SGEM#471: Are ESI Levels Accurate for Triage of Pediatric Patients?

Subscribe: RSS

Reference: Sax DR, et al. Emergency Severity Index Version 4 and Triage of Pediatric Emergency Department Patients. JAMA Pediatrics, October 2024

Date: February 12, 2025

Dr. Brandon Ho

Guest Skeptic: Dr. Brandon Ho is a graduating pediatric emergency medicine fellow at Children’s National Hospital in Washington DC and soon to be attending physician at Seattle Children’s. His research interests include AI in healthcare, medical education, and social determinants of health.

Case: You are approached by the medical director of your emergency department (ED). She has noticed that recently, there has been an increasing number of pediatric cases presenting to your facility. In some of these cases, the children ended up being more sick than initially triaged. As the institution’s evidence-based medicine enthusiast, she asks you, “What do you think of the triage system we’re using now? How accurate is it for children?”

Background: Pediatric triage is a fundamental component of emergency medicine, serving as the first critical step in managing acutely ill or injured children in the emergency department (ED). Unlike adult triage, pediatric triage presents unique challenges due to variations in physiology, developmental differences, and communication barriers in younger patients. Accurately assessing the severity of a child’s condition is essential for ensuring timely intervention while avoiding unnecessary resource utilization.

The Emergency Severity Index (ESI) is the most widely used triage system in the United States. It classifies patients based on acuity and predicted resource utilization, ranging from ESI Level 1 (requiring immediate, life-saving intervention) to ESI Level 5 (requiring no resources beyond physician evaluation). However, pediatric triage remains particularly challenging due to factors such as age-based vital sign variations, difficulty in obtaining an accurate history, and non-specific presentations of critical illness.

Typically, ESI levels 1 and 2 are used to assess acuity and risk of instability. ESI levels 3, 4, and 5) are determined by expected resource needs. Those resources can be labs, imaging, medications, consultations, etc.

Despite its widespread use, it’s imperfect with previous studies reporting mistriage rates as high as 50%. Pediatric patients can either be undertriage (assigning a lower acuity level than warranted) or overtriage (assigning a higher acuity level than necessary). This can have significant consequences when EDs are experiencing prolonged wait times, the boarding of patients, and are chronically short-staffed.

Undertriage may lead to delayed care for critically ill children, whereas overtriage can result in unnecessary resource use, increased healthcare costs, and prolonged ED crowding. Studies have shown that pediatric patients are frequently subject to both types of errors, with younger children and those presenting with atypical symptoms being at risk.

Clinical Question: How accurate is ESI version 4 in predicting acuity and resource needs among pediatric ED patients?

Reference: Sax DR, et al. Emergency Severity Index Version 4 and Triage of Pediatric Emergency Department Patients. JAMA Pediatrics, October 2024

Population: Pediatric patients (aged 0-18 years) presenting to 21 Kaiser Permanente Northern California ED’s from January 1, 2016, to December 31, 2020.
- Excluded: Missing ESI, incomplete ED time variables, transferred patients, patients who left against medical advice (AMA) or left without being seen (LWBS).
Exposure: Assigned ESI level compared to actual resource utilization and critical interventions.
Comparison: Correct triage rates were compared against undertriaged and overtriaged cases to identify patterns of mistriage.
Outcome:
- Primary Outcome: The rate of mistriage (undertriage or overtriage) of pediatric patients using ESI v4.
- Secondary Outcomes: Patient and visit characteristics associated with mistriage, including age, sex, comorbidities, mode of arrival, and race/ethnicity.
Type of Study: Multicenter retrospective cohort study

Authors’ Conclusions: “This multicenter retrospective study found that mistriage with ESI version 4 was common in pediatric ED visits. There is an opportunity to improve pediatric ED triage, both in early identification of critically ill patients (limit undertriage) and in more accurate identification of low-acuity patients with low resource needs (limit overtriage). Future research should include assessments based on version 5 of the ESI, which was released after this study was completed.”

Quality Checklist for Observational Study:

Did the study address a clearly focused issue? Yes
Did the authors use an appropriate method to answer their question? Yes
Was the cohort recruited in an acceptable way? Yes
Was the exposure accurately measured to minimize bias? Yes
Was the outcome accurately measured to minimize bias? Unsure
Have the authors identified all-important confounding factors? Unsure
Was the follow-up of subjects complete enough? Yes
How precise are the results? Fairly precise.
Do you believe the results? Yes and No.
Can the results be applied to the local population? Unsure
Do the results of this study fit with other available evidence? Yes and No
Funding of the Study: Many authors received grants from Kaiser Permanente Lokahi Risk Reduction Program. Dr. Pines received funding from some pharmaceutical companies for unrelated work.

Results: They included over one million pediatric ED encounters across 21 EDs. The mean age was 7.3 years and there were slightly more males than females.

Key Results: ESI Version 4 commonly mistriages children.

Children were correctly triaged only approximately a third of the time (34%, 95% CI 34-34.2).

