- Design
- Retrospective multicentre diagnostic cohort
- Population
- 166,943 paediatric ED encounters in 8 departments, New York
- Primary outcome
- Sensitivity and PPV for Phoenix-defined sepsis within 24 h
- Effect
- Sensitivity 36% (29–43%); PPV 3% (2–4%); under-fives 20%
This retrospective study evaluated a widely deployed electronic health record sepsis alert across eight emergency departments in one New York health system, covering 166,943 paediatric encounters from 2024 to 2025. Sepsis was defined by the Phoenix criteria.
There were 189 sepsis cases. The alert identified 36% of them, and only 30% before the child met sepsis criteria. Its positive predictive value was 3%, so 97% of alerts were false alarms. Sensitivity was lower in children under five (20%), in Black children (22%) and in departments outside tertiary children's hospitals (25%).
Automated alerts are attractive where staff are stretched, but this one would have missed two in three septic children while generating constant noise. Clinical assessment — vital signs against age norms, capillary refill, mental state, and parental concern — remains the screen.
- Do not rely on an automated sepsis alert to identify septic children.
- Check heart rate, respiratory rate and capillary refill against age-specific norms.
- Take parental concern that a child is 'not themselves' seriously.
- Be most alert in under-fives, where the tool performed worst.
- Validate any electronic alert locally before trusting it.
Why it matters
A tool many hospitals assume is protecting them missed most cases.
Don't overread it
This evaluated one product in one US system; other tools and settings may perform differently.
The statistics, in plain English
Sensitivity of 36% means 64% of septic children triggered no alert. A positive predictive value of 3% means about 33 alerts fired for every true case. The negative predictive value above 99% reflects how rare sepsis is, not that a silent alert is reassuring.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for paediatrics, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free