Objective To investigate the performance characteristics in children with moderate and minor injuries of prehospital paediatric triage tools currently in use in England for identifying seriously injured children.
Methods Eight prehospital paediatric triage tools were identified from literature review and a survey of the lead trauma clinicians across the 10 English strategic health authorities. Retrospective clinical data from 2934 patient records collected by four emergency departments were used to analyse each tool. A target sensitivity of >95% and specificity of 50–75% was set based on the literature.
Results Three tools (East Midlands, North West and Northern) demonstrated acceptable sensitivity (all 100%). The other five tools fell below the target sensitivity of >95%. All eight tools had acceptable specificity (with results between 79% and 99%).
Conclusions Three tools (East Midlands, North West and Northern) demonstrated acceptable over- and under-triage rates in this population of minor and moderately injured children. All tools reached recommended standards for over-triage, but the majority favoured under-triage.
- Emergency Department
- Major Trauma Management
- Paediatric Injury
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
There is no consensus on which triage tool to use for major trauma in children. There are several tools available and in development. This study builds on previous work assessing the accuracy of the existing triage tools in England, on a population of severely injured children.
Most prehospital triage tools used, or being developed for use, in England fail to meet recommended criteria for rates of under- and over-triage. Over-triage rates are particularly problematic in moderately to severely injured children, and under-triage rates are more of an issue for minor to moderately injured children.
As paediatric regional trauma networks become established in England, prehospital triage is likely to become increasingly important. Accurate triage tools may help to mitigate some of the challenges of the new organisation of care.1
Under the regional trauma network system, a seriously injured child is taken directly to a designated major trauma centre (MTC), unless it requires time-critical or life-saving interventions. Minor or moderately injured children, or those needing time-critical treatments, will be treated at a trauma unit. The trauma unit will have resources to provide care to most injured children and adults, except for the most severely injured. After initial treatment or stabilisation, patients can be transferred to an MTC for definitive or secondary treatment as necessary.2
The relatively low incidence of major trauma in children means that even the most experienced prehospital emergency practitioners may find it challenging to assess injury severity to determine where best to send an injured child.1 ,3 It is important to ensure that the decision is correct, so that the most seriously injured children are diverted to centres with the necessary expertise to manage their injuries; it also ensures that MTCs are not overwhelmed by children with minor or moderate injuries. In other words, it is important to avoid under- or over-triage. Under-triage occurs when a patient requiring a higher level of trauma care is triaged to receive care at a lower-level unit, and is of concern because this may result in death or significant morbidity for that seriously injured child. Over-triage results when a patient is taken to an MTC when their injuries could have been treated at a lower-level centre. Over-triage can lead to the depletion of specialist resources that should be deployed to truly severely injured children.4–8 There is currently no national or international consensus on an optimal prehospital triage tool for injured children.
This study builds on previous work using trauma registry data based in the Trauma Audit and Research Network (TARN) to assess the performance characteristics of paediatric prehospital trauma triage tools currently in use or under development in England. Previous work found that none of the tools met recommended criteria for over- and under-triage rates when interrogated against trauma registry data.4 ,9
Datasets from the trauma registry only take into account moderately to severely injured children.10 This study more thoroughly scrutinises the triage tools using retrospective clinical data from four emergency departments in England, to assess their validity in children with more minor and moderate injuries—in particular, to determine whether the high over-triage rates identified for severely injured children were common to the more moderately injured population, as this would have implications for resources and cost.
A survey of lead clinicians for trauma care across the 10 strategic health authorities in England, carried out in August 2011, identified seven prehospital triage tools for injured children currently in use; of these, two were identical.9 The original literature review, as carried out by the first two authors using PubMed and Google Scholar in August 2011, also identified other published prehospital triage tools for children, two of which were described in sufficient detail to allow analysis using historical clinical data but were also designed specifically for use in the prehospital setting, for children and trauma. These tools were used for analysis to compare with our previous work, although we acknowledge that many of the tools are likely to have been refined since that time.
The full list of tools included in the analysis is displayed in table 1.
