• 
• 

Three Part Question

In [children involved in major incidents] do [paediatric major incident triage tools] identify patients requiring [life-saving interventions]

Clinical Scenario

You are part of the response team on route to a multi-vehicle road traffic collision. Initial reports state that one of the vehicles is an overturned minibus carrying 12 children, with all sustaining varying degrees of injury. Aware that you have limited clinical resources immediately available you consider whether a paediatric major incident triage tool would help to prioritise patients needing life-saving interventions.

Search Strategy

Medline 1946 - June week 3, 2019 and EMBASE 1974 - June week 3, 2019.
[(paediatr* OR pediatr* OR child OR adolsc* OR infan*).ti,ab AND (triage OR triage tool OR triage tape OR major inciden* OR mass casualt*).ti,ab AND (trauma* OR injur* OR wound* OR polytrauma*)]

Search Outcome

The initial search produced 750 papers; review of title and abstract identified 65 papers for review. Following full review, 49 were excluded as they were not directly relevant to the research question. The remaining 16 papers are presented below.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Ramenofsky et al. 1988 United States	496 children aged ≤16 seen by emergency services in Alabama	Retrospective cohort study	Ability of Pediatric Trauma Score to identify serious injury	Sens 95.8%, spec 98.6%. No deaths with PTS score >8.	Unclear methodology to determine serious injury against which statistics were generated. Data collection 1985-1986 with likely improvement in trauma outcomes.
Eichelberger et al. 1989 United States	1334 patients aged 0-14 with blunt and penetrating trauma	Prospective cohort study	Ability of Triage Score, Revised Triage Score and Pediatric Trauma Score to identify major trauma (based on ISS ≥15).	TS ≥14 - sens 72%, spec 75%. RTS ≥11 - sens 73%, spec 74%. PTS ≥8 - sens 78%, spec 75%.	Data collection 1985-1987 with likely improvement in trauma outcomes.
Kaufmann et al. 1990 United States	376 children aged <15	Retrospective cohort study	Accuracy of Pediatric Trauma Score (PTS) and Revised Trauma Score (RTS) as defined by the total number of children triaged correctly by PTS or RTS divided by the total number of injured children.	Accuracy: PTS 68.3%, RTS 78.8%. Mortality in PTS ≤8 13%, PTS >8 0%.	Data collection 1985-1987 with likely improvement in trauma outcomes. Single centre. Triage accuracy definition.
Balik et al. 1993 Turkey	533 injured children aged under 17 presenting between 1984-1989.	Retrospective cohort study	Ability of the Pediatric Trauma Score (PTS) to predict mortality	PTS ≤8 had a mortality 24/163 (14.7%). PTS >8 had a mortality 3/370 (0.8%).	Limited statistical analysis. Single centre. Data collection 1984-1989 with likely improvement in trauma outcomes. Mortality used as a surrogate marker.
Wallis and Carley 2006 South Africa	3461 children ≤12 years old presenting to a major trauma centre within 12 hours of injury	Prospective cohort study	Ability of Paediatric Triage Tape to identify T1 (immediate) patients, as defined by patients with ISS/NISS 16 or above, and one of more Garner criteria.	ISS: Sens 37.8%, spec 98.6% NISS: Sens 26.1%, spec 98.9% Garner: Sens 41.5%, spec 98.9%	Single centre. Aged ≤12 only.
Wallis and Carley 2006 South Africa	3461 patients ≤12 years old presenting with acute injury to a major trauma centre.	Prospective cohort study	Ability of triage algorithms to discriminate T1 (immediate priority) and not-T1 (urgent or delayed priority), with sensitivity and specificity measured against ISS, NISS and modified Garner criteria.	T1 (based on ISS > 15): Paediatric Triage Tape (PTT) - Sens 37.8% Spec 98.6% CareFlight - Sens 48.4% Spec 98.8% JumpSTART (JS) - Sens 3.2% Spec 97.8% START - Sens 31.3% Spec 77.9%. T1 (based on presence of one of more modified Garner criteria) PTT - Sens 41.5% Spec 98.9% CareFlight - Sens 46% Spec 98.9% JumpSTART - Sens 0.8% Spec 97.7% START - Sens 39.2% Spec 78.7%	Most prospective triage decisions would have been based on small numbers of casualties therefore preventing true 'major incident' triage. This is partially overcome by the hypothetical scenarios (outcome two). Single centre, in a more injured population.
			Accuracy of triage scores in hypothetical major incidents with 10%, 30% and 60% T1 casualties.	10% T1 (based on ISS and Modified Garner Criteria, respectively): PTT 93% and 90% Careflight 94% and 94% JS 88% and 88% START 73% and 75%. 30% T1 (based on ISS and Modified Garner Criteria, respectively): PTT 80% and 81% Careflight 84% and 83% JS 69% and 92% START 64% and 67%. 60% T1 (based on ISS and Modified Garner Criteria, respectively): PTT 63% and 65% Careflight 69% and 68% JS 41% and 87% START 50% and 55%.
Nasr et al. 2007 Canada	628 patients <16 years old sustaining blunt trauma	Retrospective cohort study	Ability of the modified Pediatric Trauma Score (mPTS) and the Sick Kids PTS to identify high risk trauma, as defined by ISS >12.	mPTS: Sens 92%, spec 29%. Sick Kids PTS: Sens 99%, spec 21%.	Use of ISS as an output measurement; with high risk defined as ISS >12, making it difficult to compare with other studies. Included blunt injury only, but made calculations for reduction in trauma calls (which penetrating trauma and burns were indications for activation).
Cross and Cicero 2013 United States	530,695 patients including 15,114 patients aged 0-8 years and 21,781 patients aged 9-15 on the National Trauma Data Bank.	Retrospective cohort study	Accuracy of triage tools (START or JumpSTART, FDNY [Fire Dept of New York], CareFlight, Glasgow Coma Score, Sacco, unadjusted Sacco) based on mortality at hospital disposition as primary outcome; secondary measures included death in ED, ventilator use.	The Sacco score most accurately predicted mortality. FDNY performs poorly in children under 8. AUC for mortality at discharge aged 0-8: START 0.931, FDNY 0.891, CareFlight 0.9552, GCS 0.964, Sacco 0.961, Unadjusted Sacco 0.963. AUC for mortality at discharge aged 9-15: START 0.942, FDNY 0.946, CareFlight 0.955, GCS 0.949, Sacco 0.961, Unadjusted Sacco 0.958.	The use of AUC for comparison gives clear statistical information but is vague about the clinical relevance of differences. Study not specific to paediatric population. Use of mortality as a primary outcome, with less relevant secondary outcomes.