Primary Outcome

Secondary Outcomes

Race and ethnicity (Asian, Black, Hispanic, other) were more likely to be both under and overtriaged compared to White.

Undertriage was more likely if you were older (>6 years), had comorbidities, or arrived by ambulance or during normal office hours.

Overtriage was more likely if you were younger (<3 years), had no comorbidities, or arrived as a walk-in (no ambulance) during off hours.

Triage is Messy

We must acknowledge that the process of triage is by nature, imperfect. Triage decisions are often made very quickly and based on very little information with only a cursory exam. I don’t think it is realistic to be 100% accurate. We must give credit to our nurse colleagues here who do the triaging because it is hard.

One of the variables considered by the authors was the Pediatric Comorbidity Index which was based on the patients’ prior ED, inpatient, and intensive care utilization, and other coexisting illnesses. We doubt the nurses were consistently able to access this information. They are likely too busy to be pulling all these data points. Also, the patient or family’s report of other medical conditions may not always be accurate in the triage setting.

Another example of the uncertainty can be seen in their Figure under the Level 2 section. They ended up combining ESI 2 and 3 because their “algorithm did not create rules for differentiating between optimal assignments of all assigned ESI 2 and 3 visits.”

The other limitation of their study and triage itself is that disease processes are dynamic. They can evolve with time. In general, modifying an assigned ESI level is discouraged, but depending on how long the patient has had to spend in the waiting room, maybe their initial assignment of ESI 3 or 4 has gradually worsened, and they needed more resources and interventions. They did not report whether they adjusted for this. If a patient initially triaged to a lower ESI level got sicker while waiting, that doesn’t necessarily mean the initial assigned triage level was inaccurate.

Variation in ED Pediatric Volumes and Nursing Triage Experience/Training

There were 21 sites included in this study with pediatric volumes ranging from 3,600-25,000 visits per year. There was quite a range in the experience and training of the nurses who performed the triage. The details of this were not recorded by the authors, but they did state that it ranged from a minimum of 4 hours training to at least 1 year of ED clinical experience for others.

We would imagine that both the volume of pediatric visits and the experience of triage nurses contribute to inconsistencies in triage decisions but unfortunately, we can’t see this sub-analysis based on experience or based on site.

Definitions for Under and Overtriage

One of the reasons why this study may have had different rates of mistriage compared to previous studies was their definitions for under and overtriage. The authors developed their definitions through a Delphi process that included emergency medicine, pediatric emergency medicine, and critical care physicians and emergency nurses. Their system included critical interventions and resource use that you can see some examples of in the supplement.

They did several rounds of manual record review to validate these definitions. But this was a group of 8 people. Would other physicians or researchers agree with these definitions? Are some of their metrics more institution or healthcare system-specific?

Consultations and Procedures

The authors did not include specialty consultations and simple procedures (laceration repair, Foley catheter) as resources used as they were not consistently available. It’s unclear how this may have affected their over and undertriage rates. Because these additional resources were not included, it’s possible that some kids who were deemed “overtriaged” were appropriately triaged. It’s also possible that some children who were deemed appropriately triaged were under-triaged.

Clinically Significant Mistriage

The study focuses on triage accuracy rather than patient-oriented outcomes such as mortality or ICU admission. Not all mistriage is equivalent. For example, we see that in their study there were a portion of ESI level 4 patients who were classified as ESI 5. But does this matter? Did harm come to the patient? Did they require one more resource than anticipated, is that such a bad thing?

In contrast, the same ESI level 4 or 5 patient that ended up being a level 1 or 2 who needed more timely interventions and even admission are examples of clinically significant mistriage that can affect patient-oriented outcomes.

Comment on Authors’ Conclusion Compared to SGEM Conclusion: We agree with the authors’ conclusion. While we believe that ESI levels are not always accurate for triaging children, we’re not sure about the rate of mistriage that is reported in this study.

SGEM Bottom Line: We need better and more accurate methods for triaging children in the emergency department.

Case Resolution: You tell the medical director that ESI levels are not always accurate for the triage of pediatric patients. However, not all mistriage has clinically significant consequences. Rates of mistriage may be highly variable based on the triage nurse’s experience and training with pediatric patients. At this time, you recommend directing efforts towards identifying factors in previous instances when children were mistriaged that resulted in harm and focusing efforts to address those factors.

Clinical Application: ESI levels are the most popular method of ED triage in the United States. They are not always accurate, but that may also be due to the nature of triage. Accurate triage helps with prioritizing patients and allocating resources. When evaluating triage, we should focus on capturing cases of clinically significant mistriage.

What Do I Tell My Medical Director?

The ESI system is one of the most popular in the United States. You’re right that it isn’t always accurate for the triage of pediatric patients. I don’t know of any specific triage system to recommend over the ESI system currently as the triage process itself is challenging because we’re making decisions based on little information in a short period. I think we should look at the cases of mistriage that resulted in clinically significant harm and see what factors may have impacted those. In the meantime, I will keep an eye out for any further studies that may suggest that there’s a better way to triage kids.