observational study
Rogier van der Sluijs, MDa,b , Robin D. Lokerman, MDa, Job F. Waalwijk, MDa,b, Mariska
A.C. de Jongh, PhDc, Michael J.R. Edwards, PhDd, Dennis den Hartog, PhDe, Georgios F.
Giannakópoulos, PhDf, Pierre M. van Grunsven, PhDg, Martijn Poeze, PhDb,h, Luke P.H.
Leenen, PhDa, Mark van Heijl, PhDa,i, for the Pre-hospital Trauma Triage Research
Collaborative (PTTRC).
Institutions and departments
a Department of Surgery, University Medical Centre Utrecht, Utrecht, The Netherlands b Department of Surgery, Maastricht University Medical Centre, Maastricht, The Netherlands
c Network of Acute Care Brabant, ETZ Hospital, Tilburg, The Netherlands
d Department of Surgery, Radboud University Medical Centre, Nijmegen, The Netherlands
e Department of Surgery, Erasmus University Medical Centre, Rotterdam, The Netherlands
f Department of Surgery, Amsterdam University Medical Centre, Amsterdam, The
Netherlands
g Veiligheidsregio Gelderland-Zuid, Nijmegen, The Netherlands
h Network of Acute Care Limburg, Maastricht University Medical Centre+, Maastricht, The
Netherlands
i Department of Surgery, Diakonessenhuis Utrecht/Zeist/Doorn, Utrecht, The Netherlands
Collaborative group
Members of the Pre-hospital Trauma Triage Research Collaborative (PTTRC) are: Koen Lansink (ETZ Hospital Tilburg), Jens Halm (Amsterdam University Medical Centre), Wim Breeman (Regionale Ambulance Voorziening Rotterdam-Rijnmond), Timo Bevelander 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
(Regionale Ambulance Voorziening Zuid-Holland Zuid), Arjen Siegers (Regionale Ambulance Voorziening Ambulance Amsterdam-Amstelland, Regionale Ambulance Voorziening Zaanstreek-Waterland), Risco van Vliet (Regionale Ambulance Voorziening Brabant Midden-West-Noord), Thijs Verhagen (Regionale Ambulance Voorziening Utrecht), Margreet Hoogeveen (Ambulancezorg Nederland), and Leontien Sturms (Landelijk Netwerk Acute Zorg).
Abstract word count: 279 Word count: 3368
Conflict of interest: none.
Previous publication or submission: no related papers from the same study.
Corresponding author:
R. van der Sluijs, MD
Address Suite: Q02.2.301, Heidelberglaan 100, 3584 CX, University Medical Centre Utrecht, Utrecht, The Netherlands
E-mail: r.vandersluijs@umcutrecht.nl Telephone: +31 (0)6 11 91 29 40 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
Research in context
Evidence before this study
Unjustified transport of children in need of specialised care to lower-level and non-paediatric trauma centres (i.e., undertriage) is associated with adverse clinical outcomes. Conversely, overtriage – the transportation of patients without the need for specialized care to higher-level trauma centres – results in excessive costs. We updated our recent systematic review on paediatric prehospital trauma triage and searched MEDLINE, Embase, PsycINFO, and the Cochrane Register of Controlled Trials, from database inception to August 8, 2019. Search terms included paediatric trauma (study population), triage protocol (index test), accuracy (outcome), and field triage (setting). Additional studies were identified by examining the reference lists of the included studies.
No study could be identified that reported on the accuracy of the full triage strategy based on the initial transport destination. Four studies evaluated different steps of the accuracy of multiple combinations of steps from the Field Triage Decision Scheme, but none evaluated the accuracy of the combined physiologic and anatomic criteria (i.e., the only steps that induce an advice to transport the patient to the highest level of care within the system).
Added value of this study
Transporting the right patient to the right hospital is fundamental to the proper functioning of trauma systems. To our knowledge, this is the first large, multi-site, multi-centre study to evaluate the diagnostic accuracy of the full field triage strategy in terms of undertriage and overtriage in paediatric trauma patients. Additionally, this is the first study to externally validate two actively used triage protocols: the National Protocol of Ambulance Services, and the Field Triage Decision Scheme. Nearly all countries with regionalized trauma care use 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67
either the Field Triage Decision Scheme or a similar combination of physiologic, anatomic, and mechanism-related criteria, signifying the generalizability of this study.
Implications of all the available evidence
The premise of regionalised trauma systems is that centralization of patients and resources enables the most efficient trauma care. Currently, one in six-to-seven children in need of specialised trauma care is not transported to the right hospital based on both an anatomical and a resource-based reference standard. These results do not comply with the maximally recommended undertriage rate of <5%. The examined triage decision rules were unable to discriminate between low-risk patients and children in need of specialised trauma care. Efforts should be made to develop a highly sensitive and child-specific triage tool to aid Emergency Medical Services professionals during field triage.
68 69 70 71 72 73 74 75 76 77 78 79 80
Abstract
Background:
Adequate pre-hospital trauma triage is crucial to enable optimal care in trauma systems. Transport of paediatric trauma patients in need of specialised trauma care to lower-level trauma centres is associated with adverse patient outcomes. This study aimed to evaluate the diagnostic accuracy of paediatric field triage based on patient destination and triage tools.
Methods:
This was a multi-site observational study that included children (aged <16 years) transported with high priority by ambulance from the scene of injury to any trauma-receiving emergency department within seven out of 11 inclusive trauma regions in the Netherlands. Diagnostic accuracy based on the initial transport destination was evaluated in terms of undertriage and overtriage. The Dutch National Protocol of Ambulance Services (NPAS) and Field Triage Decision Scheme (FTDS) triage protocols were externally validated using an anatomical (Injury Severity Score ≥ 16) and a resource-based reference standard.