Cheung et al. 2013 United Kingdom	701 injured patients <16 years old admitted direct from the scene of accident	Retrospective cohort study	Ability of paediatric triage tools to identify patients with major trauma centre need (defined as ISS >15)	Paediatric Trauma Score - Sensitivity 39% (under-triage 61%); specificity 93% (over-triage 7%). Paediatric Triage Tape - Sensitivity 36% (under-triage 63%); specificity 84% (over-triage 16%). Six tools used regionally were measured: East Midlands: sens 97%, spec 17%; London: sens 96%, spec 28%.	Potential for bias when applying discriminator data from various tools retrospectively. Use of ISS as an output measurement. Again, no delineation of results based on age populations; Price et al. 2016 demonstrated difference in sensitivity and specificity in patients aged <8 or >8 years old.
Jones et al. 2014 United States	Simulated paediatric patients in moulage for 43 paramedics.	Prospective control study	Ability of triage tools (JumpSTART and SALT [sort, assess, lifesaving intervention, treat/transport]) to accurately triage patients in a simulated mass casualty incident.	Accuracy of triage 66% for both tools. Over-triage 23% for both tools. Under-triage 10% for SALT, 11% for JumpSTART.	Simulation based study. Small number of participants, with differences between study groups.
Ardolino et al. 2015 United Kingdom	2934 patients < 16 years old sustaining injury, who attended the emergency department without using an ambulance.	Retrospective cohort study	Ability of paediatric triage tools to identify patients with major trauma centre need (defined as ISS >15).	Paediatric Trauma Score - Sensitivity 0% (under-triage 100%), specificity 99% (over-triage 1%). Paediatric triage tape: Sensitivity 25% (under-triage 75%), specificity 97% (over-triage 3%). Six tools used regionally were measured: East Midlands, North West and Northern all demonstrated 100% sensitivity; all regional tools had specificity 79-93%.	Only four patients had ISS >15, resulting in wide confidence intervals. Potential for bias when applying discriminator data from various tools retrospectively. Use of ISS as an output measurement. No delineation of results based on age populations; Price et al. 2016 demonstrated difference in sensitivity and specificity in patients aged <8 or >8 years old.
Newgard et al. 2016 United States	2832 children (aged ≤14 years old) transported to hospital by EMS (from total 53,487 patients of all ages in study).	Prospective cohort study	Ability of National Field Triage Guidelines to identify serious injury, defined as ISS ≥ 16	Sensitivity in aged 0-14 years: 87.4% (9.6% of those identified by triage had ISS ≥ 16).	It is not possible to evaluate secondary outcomes e.g. early critical resource use for paediatric population due to lack of age separation in statistical analysis. Use of ISS as an output measurement.
Price et al. 2016 United Kingdom	31,292 injured patients <16 years old, based on Trauma Audit and Research Network data.	Retrospective cohort study	Ability of triage tools to accurately distinguish between 'immediate' and 'delayed' priority patients, based on correlation against survival, and ISS >15.	Mortality - (sensitivity and specificity, respectively) Paediatric Triage Tape - 37.8%, 66.0% Triage Sort - 96.2%, 69.6% JumpSTART / START - 91.8%, 70.5% CareFlight - 95.3%, 80.4% ISS >15 - (sensitivity and specificity, respectively) PTT - 36.4%, 66.5% Triage sort - 70.6%, 78.1% JumpSTART / START - 59.6%, 76.3% CareFlight - 64.5%, 89.8%	It was necessary to make certain clinical assumptions in order to apply triage tools to retrospective dataset e.g. weight estimates based on age; unclear whether physiological parameters were based on pre- or post- treatment initiation. Use of ISS as an output measurement.
Lerner et al. 2017 United States	5610 injured children (<15 years old) transferred to a major trauma centre.	Prospective cohort study	Accuracy of Physiologic criteria of the Field Triage Guidelines in identifying major trauma centre need, determined by non-orthopaedic surgery within 24 hours, ICU admission, or death.	51% of children requiring trauma centre resources were under triaged; 18% over triaged.	Surrogate markers of trauma centre need. Use of interviews.
Toida et al. 2018 Japan	137 patients <16 years old	Retrospective cohort study	Ability of Paediatric Physiological Anatomical Triage Scoring System (PPATS) for predicting triage priority as 'immediate'	95.8% sens, 86.7% spec, 60.5% PPV, 99.0% NPV	Immediate priority determined by area under the curve statistics, not by ISS / NISS / Garner criteria. Retrospective study, small cohort. Secondary triage tool, not primary.
			Association between PPATS and predicted mortality rate, ventilation time, ICU stay, hospital stay	Regression analysis showed a significant association between PPATS and predicted mortality (r2 = 0.139), ventilation time (r2 = 0.320), ICU stay (r2 = 0.362), hospital stay (r2 = 0.308).
Heffernan et al. 2019 United States	115 patients aged under 18 presenting to a single trauma centre	Prospective cohort study	Accuracy of triage tools to assign patients to the correct category, as defined by an expert panel's criterion standard definition of categories.	Correct triage percentage: SALT: 59.1% JumpSTART: 56.5% Triage Sieve: 55.7% CareFlight: 55.7%	Small cohort. Inclusion criteria to 18 years old; other studies have typically used 14-16 years old as an upper limit.

