense of security” — a situation where a nurse or doctor, seeing no alert fire, may feel reassured. The computer didn’t flag anything. It must be okay. We can wait.
That false reassurance can cost minutes. And in sepsis, minutes are not abstract. Every hour of delayed treatment increases the risk of death. Every hour is someone’s child.
The researchers also note something that should not be happening at all: many hospitals are actually rewarded — through Honor Roll programs and financial incentives — simply for having implemented this alert. “And yet, administrators and clinicians may be unaware of its performance,” the researchers say.
Hospitals can check a box that says “sepsis screening in place” and receive recognition for it, while the screening tool they are using fails to detect most of the children in front of them. The bureaucracy is satisfied, but the children are not protected.
What We Know About What Works
I am not a researcher. I am a mother. But I have spent years learning everything I can about sepsis since my “12-year-old” son died, because I could not bear the thought that another family would go through what ours did without understanding why.
What I have come to understand is that early recognition of sepsis is not a technology problem alone. It is a human problem, a training problem, a culture problem. Technology can help — but only if it is honest about what it can and cannot do.
The study also points to something END SEPSIS has always believed: that the best outcomes come from comprehensive care bundles — trained eyes, established protocols, fast action — not from a single alert that most of the time stays silent when it should be screaming.
Many hospitals are rewarded for having the alert switched on. Not for it working. Not for outcomes. Just for having it.
What I Want From the People Reading This
If you are a parent, I want you to know the signs of sepsis and never wait for a computer to tell you something is wrong. The signs are: a fever or abnormally low temperature; fast or difficult breathing; skin that is pale, mottled, or blueish; extreme pain that seems out of proportion to the illness; confusion or sudden change in behavior; and a feeling — in you or your child — that something is very seriously wrong. Trust that feeling. Use the words. Say to the doctor or nurse: Could this be sepsis?
If you are a clinician, I want you to hear this: the tool you may be relying on is not reliable. It is not your fault that it was sold to your hospital as something it is not. But now you know. And what you do with that knowledge matters enormously to the children who will come through your doors.
If you are a hospital administrator, or a health system leader, or a policymaker: this study is about your hospitals. These are your alert systems. These are your children.
The Work Continues
END SEPSIS was built on exactly this kind of failure — the failure to recognize sepsis in a child who came to a hospital asking for help, but was sent home, and died.
Rory Staunton was twelve years old. He scraped his arm in gym class. He was seen by medical professionals. Nobody connected the dots in time.
Since his death, we, his parents, have fought to make sure that the systems meant to catch sepsis actually catch it. We fought for Rory’s Regulations. We fight for federal funding. We are fighting for Congress to pass the SEPSIS Act. And we will keep fighting.
This new research is a reminder of exactly why that fight is not over. We have alert systems that miss most children. We have disparities that mean Black children and the youngest children are failed at even higher rates. We have financial incentives tha often reward the appearance of safety rather than its substance.
That is not good enough. It has never been good enough.
My son deserved better. Your child deserves better. Every child deserves a system that actually sees them.
– Orlaith Staunton, co-founder of END SEPSIS.
For more information about sepsis recognition, prevention, and END SEPSIS’s advocacy work, visit EndSepsis.org.