Findings:
Between January 1, 2015 and December 31, 2017, we analysed 12,915 children attending the emergency department with injuries. A total of 129 patients had an Injury Severity Score ≥ 16, and 227 patients used critical resources within a limited timeframe. Ten patients died within 24 hours of admission. One in six-to-seven patients in need of specialised trauma care was not transported to a higher-level paediatric trauma centre. The sensitivity of the NPAS ranged from 50% to 53%. The FTDS had a sensitivity of 60% to 64%. The evaluated specificities were 94 – 94% and 84 – 85% for the NPAS and the FTDS, respectively. 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
Interpretation:
Too many children in need of specialised care were transported to non-paediatric trauma centres, which is associated with increased mortality and morbidity rates. Protocols that are currently used cannot accurately discriminate between low-risk and high-risk patients.
Funding:
The Netherlands Organisation for Health Research and Development, Innovation Fund Health Insurers. 105 106 107 108 109 110 111 112 113
Introduction
Paediatric injuries are accountable for approximately 40% of all child deaths in developed countries worldwide.1 Inclusive trauma systems were established to centralize patients,
resources, and expertise in order to lower mortality rates, life-long disabilities, and costs. The higher the degree of centralization, the greater the consequences of inadequate field triage. Undertriage – transporting severely injured children to facilities without the required
resources and expertise for optimal care (i.e., lower-level trauma centres) – is associated with higher mortality rates.2,3 Conversely, overtriage – the transport of mildly injured children to
higher-level paediatric trauma centres (PTCs) with a surplus of resources – results in excessive costs and exhaustive use of scarce resources.4
Field triage can be perceived as a three-step diagnostic strategy. First, Emergency Medical Service (EMS) professionals must determine a patient’s resource-need on the scene of injury. This can be challenging because of atypical injury presentations in children, limited time, and the lack of diagnostic modalities. Second, logistical constraints need to be considered, such as trauma centre capacity, trauma centre proximity, and patient acuity. The third step is to determine the optimal transport destination in light of steps one and two.
Field triage tools can assist EMS professionals in the assessment of injury severity and subsequent resource-need. Allocation of injured children is guided by the National Protocol of Ambulance Services (NPAS) in the Netherlands.5 This protocol is partly derived from the
Field Triage Decision Scheme (FTDS), established by the American College of Surgeons
Committee on Trauma (ACSCOT).6 The FTDS or similar combinations of physiologic,
anatomic, and mechanism-related criteria are universally implemented in regionalised trauma 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
systems across the world. The NPAS was recently evaluated for a single inclusive trauma region in the Netherlands and could not adequately select severely injured adults, with a sensitivity of only 36.2%.7 Furthermore, a recent systematic review showed that few existing
triage tools were child-specific and no single tool was able to attain an undertriage rate of less
than 5%, as recommended by the ACSCOT.68 Ultimately, no evidence was available on
triage accuracy based on the transport destination of injured children: the critical question whether children requiring specialised care are in fact transported to higher-level PTCs remained unanswered.
Accurate field triage is fundamental to properly functioning trauma systems. We designed the Paediatric Pre-hospital Trauma Triage (P2-T2) study to evaluate the quality of paediatric field triage in multiple EMSs and inclusive trauma regions in the Netherlands, based on protocol accuracy, protocol compliance, and destination-based mistriage rates.
138 139 140 141 142 143 144 145 146 147 148 149 150 151
Methods
Study Design and Setting
The P2-T2 study was a multi-site, observational diagnostic study that aimed to evaluate the quality of paediatric trauma field triage in the Netherlands. This study was reported in accordance with the Standards for Reporting of Diagnostic Accuracy Studies (STARD) guidelines.9 This study received an approval letter from the Institutional Review Board of the
University Medical Centre Utrecht that confirmed that the Medical Research Involving Human Subjects Act did not apply.
Children transported by eight different EMSs (Utrecht, Rotterdam-Rijnmond, Zuid-Holland Zuid, Amsterdam-Amstelland, Zaanstreek-Waterland, Brabant Midden-West, Brabant-Noord, and Gelderland-Zuid) were included between January 1, 2015 and December 31, 2017. These EMSs transport approximately 390,000 high-priority patients annually, within an 8,063 km2 region, with a population of 6·5 million people. All EMSs in the Netherlands are
provided by the government in partnership with private companies. Pre-hospital care systems are required to comply with the protocols of the Dutch Institute of Ambulance Care. All ground ambulances are staffed by a nurse, which is licensed to administer medical treatment at advanced life support level, and a dedicated driver. The initial transportation destination is controlled by the ambulance nurse. The participating EMSs are integrated into six regional, inclusive trauma regions. Seven out of 11 inclusive trauma regions in the Netherlands, encompassing six level-I PTCs, one level-I adult trauma centre, and 60 level-II/III PTCs or adult trauma centres, participated in this study. Each of the designated level-I PTCs (i.e., higher-level trauma centres) featured a paediatric intensive care unit and offered trauma care at the highest level for severely injured children. Surrounding level-II/III facilities were 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175
considered lower-level trauma centres designated to treat mildly and moderately injured patients and did not feature paediatric ICUs. All participating EMSs and trauma regions are part of the Pre-hospital Trauma Triage Research Collaborative (PTTRC).
Patients
All paediatric trauma patients (<16 years of age) transported by EMSs with high priority were eligible for inclusion. Patients not transported from the scene of injury to a trauma receiving emergency department (ED and patients transported to a trauma-receiving ED in one of the four non-participating trauma regions were excluded. Patients transported by helicopter were ineligible because of different triage strategies and protocols. Few patients are transported by helicopter in the Netherlands because of the relatively short distances.
A novel tool (SelectAssist) was developed to aid trauma patient selection in the pre-hospital setting (see Appendix, p2). Validation in a hold-out test dataset indicated an overall accuracy of 98·9% (95%-CI, 98·4 – 99·3). All initially selected patients were manually reviewed. The full patient flow, including the selection strategy, is depicted in Figure 1.