Comment(s)

A number of paediatric major incident triage tools have been presented in the literature, each with limited data to support either their development of their performance. There are no studies to date utilising data derived from major incidents - the setting in which we are expecting to use the tools; instead trauma registries or data from consecutive trauma patients are used as surrogates. Both of these surrogates are associated with limitations. With retrospective studies, the injury pattern observed following a major incident may not reflect that which is recorded on the databases. Additionally, incomplete data may result in selection bias when analysing retrospective records. Furthermore, when performing the analysis on both surrogate categories, we are most frequently applying the triage tools to the recorded physiological parameters in isolation, rather in the context of any treatment that may have been given (e.g. fluids or analgesia that might influence haemodynamic stability). Therefore, this may not truly reflect the dynamic nature of triage that would be encountered within a traditional major incident. Whilst prospectively collected data from trauma patients may be more accurate in terms of completeness of data, the triage tool is performing in a different setting to that in which we are expecting it to function - this limits our ability to validate major incident triage tools. Paediatric trauma generates further challenges for triage tools to acknowledge, related to the variety of physiological norm based on patient age. There is a lack of standardised outcome to allow comparison of studies, with numerous surrogates used to constitute trauma centre need. Commonly, the outcome measure determining trauma centre need was the Injury Severity Score (ISS), which itself is a retrospective measure. This marker is an established definition of major trauma, but may not represent the optimal marker for trauma centre need; this correlation is yet to be fully elucidated, with evidence that ISS performance is different in paediatric populations.(17) Current NHS England clinical guidelines for major incident advocate the use of JumpSTART in emergency department paediatric triage(18); the ambulance service use Paediatric Triage Tape in accordance with National Ambulance Resilience Unit guidance.(19) There is limited evidence to support the use of these tools.

