Introduction
Disasters occur with little or no warning and often overwhelm medical response and resources. The purpose of triage in disaster is to categorize and prioritize patients based on urgency with the ultimate goal of saving as many lives as possible. Pediatric disaster triage offers unique challenges, including lack of provider familiarity with pediatric patients and physiology, children with complex health needs, differing development and psychological states, and heightened emotions.Reference Koziel, Meckler and Brown1, Reference Lozon and Bradin2
Although initial triage usually occurs at the scene of the event, triage is a dynamic process and occurs in various settings. It is important for clinicians at all levels to be familiar with and apply a standardized disaster triage system. Although there is growing interest in disaster medicine among pediatric trainees, especially in the context of the COVID-19 pandemic, disaster medicine has not yet been incorporated into pediatric residency training.Reference Boggs, Goodwin and Simpson3, Reference Cicero, Blake and Gallant4 Additionally, there is high variability in how disaster education is delivered.Reference Sarin, Cattamanchi, Alqahtani, Aljohani, Keim and Ciottone5, Reference Ngo, Schertzer, Harter and Smith-Coggins6
We sought to evaluate the consistency of triage decisions made by physicians of various training levels (residents, fellows, attendings) using the Sort, Assess, Life-Saving Intervention, Treatment/Transport (SALT) triage system7 at our institution. We chose to use SALT triage because it was derived from consensus opinion by national experts from multiple medical societies and incorporated aspects of existing triage systems. Additionally, SALT has been shown to have similar accuracy compared to popular pediatric triage systems such as JumpSTARTReference Jones, White and Tofil8, Reference Nadeau and Cicero9 and could be learned and applied quickly by paramedics or laypersons.Reference Cone, Serra, Burns, MacMillan, Kurland and Van Gelder10, Reference Celik, Mencl, Debacker, Kurland, Wilber and Frey11
Methods
This project was submitted to the Children’s National Hospital Institutional Review Board (IRB) and deemed to be quality improvement and not human subjects research, thus exempt from oversight of the Institutional Review Board.
Creation of Fictional Triage Patients
Descriptors for 20 fictional triage patients were created and used in the pre- and post-test triage activity (Figure 1). Particular care was taken to include patients who were infants and children and youth with special health care needs (CYSHCN). The investigators (DR, TG, JS) reached a consensus on the accurate triage category (green, yellow, red, black) on the appropriate triage category for each fictional patient based on existing expert opinion and literature.Reference Lerner, McKee and Cady12, Reference Donofrio, Kaji and Claudius13 The 20 patients were triaged into the following categories: 7 green, 5 yellow, 6 red, 2 black. A full list of the fictional triage patients and consensus triage categories is available in Supplemental Appendix.
Figure 1. Sample Triage Patients.
Recruitment of Participants
Pediatric residents, pediatric emergency medicine fellows, and pediatric emergency medicine attendings from our institution were recruited to participate via email. Attendings are doctors who have completed post-graduate training. Fellows are doctors in training to become attending physicians. Residents are medical school graduates in post-graduate training. Participation was voluntary. Pediatric residents were offered elective credit for their participation.
Just-in-Time Triage Module
We created a pre-test, narrated didactic module on SALT triage, and post-test. During the pre-test triage activity, participants were tasked with triaging 20 fictional pediatric patients. To add a realistic level of stress, participants were given an average of 20 seconds to designate a triage category for each patient. Afterward, they listened to a narrated didactic module that covered basic concepts of triage, the SALT triage system, and pediatric limitations of triage systems. Finally, they repeated the same triage activity during a post-test.
Statistical Analysis
We performed statistical analysis with SPSS.14 We analyzed test scores using non-parametric tests and report median percent accuracy. We compared pre- and post-test scores using the Wilcoxon signed rank test and pairwise comparison between groups using the Kruskal-Wallis test, reporting medians and interquartile ranges of triage accuracy.
Results
Demographics
Forty-eight participants enrolled in the module: 27 pediatric emergency medicine (PEM) attendings (56%), 9 PEM fellows (19%), and 12 pediatric residents (25%). Approximately half (53%) reported previous experience with disaster triage including didactics, courses, and real-life experience.
Pre- and Post-Test Accuracy
The median (interquartile range [IQR]) pre-test percent accuracy across all participants was 75 (65-85). Attendings scored 80 (70-90). Fellows scored 75 (67.5-87.5). Residents scored 62.5 (55-68.8) (Figure 2). Attendings scored significantly higher than residents, P < 0.01, but not higher than fellows. For the 44 participants who completed both the pre- and post-test, median score significantly improved from 75 (65-85) to 80 (75-90), P < 0.01 (Figure 3), with most improvement seen among the resident group (Table 1). There were no significant differences in post-test scores among groups.
Figure 2. Pre-test Triage Accuracy by Training Level.
Figure 3. Pre and Post-test Accuracy (n=44).
Table 1. Pre- and post-test scores by training level
Discussion
We created a just-in-time web-based module to evaluate and improve the accuracy of pediatric triage decisions among PEM attendings, PEM fellows, and pediatric interns and residents. We demonstrated that PEM attendings were the most accurate, although not significantly better than PEM fellows. PEM attendings were significantly more accurate in comparison to pediatric residents. After completing a brief didactic module, accuracy of triage decisions significantly increased across all participants but was most pronounced in pediatric residents.
Limitations
This study was performed at a single academic pediatric institution and results may not be generalizable to other settings. All participation was voluntary, so it is possible that the group of participants was self-selecting in having previous knowledge or interest in disaster triage although only about half reported previous experience. Additionally, most participants were PEM attendings with a smaller number of PEM fellows and residents participating. While there was significant improvement across all participants after completion of the just-in-time triage module, it is possible that this improvement would have been more pronounced with the inclusion of more PEM fellows or pediatric residents. Finally, although we attempted to add a stress component by limiting time to make a triage decision, we recognize that a text-based description of fictional patients may not realistically simulate the multiple stressors of making triage decisions during a real disaster. Future efforts should go towards increasing the fidelity of the triage activity to better replicate real-life disaster scenarios and expanding this educational offering to additional health care professionals including pre-hospital and volunteer medical personnel as they are likely to be making or assisting in the initial triage decision at the scene of the disaster.
Conclusion
There is variation in accuracy of triage decisions among clinicians of different training levels. An online module is a feasible method of delivering just-in-time triage training that improves accuracy of disaster triage decisions for pediatric patients, especially among pediatric residents.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/dmp.2025.10268.
Author contribution
Drs. Ren, Goodwin, and Simpson conceived and designed the study. Dr. Ren drafted the manuscript and collected the data. Dr. Breslin conceived and performed the data analysis. All authors reviewed and approved the final manuscript prior to submission.
Competing interests
The authors declare no conflict of interest or funding for this study.
Open access