Data Collection
Pre-hospital electronic health records (EHR) from participating EMSs were prospectively collected in a standardised manner from January 1, 2015 to December 31, 2017. All EHRs were structured according to the template of the Dutch Basic Set of Ambulance Care and included patient demographics, vital signs, a description of the mechanism of injury, medical treatments administered, a primary survey, and a secondary survey.10 Physiologic
characteristics consisted of, among others, blood pressure, respiratory rate, and the Glasgow 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199
Coma Scale (GCS). Mechanisms of injury, suspected injuries, triage tool criteria, and relevant patient outcomes were collected from corresponding input fields. Free text fields available in EHRs were abstracted by research assistants and investigators to complement the collected data. If there was any uncertainty on the classification of a report, one different investigator was asked to review the patient record. If these investigators disagreed, a third investigator was asked to discuss the record until consensus was reached.
Pre-hospital data was linked with in-hospital data from the Dutch National Trauma Registry to construct the final dataset.10 This is a nationwide registry covering all trauma-related
hospital admissions from every ED in the Netherlands. The registry includes relevant patient outcomes, such as: Injury Severity Scores (ISS), mortality, Intensive Care Unit (ICU)
admission, and early critical-resource use.11 The Abbreviated Injury Scale version 2005,
update 2008, was used by trained trauma data managers to calculate the ISS after the final diagnosis was made.12 Patients discharged at the ED to their home environment were
considered not to be severely injured, nor to have used any critical resource within a predefined time frame. These assumptions were verified and confirmed in hand-collected data from a previous study (n=4,950).7
A combined deterministic and probabilistic linkage scheme was used to match pre-hospital patient records to patient outcomes. A novel tool (LinkAssist, see Appendix p3) was developed for probabilistic record linkage of pre-hospital and in-hospital data. The record linkage strategy was validated to be 100·0 (95%-CI, 100·0 – 100·0) accurate in previously unseen data. All patients requiring specialised trauma care admitted to any of the
participating EDs were checked by hand. 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
Test methods
First, the diagnostic accuracy of the full triage strategy was evaluated based on initial transport destination. Patients requiring specialised trauma care transported to level-I PTCs and patients not in need of specialised trauma care transported to lower-level trauma centres were considered adequately triaged. All other patients were considered erroneously triaged. Second, two actively used field triage decision schemes were externally validated: the NPAS and the FTDS (Table 1). The NPAS was used in daily practice by all EMSs in the
Netherlands to select children in need of specialised trauma care during field triage. The NPAS (Triage Choice of Hospital, version 8·1) advised transport to a level-I PTC when at least one physiologic or injury-related criterion was met. The FTDS (version of 2011) was the only actively used and evaluated protocol identified in our recent systematic review on the accuracy of paediatric field triage.8 The FTDS advised to transport a patient to the highest
level of care within a trauma system when at least one physiologic or injury-related criterion was met.
The primary reference standard for need of specialised care was defined as an ISS 16, as is suggested by the ACSCOT to evaluate triage accuracy.6 Since the ISS is based on anatomical
criteria, it is assumed to be consistent with the patient status on-scene and was therefore used as a diagnostic reference standard. A secondary, resource-based, reference standard was adopted that was targeted on early critical-resource use. Early critical-resource use was defined as a composite endpoint consisting of intubation in the pre-hospital setting, ICU admission after discharge from the ED, major surgical intervention within 12 hours, major radiological intervention within 12 hours, or death within 24 hours after ED arrival (see Table 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247
S1, Appendix, p3). This definition is similar to previous research with resource-based reference standards.10,13
Statistical analysis
The primary outcome was the diagnostic accuracy (undertriage, overtriage) of the full triage strategy based on transport destination. Undertriage was defined as the proportion of patients in need of specialised trauma care initially transported to a lower-level paediatric or adult trauma centre, whereas overtriage was defined as the proportion of patients not requiring specialised trauma care transported to a level-I PTC.
The secondary outcomes were diagnostic accuracy (sensitivity, specificity, negative
predictive value [NPV], positive predictive value [PPV], positive likelihood ratio [LR+], and negative likelihood ratio [LR-]) of the NPAS and the FTDS triage tools, and the compliance of EMS professionals to the NPAS. Sensitivity was defined as the proportion of patients in need of specialised trauma care (ISS ≥ 16 or early critical-resource use) that was correctly identified by the triage tool, regardless of the destination facility. Specificity was defined as the proportion of children not in need of specialised care that was accurately classified by the triage tool, again disregarding the destination facility. Higher-level trauma centre criteria from both the NPAS and the FTDS were retrospectively applied to resemble strict adherence to the triage criteria. Triage tool compliance by EMS professionals was assessed by
quantifying the discrepancy between the trauma centre level suggested by the triage tool and the designated level of the actual transport destination.
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
Descriptive statistics were calculated using frequencies and percentages for categorical variables, and median values with interquartile ranges for continuous variables. Contingency tables were constructed for each pair of index tests and reference standards. Diagnostic accuracy metrics were derived from each contingency table. Binomial Agresti-Coull 95%-CIs were calculated where applicable.14 The log-method was used to define 95%-CIs for ratios.15
Patient height was estimated based on age, gender, and mean values per age group in the Dutch population. The haversine method was used to calculate the great-circle-distance (i.e. as the crow flies) between the scene of injury and surrounding hospitals. The sample size was determined based on prior research in an adult population.7
Multiple imputation was used to address missing values. Variables present in the triage criteria with missing values were GCS, systolic blood pressure, and respiratory rate. Missing values were imputed based on a predictor matrix that included, among others, vital signs measured in the ED, age, gender, and patient outcomes. Multi-level multiple imputation methods that account for clustering across sites were adopted using the R-package micemd.16
Forty-eight datasets were generated based on 20 iterations per set. Analyses were applied to each of the 48 datasets. Results for all analyses were averaged to calculate point estimates. Confidence intervals were calculated in accordance with Rubin’s rules.17
Role of the funding source
The funders of the study had no role in the study design, data collection, analyses,
interpretation of the data, or the writing of the final report. RvdS, JFW, RDL had full access to the data and the corresponding author had the final responsibility to submit for publication. 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293
Results
Between January 1, 2015 and December 31, 2017, approximately one and a half million patient records were identified of which 631,475 patients were transported to an ED with high priority (Figure 1). After excluding non-trauma patients (n=466,071; 74%) and patients 16 years of age (n=152,473; 92%), 12,931 injured children remained eligible for inclusion. Sixteen patients were lost to follow-up, because they were transported to an ED in a non-participating trauma region or to a different country.
