A guide to triage

(1)

Contents lists available atScienceDirect

International Emergency Nursing

journal homepage:www.elsevier.com/locate/aaen

The accuracy of acuity scoring tools to predict 24-h mortality in traumatic brain injury patients: A guide to triage criteria

Zohre Naja ﬁ

^a,1

, Hossien Zakeri

^b,2

, Amir Mirhaghi

^c,⁎

aDepartment of Medical-Surgical Nursing, Torbat Heydariyeh University of Medical Sciences, Torbat Heydariyeh, Iran

bEmergency Medicine, Hasheminejad Hospital, School of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran

cEvidence-Based Caring Research Center, Department of Medical-Surgical Nursing, School of Nursing and Midwifery, Mashhad University of Medical Sciences, Mashhad, Iran

A R T I C L E I N F O

Keywords:

Traumatic brain injury Triage

Prehospital Emergency

A B S T R A C T

Background and aim:Prompt identiﬁcation of traumatic brain injury (TBI) is vital for patients in critical con- dition; however, it is not clear which acuity scoring tools are associated with short-term mortality. The aim of this study was to determine the accuracy of acuity scoring tools and 24-h mortality among TBI patients in both prehospital and hospital settings.

Methods:This study was an observational, prospective cohort, in which patients with TBI were followed from the accident scene to the hospital. Vital signs and acuity scoring tools, including the Revised Trauma Score (RTS), Injury Severity Score (ISS), National Early Warning Score (NEWS), Shock Index (SI), Modiﬁed Shock Index (MSI) and Trauma and Injury Severity Score (TRISS), were collected both on the scene as well as at the hospital. A logistic regression was performed to ascertain the eﬀects of clinical parameters on the likelihood of survival of patients with TBI regarding 24-h mortality.

Results::A total of 185 patients were included in this study. The mortality rate was 14% (25/185). The logistic regression model was statistically signiﬁcant atχ²= 60.8, p = 0.001. A hierarchical forward stepwise logistic regression analysis showed that age, hospital RTS and prehospital NEWS signiﬁcantly improved mortality predictions. The model explained the 51.2% variance in survival of patients with TBI.

Conclusions:The NEWS and the RTS may be used to triage TBI patients for prehospital and hospital emergency care, respectively. Therefore, because traditional vital signs criteria may be of limited use for the triage of TBI patients, it is recommended that acuity scoring tools be used in such cases.

1. Introduction

Traumatic brain injury (TBI) is the main cause of disability as well as neurologic morbidity in young adults [1]. Severe trauma is considered a serious health problem, because disability affects victims’ roles in both family and society[2]. Notably, TBI is also associated with high socio-economic costs[3]. Thefirst hour of trauma management is crucial for TBI patients due to the time-sensitive care required; therefore, mortality may be decreased if critically ill patients are recognized more readily and transferred promptly to trauma centres [4]. Para- medics are generally thefirst to assess and treat trauma patients in the prehospital environment, which makes them responsible for identifying life-threatening injuries and improving patients’ quality of care in stressful situations[5]. As such, it is essential to develop prehospital

emergency criteria to promote prompt recognition of severe TBI patients[6].

In the prehospital phase, the initial steps are to assess the level of consciousness (LOC), maintain both the airway and oxygenation, in- itiateﬂuid replacement, immobilize the spine and promptly transfer the patient to a high-level trauma centre[7]. Trauma scoring systems are useful for the recognition of critically injured patients and are a pre- requisite for establishing performance improvement among paramedics, which results from better outcome prediction and triage allo- cation as well as choosing the optimum hospital destination [8].