Retrospective clinical data from four emergency departments in England were used to interrogate each of the triage tools. Analysis was performed using anonymised data that were routinely collected. Of the emergency departments, one represented a children's MTC (Birmingham Children's Hospital), one a combined children's and adult MTC (St Mary's Hospital, London), and the other two were trauma units but with specialised children's services (Bristol Royal Hospital for Children and Leicester Children's Hospital). They also represent a reasonable geographical spread across England.
The Injury Severity Score (ISS) was used to identify children with serious injuries who would require treatment at an MTC. The ISS is a well-recognised anatomically based scoring system for the assessment of injury severity: the higher the score the more severe the injury, with a score >15 being classified as a major or polytrauma.11
Inclusion criteria were people aged below 16 years sustaining injury or trauma between January 2011 and June 2012 who attended the emergency department without using an ambulance. Key discriminators used to analyse the prehospital triage tools were the same as in Cheung et al9 and are shown in table 2. Where discriminator data were incomplete, they were assumed to be within normal ranges.12
Each triage tool was broken down into constituent discriminators (such as physiological signs or mechanism of injury). The constituent discriminators for each tool were then retrospectively applied to the dataset of injured children from each emergency department to derive sensitivity, specificity, positive predictive value, negative predictive value and likelihood ratio measures for each tool's efficacy in identifying seriously injured children (defined as those with ISS >15). A future study will aim to look at each discriminator in turn and its predictive value.
Where discriminators describe physiological values relative to a ‘normal’ range, Advanced Life Support Group guidelines ranges were used.13
Data were analysed using SPSS (V.16) and STATA (V.11) statistical software.
In total, 3003 patient records were identified that fitted the inclusion criteria; 69 did not have an ISS recorded so were excluded, leaving 2934. Of these, four were recorded as having an ISS >15. Demographic data for the study population are shown in table 3.
As in the previous study, the paediatric triage tape was analysed only against children who had an estimated or recorded weight under 32 kg, as the tape is only suitable for children up to 32 kg in weight. A total of 1811 children were identified as weighing <32 kg, and, of these, four had an ISS >15.
The results for each tool are shown in table 4.
In this population of children with minor to moderate injuries, the prehospital triage tools in common use or in development across England all show high specificity, with the majority sacrificing this for sensitivity. Five of the eight tools analysed (London, South West London/Surrey, Wessex, paediatric trauma score and paediatric triage tape) show a tendency to under-triage.
Using ISS level >15 as the criterion for measuring sensitivity and specificity, target levels for under-triage of 0–5% (ie, sensitivity >95%) and over-triage of 25% or less (ie, specificity of 50–75% or more) have been defined by the American College of Surgeons Committee on Trauma.4 ,8
Three of the tools fulfilled acceptable criteria for both over- and under-triage (East Midlands, North West and Northern), with over-triage rates of 18%, 21% and 19%, respectively, and under-triage rates of 0% in each.
All of the eight triage tools analysed here fulfilled the target criterion for over-triage. This is important to ensure MTCs and prehospital services are not overloaded in terms of cost and capacity, which can result in compromise of clinical care and outcomes. However, under-triage, as seen in the other five tools, which all underestimate the severity of trauma in an unacceptable proportion of injured children, has implications for increased mortality and morbidity in this vulnerable group.
As in our previous study in which we analysed the tools against trauma registry data, this paper uses the classical definition of (1−sensitivity) for under-triage and (1−specificity) for over-triage.4 Other studies have used (1−negative predictive value) for under-triage and (1−positive predictive value) for over-triage.14 ,15 If we apply this method to our dataset, all the triage tools fulfil under-triage rates, but they all demonstrate a tendency to over-triage. This is likely to be different because positive and negative predictive values depend heavily on prevalence. In our population data, the prevalence of severely injured children (ISS >15) is low (0.14%). Such a low prevalence and high specificity results in a high negative predictive value and hence a low under-triage rate; however, the positive predictive value will be low and results in the higher over-triage rate. On the basis of this study, the tools need significant modification and validation, which corroborates the findings from our previous study.9 It is our intention that the next stage of the study will be to identify key discriminators to inform the development of any future paediatric prehospital triage tools.