Pre-hospital variables with missing values were systolic blood pressure (30%), GCS (24%), and respiratory rate (37%).
The median age of the included patients was 10·3 years, 58% was male, and 42% was female (Table 2). The median systolic blood pressure increased per age group from 106 (children 0 – 1 years) to 124 (children 11 – 15 years), whereas median respiratory rates and heart rates decreased with increasing age. The median distance to a higher-level trauma centre was 11·3 km (IQR, 5·3 – 26·7). The closest trauma centre was bypassed 1314 times (10%). The most common injury patterns were head injuries (29%; 3,683/12,915) and injuries in the
extremities (29%; 3,780/12,915). Injury patterns differed considerably between age groups. Head injuries were more prevalent in infants, toddlers, and pre-schoolers. Conversely, grade-schoolers, and teenagers suffered injured extremities more often. More than half of the patients fell (50%; 6,492/12,915), 19% (2,464/12,915) had a bicycle incident, 16%
(2,017/12,915) was involved in a motor vehicle incident, and 4% (478/12,915) suffered burn injuries. Approximately one in five patients had a sport-related injury.
294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317
A total of 4,091 (32%) patients were hospitalised after ED evaluation. The median ISS of hospitalised patients was 2 (IQR, 1 – 4). A total of 129 hospitalised patients with an ISS 16 were considered in need of specialised trauma care based on the primary reference standard. A total of 202 admitted patients (2%) had an Abbreviated Injury Scale 3 in the head or neck. Likewise, 225 (2%) had a score of 3 in the extremities. Early critical resources were used by 227 (2%) of the included patients, of which 90 (1%) were intubated in the out-of-hospital setting, 48 (<1%) underwent a major surgical intervention (e.g. craniotomy, intracranial pressure monitoring, damage control thoracotomy/laparotomy) or radiological intervention, and ten patients died within 24 hours after ED arrival. All deceased patients had impaired vital signs and a median Glasgow Coma Scale of 3 (IQR, 3 – 7). Seven of these patients suffered from submersion injuries, two patients were involved in motor vehicle accidents, and one patient died as a result of an explosion. The percentage of children transported to a level-I PTC decreased from 30% to 16% with increasing age groups.
Presence of the individual NPAS and FTDS criteria in the current cohort are shown in Table 3. One or more of the NPAS physiological criteria were met by 799 (6%) children. In contrast, 1,993 (15·4%) children fulfilled at least one criterion of the FTDS physiological criteria. Anatomical criteria of the NPAS and FTDS were met by 47 (0·4%) and 121 (1%) children, respectively. The NPAS was considered positive in 835 (6%) patients, whereas the FTDS generated a positive test result in 2,090 (16%) children.
The undertriage rate in the current cohort was 16% (95%-CI, 11 – 24) based on the primary reference standard (ISS ≥ 16), whereas the overtriage rate was 21% (95%-CI, 21 – 22; Table 4). Evaluation using the secondary reference standard (early critical-resource use) resulted in 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341
an undertriage rate of 15% (9%-CI, 11 – 20) and an overtriage rate of 21% (95%-CI, 20 – 21). A total of 50/187 (27%; 95%-CI, 21 – 34) patients with severe traumatic brain injury (Abbreviated Injury Scale 3) were first transported to lower-level trauma centres. Eight out of ten deceased patients were initially transported to a level-I PTC. The median distance from the scene of injury of undertriaged patients to a higher-level trauma centre was 22·4 km (IQR, 9.0 – 31.6) with a maximum of 51·0 km.
The sensitivity and specificity of the NPAS were 53% CI, 44 – 63) and 94% (95%-CI, 93 – 95), respectively, based on the primary reference standard. The FTDS was 64% (95%-CI, 54 – 74) sensitive to select children with an ISS 16 and had a specificity of 84 (95%-CI, 83 – 86). Evaluation of the NPAS using the secondary, resource-based, reference standard resulted in a sensitivity of 50% (95%-CI, 44 – 57) and a specificity of 94 (95%-CI, 94 – 95). The calculated sensitivity and specificity rates for the FTDS based on the secondary reference standard were 60 (95%-CI, 53 - 68) and 85 (95%-CI, 83 - 86). EMS professionals complied to a positive result of the NPAS in 45% (95%-CI, 41 – 49) of the cases.
Compliance to a negative advice was 80% (95%-CI, 79 – 80), although the NPAS does not obligate EMS professionals to transport mildly injured patients to lower-level trauma centres. 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358
Discussion
This multi-site study investigated the accuracy of paediatric pre-hospital trauma triage based on patient transport destination, triage protocols, and protocol compliance. We found that one in six-to-seven children requiring specialised trauma care was not transported to a level-I PTC. The full triage strategy was therefore unable to attain a satisfactory undertriage rate of <5%, which might have led to avoidable adverse patient outcomes. Moreover, both the NPAS and FTDS triage protocols were externally validated in the current cohort and were found to be insensitive and moderately to highly specific based on both an anatomical and a resource-based reference standard. Triage protocols, however, must have very high sensitivities, indicating that children designated as low risk do not include patients in need of specialised trauma care, to avoid undertriage. Predictive values were either high (NPV) or low (PPV) for both protocols. This was likely influenced by the low prevalence of a positive reference standard.18 Conformity to the NPAS in daily practice was low, but did lead to lower
undertriage rates as compared to strict protocol adherence.