Trauma scoring systems are also useful for risk stratiﬁcation. This is especially true for paramedics because they are usually working with little clinical information in theﬁeld. Paramedics must prioritise by transferring severe TBI patients for advanced care sooner than other TBI

http://dx.doi.org/10.1016/j.ienj.2017.08.003

Received 31 May 2017; Received in revised form 11 August 2017; Accepted 21 August 2017

⁎Corresponding author at: Department of Medical-Surgical Nursing, School of Nursing and Midwifery, Chahrrah-e-Doktorha, Mashhad, Razavi Khorasan 9137913199, Iran.

1Department Medical-Surgical Nursing, Torbat Heydariyeh University of Medical Sciences, Razi Street, Torbat Heydariyeh, Razavi Khorasan 9516915169, Iran.

2Department of Emergency Medicine, Hasheminejad Hospital, School of Medicine, Azadi square, Mashhad, Razavi Khorasan 9177948564, Iran.

E-mail addresses:najaﬁ[email protected](Z. Najaﬁ),[email protected](H. Zakeri),[email protected](A. Mirhaghi).

T

(2)

patients with lower priorities.

Most previous studies have focused on the accuracy of the trauma scoring system and long-term outcomes; therefore, the associations between clinical criteria and short-term outcomes remain unclear.

Notably, only short-term outcomes are specific for the triage of TBI patients in both prehospital and emergency settings, while trauma- specific triage guidelines for TBI patients have rarely been developed [9]. Evidence that supports the trauma scoring system based on 24-h outcomes is thus required. To be more specific, the Glasgow Coma Scale (GCS) was found to be a good prognostic factor of long-term mortality among TBI patients in the emergency department (ED); conversely, its value in the prehospital setting remains unclear. Several studies have shown that the GCS improved the prediction of 48-h mortality, and motor scores were also significant predictors of long-term mortality (2 weeks to 6 months) [2,7,10–14]. However, the validity of these findings regarding 24-h outcomes are unclear.

Overall, an accurate tool is needed to identify the severity of TBI during early trauma, especially regarding 24-h outcomes, which could be employed for more accurate clinical decisions. In addition, because traumatic injuries are increasingly recognized as a leading source of morbidity and mortality in developing countries, context-speciﬁc research is necessary to identify opportunities for prevention and improved treatment. Therefore, the aim of this study was to determine the accuracy of acuity scoring tools and 24-h mortality among traumatic brain injury patients in both the prehospital and hospital settings.

2. Methods 2.1. Design

This study was an observational, prospective cohort that followed TBI patients from the accident scene to the hospital between February and September 2016.

2.2. Ethics

Data collection was carried out after receiving approval from the ethics committee at Mashhad University of Medical Sciences (No.

940948).

2.3. Setting

This study was conducted in the Hasheminezhad Hospital in Mashhad, Razavi Khorasan, Iran, which is the second largest Level 1 trauma centre (320 beds) in the city. The hospital provides several specialties, including neurosurgical, emergency medicine, orthopaedic, surgical and internal medicine services, 24 h per day. The hospital ED receives 14,500 trauma patients annually, most of whom arrive by ambulance. All nurses in the ED have both a Bachelor of Science (B.S.) in nursing and Trauma Certiﬁed Registered Nurse (TCRN) certiﬁcation.

All emergency physicians in the ED are specialists in emergency medicine. Emergency medical services (EMS) are provided by professional individuals who are trained to provide basic trauma life support, such as immobilization, airway management and intravenousﬂuid therapy, during ambulance transfers. All paramedics must have a B.S. in emergency medical services. They transfer trauma patients from the scene to the EDs in the shortest possible time with the aim of reducing morbidity as well as mortality.

2.4. Data collection

TBI patients who were received by ambulance and admitted to the ED were assessed. Patient records were used to collect relevant data, including age, gender, mechanism of injury, medical history, GCS, respiratory rate (RR), oxygen saturation (SpO2), temperature (T), heart rate (HR), systolic blood pressure (SBP), diastolic blood pressure (DBP)

and pupillary status display (size and reaction to light), in both the prehospital and hospital settings. The data were documented by paramedics in the prehospital setting and by emergency nurses in the ED.

