Editor’s note: Electronic sepsis alerts are designed to help us recognize sick children sooner, but how well do they perform in practice? A new JAMA Pediatrics study examines that question and raises concerns about missed cases, timely detection, and differences in performance across patient groups and ED settings. The study’s lead author, Dr. Chelsey Mitchell, was my co-fellow during pediatric emergency medicine fellowship at Columbia. I’m excited to highlight her work and its implications for the children we care for in the ED.
— Dr. Josh Belfer, Editor, HipPEMcrates
The utility of screening for pediatric sepsis has long been debated, and with the recent changes to the Surviving Sepsis Guidelines stating that there is insufficient evidence to recommend implementing systemic sepsis screening – in addition to established clinical protocols to improve recognition – it makes you wonder…How useful is the most used screening tool in the United States?
Studies such as this have been difficult to perform due to an unclear definition of pediatric sepsis. With the novel and now established Phoenix Sepsis Scoring system, it made studying the screening alert more straightforward, with clearly defined outcomes that are not dependent on physician opinion or treatment and rather a clear, defined, and established criteria.
In our recent JAMA Pediatrics study, we evaluated the Epic System’s Pediatric Sepsis Early Detection Model (PESM) alert. In our cohort, the alert identified only about one-third of children who met the study’s sepsis definition. Here is what ED clinicians should know.
What we studied
Our retrospective study included 166,943 pediatric encounters across 8 EDs in one New York health system from January 2024 through June 2025. We evaluated whether the PESM alert triggered during the first 24 hours after ED arrival and whether it triggered before patients met Phoenix sepsis criteria. There were 189 sepsis cases, defined by a Phoenix score of at least 2.
What we found
- Overall sensitivity was 36%. The alert did not trigger in approximately 64% of children who met the study’s sepsis definition.
- Early detection sensitivity was 30%. When the alert had to fire before Phoenix criteria were met, it identified fewer than one-third of cases.
- Positive predictive value was 3%. Only about 3 in 100 encounters with an alert met the study’s sepsis definition within 24 hours.
- Sensitivity was lower in several subgroups. It was 22% in Black children, 20% in children younger than 5 years, and 25% in EDs outside of tertiary children’s hospitals. These subgroup estimates had wide confidence intervals.
What this means for the ED
Our interpretation is that an absent alert should not reassure clinicians when the history, examination, or trajectory raises concern for sepsis. The reported negative predictive value exceeded 99%, but sepsis occurred in only about 0.11% of encounters. In such a low-prevalence population, a high negative predictive value can coexist with poor sensitivity.
For ED leaders, these findings support asking how the alert performs locally: how often it misses sepsis, whether it fires early enough to change care, and whether performance differs by patient group or care setting. A tool’s availability in the electronic health record does not establish its effectiveness in your ED.
How far can we take these findings
This was a retrospective evaluation in one health system using a specific sepsis definition and 24-hour window. The findings describe the alert as evaluated here; they do not establish the performance of every sepsis screening tool or show that the alert itself caused treatment delays or worse outcomes. The subgroup findings raise equity concerns but do not establish why performance differed.
The clinical takeaway: use electronic alerts cautiously as one part of sepsis recognition. Continue bedside assessment and reassessment when a child’s presentation raises concern, even when the screen is quiet.
Read the study
Mitchell CM, Zhang Z, Ranard BL, et al. Evaluation of a Commercial Pediatric Sepsis Early Detection Alert in the Emergency Department. JAMA Pediatr. Published online October 5, 2026. doi:10.1001/jamapediatrics.2026.4575.
Dr. Chelsey Mitchell, MD, MS, is a Pediatric Emergency Medicine physician and Assistant Medical Director at NewYork-Presbyterian Morgan Stanley Children’s Hospital Emergency Department. She can be reached at cmm2308@cumc.columbia.edu.