The strengths of this study are its generalisability, the robust and sensitive methods used to select patients, the record linkage strategy, and the methodology used to construct the cohort.
First, we selected EMSs with urban, suburban, and rural service areas. These
heterogeneous EMSs were chosen to increase the generalisability of the results. The effect of inadequate triage tools is likely transportable to other inclusive trauma systems with similar (dichotomous) triage strategies, whereas destination-based accuracy might be different because of geographical differences, local regulations, and dissimilarities in the education of EMS professionals. These results might also extrapolate to trauma team activation protocols 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381
based on similar criteria in both inclusive trauma systems and exclusive trauma centres, although this needs to be verified in separate studies first.
Second, children were not solely selected based on the chief complaint (e.g.
traumatology), but all children transported by EMSs to EDs in participating trauma regions were thoroughly screened in order not to miss any patient that needed trauma-related ED evaluation. Only sixteen patients were lost to follow-up after transport to a non-participating trauma centre.
Third, all EDs in the participating trauma regions contributed to data collection. This enabled the most appropriate type of study population to investigate triage rates and to validate triage protocols. Another advantage of this study was the standardised data
collection, and the calculation of ISSs based on the Abbreviated Injury Scale. This approach is superior to ICD-9 derived ISSs.19
A number of remarks should be made before interpreting these results. The current study was not statistically powered to calculate triage rates in different subgroups (e.g., per resource, age group, or region). This would have been insightful, since injury patterns differ between age groups, and the density, proximity, and capacity of PTCs differ between trauma regions. This is mainly caused by the low prevalence of children in need of specialised trauma care.
Both an anatomical and a resource-based reference standard were used to evaluate the diagnostic accuracy of the NPAS and the FTDS. Trauma systems are traditionally evaluated using the ISS, but resource-based reference standards are increasingly being proposed as a – supposedly better – alternative to determine need of specialised trauma care.13,20 It could be
argued that resource use directly anticipates on the consequences of trauma-care 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405
regionalization (e.g. centralization of resources), whereas ISS merely acts as a surrogate marker. The downside of a resource-based reference standard, however, is the fact that it is a composite endpoint dependent on the current triage strategy (i.e., resources might be
unavailable at the transport destination and thus not used), whereas the ISS is not.
The high undertriage rate might be partly attributable to the low sensitivity of the NPAS, although it was higher in comparison to a more select adult population.7 The diagnostic
accuracy of the FTDS was previously evaluated in four studies conducted in the USA. Lerner and colleagues evaluated the accuracy of the physiological step and the complete FTDS (the entire triage algorithm, including mechanism of injury, and special considerations) in a prospective study in three PTCs based on a resource-based reference standard.13,21 The
reported sensitivities were 49% and 65%, respectively. These results cannot be directly compared to the current study, since they were derived from a selected population of children admitted to PTCs and focused on different steps of the FTDS.13,21 Two studies by Newgard et
al. evaluated the complete FTDS (version 2006) based on an ISS 16.19,22 Sensitivity rates
ranged between 84% and 87%. Judgment by EMS professionals was the most frequently used triage criterion. This suggests that the triage criteria were unable to capture the heterogeneous population in need of specialised trauma care. The current study evaluated the physiologic and anatomic criteria, which makes it unsuitable for comparison. Finally, one study evaluated the compliance of EMS professionals to the physiologic step of the FTDS based on transport destination.23 It showed that more than one-quarter of the children positively identified by the
FTDS physiologic criteria was transported to a lower-level trauma centre. Our study reports even higher non-compliance to the NPAS.
406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429
The poor performance of the evaluated field triage protocols is likely multi-causal. Several scenarios, mechanisms of injury, and heterogeneous injury patterns can lead to severe injuries. Criteria in current decision schemes are generic, do not interact with other criteria, and are therefore unable to produce an advice on patient-level. A probability or risk score might be a better alternative, since triage protocols are only one component of the full triage strategy.10,24 Furthermore, dichotomization of physiologic criteria leads to a loss of
information and the current cut-off points are not child-specific (see Figure S1, Appendix, p5).25
Further research could focus on child-specific, highly discriminative predictors for need of specialised trauma care. New prediction models could be developed and validated in order to aid EMS professionals during the challenging process of field triage.
Undertriage and overtriage rates are key metrics to evaluate the functioning of trauma systems. This study reveals undertriage rates that need improvement to enable maximally cost-efficient care. After all, centralization of resources and expertise (i.e., regionalised trauma care) could be detrimental if patients are not transported to the right hospital.
Declaration of interests
We declare no competing interests.
Acknowledgements
This study was partly funded by grants from the Netherlands Organisation for Health
Research and Development (ZonMw) and the Innovation Fund Health Insurers. We thank the 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453
research staff from all participating sites and all participating members of the Pre-hospital Trauma Triage Research Collaborative. We thank our research assistants: Alexander de la Mar, Annemiek Vuurens, Dunja Scheepmaker, and Toril Lintzen. Finally, we would like to thank Jill Whittaker for her advice on writing style.
454 455 456 457
Table 1
Triage tools
Elaboration
NPAS* FTDS†
Physiologic
characteristics Glasgow Coma Scale <9 or deteriorating Glasgow Coma Scale <14
Patients fulfilling any of the physiological criteria should be transported to a level-I paediatric trauma centre or the highest level of care within the trauma system in case of the FTDS.