The results of this study and our original study analysing the triage tools against trauma registry data suggest that the tools in use or in development in England tend to over-triage children with moderate to severe injuries, yet under-triage those with minor to moderate injuries. This is perhaps acceptable for those children at the extremes of the spectrum, but there is a concern that there will be a proportion of children who are triaged inappropriately to a centre that is not optimised to manage their injuries.
Strengths and limitations of the study
The use of emergency department data has allowed us to assess the efficacy of each triage tool against a large population of children. However, data have been collected from only four departments, which does not take into account geographical or historical differences in local trauma management. It is worth noting that one of the four departments (St Marys) was already functioning as an MTC at the time of data collection, as London was the first area to establish Regional Trauma Networks, but the other three departments were not working to that model when the data were collected. It is possible that this may have had an effect on trauma care over the time period analysed.
As with our original study, the triage tools may well have undergone modification since they were collated as each network learnt from their experiences, which may alter our findings. There is again the potential for bias in applying discriminator data from various triage tools retrospectively to a set of data points already collected. We used the same clinical interpretations of discriminators as in the original study.
Given that there were only four cases with ISS >15 in this population of minor to moderately injured children, the CIs for the under-triage values are very wide. This should be taken into account when interpreting the data on under-triage. As the incidence of major trauma is so low, much larger datasets would be needed to pick up more positive test results and therefore improve the accuracy of the analysis.
A possible selection bias was introduced by including only children who arrived at the emergency department without the use of an ambulance. It may be assumed that this cohort of patients had less severe injuries. Including data from ambulance patient records would have made this section of the study more robust in terms of eliminating the selection bias. It is the prehospital teams who will be using the tools being evaluated in the study, so it would have been more relevant to have included these data in our work.
The results of this study are intended to be used alongside our earlier results obtained using trauma registry data to analyse the triage tools.9 The present study analysed the tools using a population of children with mild–moderate injuries—as determined by ISS <15—compared with the population of severely injured children in the original study. In this way, all children affected by trauma, from those with minor to severe injuries, are taken into account. To complete the assessment and validation of these tools, however, they also need to be tested prospectively in the prehospital environment for which they are designed. We intend to pursue a prospective, unselected analysis of injured children in the next phase of our study including the use of ambulance patient record data. That study will also aim to break down the individual parameters of each triage tool to identify which are most predictive.
Most prehospital triage tools used, or being developed for use, in England fail to meet recommended criteria for rates of under- and over-triage. Over-triage rates are particularly problematic in moderately to severely injured children, and under-triage rates are more of an issue for minor to moderately injured children. This has major implications for regional trauma networks for children in England in terms of both operational, resource and safety issues and the potential for mortality and morbidity. This paper supports our original finding that there is an urgent need for the development of evidence-based tools to allow accurate risk stratification of injured children in the prehospital setting.
Thank you to Catrin Dyer, Isabel Foster and Mary Reeves for their input into data collection for the Birmingham cohort. Work was conducted on behalf of PERUKI (Paediatric Emergency Research in the UK and Ireland). This is a collaborative organisation which has the aim of improving emergency care for children through robust multicentre research. There are 39 member sites, representing a mix of generic and paediatric-specific emergency departments. Prehospital triage of injured children has been identified as a priority area for research in our setting by PERUKI, and this study was performed as part of that work stream.
Contributors AA and CRC designed the project protocol, collated the data and wrote the paper. TL and OB conducted most of the data analysis. FL, KB, ML and DR helped to obtain data and to write up the paper at various stages throughout the process; they all also helped with data interpretation. NK, SI and AC collected data and assisted with the interpretation and analysis of these data. IM acts as guarantor for the work having had the original idea for the project, designed the project protocol and helped with all aspects of the work at various stages. All authors were asked to comment on and alter the paper before submission. They all reviewed and revised the manuscript and approved the final version submitted.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.