The Injury Severity Score (ISS) was documented by emergency physicians in the ED. To assess in-hospital mortality, we examined death rates in theﬁrst 24 h post-injury. All included patients were followed up for mortality over 24 h in the ED. Therefore, TBI patients were divided into two groups based on their outcome (dead or alive) in theﬁrst 24 h post-injury.

2.5. Patient selection

During the study period, TBI patients who met the following criteria were included in the study: the mechanism of injury (MOI) was a traﬃc accident, they were transported directly from the scene by an emergency medical services ambulance, they had an ISS > 9, they were older than 18 years but less than 85 years of age and they had at least one vital sign parameter higher than zero at the scene (in order to in- clude only cases without clinical death). Exclusion criteria included incomplete data in either the prehospital or hospital patient records (i.e., vital signs parameters), pregnancy, comorbidities (diabetes, car- diovascular disease, chronic obstructive pulmonary disease etc.), transfer time from the scene to the hospital of more than 60 min and transfer of patients to other health care centres in theﬁrst 24 h.

2.6. Variables

GCS is used to assess the level of consciousness in patients with brain injuries. This scale is composed of three subscales that independently measure motor response, eye opening and verbal response.

Scores range from 3 to 15, with a score of 3 indicating the lowest degree of consciousness and a score of 15 indicating alertness[4]. Mean ar- terial pressure (MAP): MAP is used as an indicator of bloodflow. It may represent tissue perfusion better than SBP because it considers diastolic pressure. The Shock Index (SI) and Modified Shock Index (MSI): Both the SI and MSI have been used for prompt identification of hypovolemic shock in patients with trauma. The SI normal range varies from 0.5 to 0.7 in healthy adults[15]. The Injury Severity Score (ISS): The ISS is an anatomical scoring system that represents an overall score for trauma patients. Each injury from six body regions (head, face, chest, abdomen, extremities including pelvis and external) has a relevant Abbreviated Injury Scale (AIS) score. Only the highest AIS score in each body region is used to calculate the ISS[10]. The Revised Trauma Score (RTS): The RTS is the sum of 12 weighted, coded values for the GCS score, RR, and SBP[4]. The trauma and injury severity score (TRISS): The TRISS is used to determine the probability of survival by using the ISS, the RTS and the patient’s age. It includes both anatomic and physiologic criteria to predict survival in relation to the severity of the injury[10]. The National Early Warning Score (NEWS): The NEWS is used to assess critically ill patients, especially in the prehospital setting. It compro- mises six physiological factors including the RR, SpO2, T, SBP, HR and level of consciousness. Each factor scores from 0 to 3, except supple- mental oxygenation. The NEWS ranges from 0 to 20[16].

2.7. Confounders analysis

We considered several potential confounders of the association between severity of head injury and mortality during thefirst day of admission. These included mechanism of injury, an ISS lower than 9 and other life threatening injuries. For purposes of analysis, traffic accident survivors (driver, passenger, motorcycle, bicycle and pedestrian) were included in the study. Other MOIs were excluded because they may have coincided with different patterns of injury. TBI patients with a minimum ISS score of 9 were included only if that score belonged to a brain injury. TBI is often associated with a high-velocity circumstance;

hence, injuries to other parts of the body are more commonly seen.

(3)

2.8. Statistical analyses

Statistical analyses were performed with SPSS version 18.0 software (SPSS, Chicago, Illinois, USA). Data are presented as a mean (Standard Deviation) for continuous variables and as proportions for categorical variables. The Fisher’s exact test was used to evaluate the relationship between the categorical variable and the 24-h mortality for categorical variables. Continuous variables that were not normally distributed between the two independent groups were compared using the Mann-WhitneyU test. Acuity scales for survival probability in TBI patients, such as the GCS, SI, MSI, MAP, ISS, RTS, TRISS and NEWS, were calculated based on both prehospital and hospital parameters. All scales were calculated using Microsoft Excel software (2013), except the TRISS.