Airway, breathing, or circulation cannot be stabilised Systolic blood pressure (mmHg) < 90 mm Revised Trauma Score <11 or Paediatric Trauma Score
<9
Respiratory rate <10 or >29 per minute or <20 in infant aged <1 year Anisocoria Hypothermia ≤32 °C Anatomy of suspected injuries
All penetrating injuries to the head, thorax, and abdomen
All penetrating injuries to head, neck, torso, and extremities proximal to elbow and knee
Patients suspected of any of these injury characteristics should be transported to a level-I paediatric trauma centre (NPAS) or the highest level of care within the trauma system (FTDS).
Flail chest Flail chest
Two or more long-bone fractures Two or more proximal long-bone fractures Amputation proximal to wrist and ankle Amputation proximal to wrist and ankle
Unstable pelvic fracture Pelvic fractures
Paralysis Paralysis
Open or depressed skull fracture Crushed, degloved, or mangled extremity Mechanism of
injury
Falls: Falls: Patients should be transported
to a regional trauma centre, but not necessarily the highest level of care within the system. >5 meter or three times the height of the patient >6 meter, or >3 meter or two to three times the height of the
child for children <15 years of age
High-risk motor vehicle crash: High-risk auto crash:
– Deformity >50 cm or intrusion > 30cm occupant site – Intrusion: >30 cm occupant site; >46 cm any site – Ejection from automobile – Ejection (partial or complete) from automobile
– Death in same vehicle – Death in same passenger compartment
– More than 65 km/hour – Motorcycle crash >32 km/hour
– Auto vs. pedestrian >10 km/hour – Auto vs. pedestrian/bicyclist thrown, run over, or with significant (>32 km/hour) impact
– Vehicle telemetry data consistent with high risk of injury
The order of variables of the NPAS and the FTDS were changed to facilitate comparison between the two triage tools. Units in the FTDS were converted to the SI system of units. Special patient or system considerations were removed (e.g. pregnancy). NPAS=National Protocol of Ambulance Services. FTDS=Field Triage Decision Scheme. *Triage Choice of Hospital version 8·1. †Field Triage Decision Scheme version of 2011.
Table 1: Criteria, variable groups, and elaboration of the NPAS and FTDS triage tools 458
459 460
Table 2 Age groups All patients 0 – <2 2 – <6 6 – <11 11 – <16 Demographic characteristics Age, y 10·3 (4·2 - 13·6) 1·1 (0·7 - 1·5) 3·7 (2·8 - 4·9) 8·8 (7·4 - 10·1) 13·8 (12·6 - 14·9) Boys 7503 (58·1) 936 (54·7) 1524 (61·5) 1645 (58·8) 3398 (57·3) Girls 5412 (41·9) 775 (45·3) 953 (38·5) 1155 (41·2) 2529 (42·7)
Prehospital physiological characteristics
Systolic blood pressure 120 (110 - 130) 106 (92 - 126) 110 (100 - 120) 114 (105 - 124) 124 (114 - 134)
Systolic blood pressure <90 488 (3·8) 93 (5·4) 140 (5·6) 108 (3·8) 148 (2·5)
Respiratory rate 18 (15 - 20) 24 (16 - 30) 20 (16 - 24) 18 (16 - 20) 16 (14 - 20)
Respiratory rate <10 or >29 1046 (8·1) 521 (30·5) 237 (9·6) 102 (3·7) 186 (3·1)
Heart rate 93 (80 - 110) 120 (98 - 134) 103 (90 - 120) 92 (82 - 103) 88 (78 - 100)
Saturation 99 (98 - 99) 99 (97 - 100) 99 (98 - 99) 99 (98 - 99) 99 (98 - 99)
GCS 15 (15 - 15) 15 (15 - 15) 15 (15 - 15) 15 (15 - 15) 15 (15 - 15)
Characteristics of suspected injuries
Head 3683 (28·5) 779 (45·5) 1019 (41·1) 660 (23·6) 1225 (20·7) Thorax 218 (1·7) 4 (0·2) 19 (0·8) 59 (2·1) 136 (2·3) Abdomen 263 (2) 6 (0·4) 34 (1·4) 87 (3·1) 136 (2·3) Extremities 3780 (29·3) 55 (3·2) 370 (14·9) 908 (32·4) 2447 (41·3) Mechanism of injury Falls 6492 (50·3) 1122 (65·6) 1508 (60·9) 1505 (53·8) 2357 (39·8) >3 meters 140 (1·1) 10 (0·6) 35 (1·4) 48 (1·7) 47 (0·8)
≥3 times body length 218 (1·7) 152 (8·9) 34 (1·4) 22 (0·8) 10 (0·2)
Fall from stairs 1042 (8·1) 349 (20·4) 441 (17·8) 114 (4·1) 138 (2·3)
Motor vehicle incident 2023 (15·7) 87 (5·1) 297 (12) 470 (16·8) 1169 (19·7)
Bicycle incident 2464 (19·1) 74 (4·3) 308 (12·4) 466 (16·6) 1616 (27·3) Pedestrian vs auto 399 (3·1) 11 (0·6) 110 (4·4) 157 (5·6) 121 (2) Bicycle vs auto 773 (6) 7 (0·4) 42 (1·7) 117 (4·2) 607 (10·2) Burns 478 (3·7) 218 (12·7) 100 (4) 72 (2·6) 88 (1·5) Suffocation 133 (1) 71 (4·1) 45 (1·8) 6 (0·2) 11 (0·2) Submersion 75 (0·6) 17 (1) 34 (1·4) 10 (0·4) 14 (0·2) Sport related 2688 (20·8) 7 (0·4) 85 (3·4) 592 (21·1) 2004 (33·8) Violence 214 (1·7) 6 (0·4) 10 (0·4) 19 (0·7) 179 (3) Outcomes In-hospital stay 4091 (31·7) 893 (52·2) 1050 (42·4) 901 (32·2) 1247 (21) ISS* 2 (1 - 4) 2 (1 - 2) 2 (1 - 4) 4 (2 - 5) 4 (2 - 5) ISS ≥ 16* 129 (3·2) 21 (2·4) 27 (2·6) 29 (3·2) 52 (4·2)
ISS score ≥ 3, per region*
Head and neck 202 (1·6) 36 (2·1) 45 (1·8) 44 (1·6) 77 (1·3)
Face 4 (0) 1 (0·1) 0 (0) 2 (0·1) 1 (0)
Thorax 71 (0·5) 10 (0·6) 17 (0·7) 15 (0·5) 29 (0·5)
Abdomen 39 (0·3) 1 (0·1) 3 (0·1) 10 (0·4) 25 (0·4)