Editor Comment

J of Trauma

Clinical Bottom Line

Within the UK, two forms of paediatric major incident triage are used, the JumpSTART method and the Paediatric Triage Tape. Both demonstrate poor performance at identifying patients in need of life-saving intervention. There is limited evidence to support the use and validity of existing paediatric major incident triage tools. Further research is required in order to attempt to derive and validate a triage tool that is able to identify children in need of life-saving interventions.

References

1. Ramenofsky ML, Ramenofsky MB, Jurkovich GJ, et al. The predictive validity of the pediatric trauma score. J Trauma. 1988 Jul;28(7):1038-42.
2. Eichelberger MR, Gotschall CS, Sacco WJ. A comparison of the trauma score, the revised trauma score, and the pediatric trauma score. Ann Emerg Med. 1989 Oct;18(10):1053-8.
3. Kaufmann CR, Maier RV, Rivara FP, et al. Evaluation of the pediatric trauma score. JAMA. 1990 Jan 5;263(1):69-72.
4. Balik E, Ozok G, Ulman I. Pediatric trauma score: is it reliable in predicting mortality? Pediatr Surg Int. 1993;8:54-55.
5. Wallis LA, Carley S. Validation of the paediatric triage tape. Emerg Med J. 2006 Jan;23(1):47-50.
6. Wallis LA, Carley S. Comparison of paediatric major incident primary triage tools. Emerg Med J. 2006 Jun;23(6):475-8.
7. Nasr A, Mikrogianakis A, McDowall D, et al. External validation and modification of a pediatric trauma triage tool. J Trauma. 2007 Mar;62(3):606-9.
8. Cross KP, Cicero MX. Head-to-head comparison of disaster triage methods in pediatric, adult, and geriatric patients. Ann Emerg Med. 2013 Jun;61(6):668-676.
9. Cheung R, Ardolino A, Lawrence T, et al. The accuracy of existing prehospital triage tools for injured children in England - an analysis using trauma registry data. Emerg Med J. 2013 Jun;30(6):476-9.
10. Jones N, White ML, Tofil N, et al. Randomized trial comparing two mass casualty triage systems (JumpSTART versus SALT) in a pediatric simulated mass casualty event. Prehosp Emerg Care. 2014 Jul-Sep; 18(3):417-23.
11. Ardolino A, Cheung CR, Lawrence T, et al. The accuracy of existing prehospital triage tools for injured children in England: an analysis using emergency department data. Emerg Med J. 2015 May;32(5):39
12. Newgard CD, Fu R, Zive D, et al. Prospective validation of the national field triage guidelines for identifying seriously injured persons. J Am Coll Surg 2016 Feb;222(2):146-58.
13. Price CL, Brace-McDonnell SJ, Stallard N, et al. Performance characteristics of five triage tools for major incidents involving traumatic injuries to children. Injury 2016 May;47(5):988-92.
14. Lerner EB, Drendel AL, Cushman JT, et al. Ability of the physiologic criteria of the field triage guidelines to identify children who need the resources of a trauma center. Prehosp Emerg Care. 2017 Mar-Apr;21(2):180-184.
15. Toida C, Muguruma T, Abe T, et al. Introduction of pediatric physiological and anatomical triage score in mass-casualty incident. Prehosp Disaster Med. 2018 Apr;33(2):147-152
16. Heffernan RW, Lerner EB, McKee CH, et al. Comparing the accuracy of mass casualty triage systems in a pediatric population. Prehosp Emerg Care. 2019 May-Jun;23(3):304-308.