Extremities 225 (1·7) 10 (0·6) 38 (1·5) 75 (2·7) 102 (1·7)
External 56 (0·4) 19 (1·1) 25 (1) 7 (0·2) 5 (0·1)
Early critical-resource use* 227 (1·8) 38 (2·2) 55 (2·2) 52 (1·9) 82 (1·4)
Discharge from ED to ICU 155 (1·2) 31 (1·8) 43 (1·7) 34 (1·2) 47 (0·8) Out-of-hospital intubation 90 (0·7) 13 (0·8) 22 (0·9) 14 (0·5) 41 (0·7) Major interventions <12h 48 (0·4) 4 (0·2) 5 (0·2) 19 (0·7) 20 (0·3) Mortality <24h 10 (58·8) 1 (33·3) 5 (62·5) 2 (100) 2 (50) Transport to level-I PTC 2823 (21·9) 515 (30·1) 700 (28·3) 647 (23·1) 961 (16·2) Transport to non-PTC 10092 (78·1) 1196 (69·9) 1777 (71·7) 2153 (76·9) 4966 (83·8)
Data are median (IQR) or n (%). Age ranges include both the begin and endpoint. Values derived from multiply imputed variables were rounded to zero decimals. GCS=Glasgow Coma Scale. ISS=Injury Severity Score. ED=emergency department. ICU=intensive care unit. PTC=Paediatric Trauma Centre. *Hospitalised patients only.
Table 2: Patients characteristics in the validation cohort per age group 461 462 463 464 465 466 467 468
Table 3 469 470 471 472 473
Table 4
Validation cohort (n=12915) NPAS and FTDS
Flail chest 1 (0)
Amputation proximal to wrist and ankle 1 (0)
Paralysis 14 (0·1) NPAS GCS <9 or deteriorating 171 (1·3) RTS <11 or PTS <9 490 (3·8) Anisocoria 5 (0) Hypothermia ≤32 °C 9 (0·1)
All penetrating injuries to the head, thorax, and abdomen 19 (0·1)
Two or more long-bone fractures 8 (0·1)
Unstable pelvic fracture 7 (0·1)
Fall >5 meter or three times the height of the patient 227 (5·8)
MV deformity >50 cm or intrusion > 30cm occupant site 12 (0·1)
MV ejection 3 (0)
MV incident with death in same vehicle 2 (0)
MV incident >65 km/hour 153 (1·2)
Pedestrian vs auto >10 km/hour 395 (3·1)
FTDS
GCS <14 693 (5·4)
Systolic blood pressure (mmHg) < 90 mm 488 (3·8)
Respiratory rate <10 or >29 per minute or <20 in infants aged <1 year 1001 (7·7) All penetrating injuries to head, neck, torso, and extremities proximal to elbow and
knee 24 (0·2)
Two or more proximal long-bone fractures 4 (0)
Pelvic fractures 50 (0·4)
Open or depressed skull fracture 20 (0·2)
Crushed, degloved, or mangled extremity 13 (0·1)
Fall >6 meter, or >3 meter or two to three times the height of the child for children
<15 years of age 614 (15·6)
MV intrusion: >30 cm occupant site; >46 cm any site 12 (0·1)
MV ejection (partial or complete) 3 (0)
MV incident with death in same passenger compartment 0 (0)
Motorcycle crash >32 km/hour 50 (0·4)
Auto vs pedestrian/bicyclist, thrown, run over, or with significant (>32 km/hour)
impact 477 (3·7)
Vehicle telemetry data consistent with high risk of injury 0 (0)
NPAS=National Protocol of Ambulance Services: Triage Choice of Hospital version 8·1. FTDS=Field Triage Decision Scheme version of 2011. GCS=Glasgow Coma Scale. RTS=Revised Trauma Score. PTS=Paediatric Trauma Score. MV=Motor Vehicle. Values are presented as n (%).
Table 3: Presence of NPAS and FTDS criteria in the validation cohort 474 475 476 477 478 479 480 481
Outcomes
Injury Severity Score ≥ 16 Early critical-resource use Destination-based Transported to level-I PTC With outcome (TP) 108 194 Without outcome (FP) 2715 2629 Transported to lower-level PTC With outcome (FN) 21 33 Without outcome (TN) 10071 10059 Undertriage (95%-CI) 16·3 (10·8 - 23·7) 14·5 (10·5 - 19·7) Overtriage (95%-CI) 21·2 (20·5 - 22·0) 20·7 (20·0 - 21·4) NPAS* Positive on criteria With outcome (TP) 69·0 114·5 Without outcome (FP) 765·7 720·2 Negative on criteria With outcome (FN) 60·0 112·5 Without outcome (TN) 12020·3 11967·8 Sensitivity (95%-CI) 53·3 (43·9 - 62·9) 50·4 (43·6 - 57·3) Specificity (95%-CI) 94·0 (93·4 - 94·6) 94·3 (93·7 - 94·9) PPV (95%-CI) 8·3 (6·5 - 10·4) 13·7 (11·4 - 16·4) NPV (95%-CI) 99·5 (99·4 - 99·6) 99·1 (98·9 - 99·2) LR+ (95%-CI) 8·9 (4·4 - 18·2) 8·9 (4·9 - 16·2) LR- (95%-CI) 0·5 (0·4 - 0·6) 0·5 (0·5 - 0·6) FTDS† Positive on criteria With outcome (n) 83·2 137·3 Without outcome (n) 2006·3 1952·3 Negative on criteria With outcome (n) 45·8 89·7 Without outcome (n) 10779·7 10735·7 Sensitivity (95%-CI) 64·5 (54·1 - 74·1) 60·5 (52·8 -67·8) Specificity (95%-CI) 84·3 (83·1 - 85·5) 84·6 (83·4 - 85·8) PPV (95%-CI) 4·0 (3·2 - 5·0) 6·6 (5·5 - 7·8) NPV (95%-CI) 99·5 (99·4 - 99·6) 99·2 (99·0 - 99·3) LR+ (95%-CI) 4·1 (2·8 - 6·1) 3·9 (3·0 - 5·1) LR- (95%-CI) 0·4 (0·3 - 0·5) 0·5 (0·4 - 0·6)
PTC=Paediatric Trauma Center. TP=true positive. FP=false positive. FN=false negative. TN=true negative. NPAS=National Protocol of Ambulance Services. PPV=positive predictive value. NPV=negative predictive value. LR+=positive likelihood ratio. LR-=negative likelihood ratio. FTDS=Field Triage Decision Scheme. *Triage Choice of Hospital version 8·1. †Field Triage Decision Scheme version of 2011.
Table 4: Diagnostic accuracy of paediatric field triage, based on destination, the NPAS, and the FTDS
department. P=Probability. 486
RvdS, MP, LPHL, and MvH wrote the statistical analysis plan, analysed the data, and drafted the manuscript. All authors contributed to the study design, interpreted the data, and revised the manuscript.
488 489 490
1. World Health Organization. World report on child injury prevention. Geneva, 2008. 2. Potoka DA, Schall LC, Gardner MJ, Stafford PW, Peitzman AB, Ford HR. Impact of pediatric trauma centers on mortality in a statewide system. J Trauma 2000; 49(2): 237-45. 3. Sathya C, Alali AS, Wales PW, et al. Mortality Among Injured Children Treated at Different Trauma Center Types. JAMA Surg 2015; 150(9): 874-81.
4. Newgard CD, Staudenmayer K, Hsia RY, et al. The cost of overtriage: more than one-third of low-risk injured patients were taken to major trauma centers. Health Aff (Millwood) 2013; 32(9): 1591-9.
5. Ambulancezorg Nederland. Landelijk Protocol Ambulancezorg. Zwolle, 2016.
6. American College of Surgeon Committee on Trauma. Resources for Optimal Care of the Injured Patient: 2014. Chicago, IL, 2014.
7. Voskens FJ, van Rein EAJ, van der Sluijs R, et al. Accuracy of Prehospital Triage in Selecting Severely Injured Trauma Patients. JAMA Surg 2018; 153(4): 322-7.
8. van der Sluijs R, van Rein EAJ, Wijnand JGJ, Leenen LPH, van Heijl M. Accuracy of Pediatric Trauma Field Triage: A Systematic Review. JAMA Surg 2018; 153(7): 671-6. 9. Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ 2015; 351: h5527.
10. Van der Sluijs R, Debray TPA, Poeze M, Leenen LPH, van Heijl M. Development and validation of a novel prediction model to identify patients in need of specialized trauma care during field triage: design and rationale of the GOAT study. Diagnostic and Prognostic Research 2019; 3(12). 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
describing patients with multiple injuries and evaluating emergency care. J Trauma 1974;
14(3): 187-96.
12. A. GT, E. W. Abbreviated Injury Scale 2005: update 2008 Association for the Advancement of Automative Medicine; 2008.
13. Lerner EB, Cushman JT, Drendel AL, et al. Effect of the 2011 Revisions to the Field Triage Guidelines on Under- and Over-Triage Rates for Pediatric Trauma Patients. Prehosp Emerg Care 2017; 21(4): 456-60.
14. Agresti A, Coull BA. Approximate is better than "exact" for interval estimation of binomial proportions. Am Stat 1998; 52(2): 119-26.
15. Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol 1991; 44(8): 763-70.
16. Audigier V, Resche-Rigon M. micemd: Multiple Imputation by Chained Equations with Multilevel Data. 1.6.0. ed; 2019.
17. Rubin DB. Multiple imputation for nonresponse in surveys: John Wiley & Sons, inc.; 1987.
18. Altman DG, Bland JM. Diagnostic tests 2: Predictive values. BMJ 1994; 309(6947): 102. 19. Fleischman RJ, Mann NC, Dai M, et al. Validating the Use of ICD-9 Code Mapping to Generate Injury Severity Scores. J Trauma Nurs 2017; 24(1): 4-14.
20. Newgard CD, Hedges JR, Diggs B, Mullins RJ. Establishing the need for trauma center care: anatomic injury or resource use? Prehosp Emerg Care 2008; 12(4): 451-8.
525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554
Field Triage Guidelines to Identify Children Who Need the Resources of a Trauma Center. Prehosp Emerg Care 2017; 21(2): 180-4.
22. Newgard CD, Zive D, Holmes JF, et al. A multisite assessment of the American College of Surgeons Committee on Trauma field triage decision scheme for identifying seriously injured children and adults. J Am Coll Surg 2011; 213(6): 709-21.
23. Lerner EB, Studnek JR, Fumo N, et al. Multicenter Analysis of Transport Destinations for Pediatric Prehospital Patients. Acad Emerg Med 2019; 26(5): 510-6.
24. van Rein EAJ, van der Sluijs R, Voskens FJ, et al. Development and Validation of a Prediction Model for Prehospital Triage of Trauma Patients. JAMA Surg 2019.
25. Fleming S, Thompson M, Stevens R, et al. Normal ranges of heart rate and respiratory rate in children from birth to 18 years of age: a systematic review of observational studies. Lancet 2011; 377(9770): 1011-8. 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572