The effect of each predictor on the outcome variable was assessed individually by performing a univariate analysis. Variables showing a statistically significant association at a 5% level of significance in the univariate analysis were included in the multivariate analysis.

Multivariate logistic regression was used for developing a prognostic model for the outcome (mortality: no/yes) of theﬁrst 24 h of admission. Models using a predeﬁned group of predictor variables (method

‘enter’) in the univariate analysis, as well as a hierarchical forward stepwise LR method for inclusion of predictor variables in multivariate analysis, were employed.

The independent predictor variables were the GCS-Eye, GCS-Verbal, GCS-Motor, Total GCS, Right-Pupillary size, Right-Pupillary reactivity, Left-Pupillary size, Left-Pupillary reactivity, RR, SpO2, HR, SBP, DBP, MAP, SI, MSI, RTS, TRISS and the NEWS. In prognostic research, the logistic regression model is a commonly used statistical method, which is estimated by using maximum likelihood methods and when the outcome variable follows a binomial distribution. Coefficients of the final model are presented together with the respective odds ratio (OR) and a corresponding 95% confidence interval (CI). No adjustments for age, sex, injury severity or other relevant variables were performed.

The ability of variables to predict both mortality and TBI presence were evaluated using the receiver operating characteristic (ROC) curve analysis and expressed as the area under the ROC curve (AUROC) with a 95% CI. Discrimination is a variable’s ability to separate patients with different outcomes. It is quantified by using the area under the receiver’s operating characteristic curve (AUC), which determines whe- ther those with fewer predicted risks are less likely to have a poor outcome among all possible pairs of patients with different outcomes.

The ROC curve is based on sensitivity and speciﬁcity for various cut-oﬀ points. An AUC of 1.0 means perfect discrimination (for example, perfect separation of survivors from non-survivors), and an AUC of 0.5 means no discriminative power at all (for example, separation of survivors from non-survivors shows no better results than chance alone).

The larger the AUROC, the better the ability of a variable to separate survivors from non-survivors.

3. Results

3.1. Patient characteristics

Clinical characteristics of the 185 patients are demonstrated in Table 1. The median age of the patients was 33 (IQR: 24–50) years old, and 76% were men. Overall, 116 (71%) patients were pedestrians or motorcycle riders, and 170 (92%) had a blunt injury. The median transfer time was 32 (IQR: 24–43) minutes. The survival rate was 86%

(160/185). Patients in the mortality group were not signiﬁcantly older in either the prehospital or hospital setting (p > 0.05). The proportions of both groups also did not diﬀer regarding sex status. Two patients from survival group were excluded because of incomplete data.

3.2. Survival analysis

Clinical variables that independently predict 24-h survival are

demonstrated inTable 2. All odds ratios, except RR, SpO2, SI and MSI, were increased from the prehospital to the hospital setting. A logistic regression was performed to ascertain the effects of clinical parameters on the likelihood of survival of patients with TBI. The model provided a statistically significant improvement over the constant-only model. The logistic regression model was statistically significant at χ²= 60.8, p = 0.001. The Hosmer-Lemeshow test of the goodness offit showed that the model is a goodfit to the data because p = 0.517. The model explained the 51.2% variance in survival of patients with TBI. Na- gelkerkeR²(51.2%) indicated a moderate relationship between prediction and grouping. The correct prediction rate was 93% (99.4% for survival and 52.0% for mortality). A hierarchical forward stepwise logistic regression analysis showed that age and the hospital’s RTS (≥4) both significantly improved predictions (p < 0.05) (Table 3). In- creasing age predicted an increased likelihood of mortality. NEWS≤7 also improved prediction significantly (Table 3). The accuracy of survival rates, which was based on the RTS, TRISS, MAP, SI, MSI, NEWS and ISS scores, are reported inTable 4 and 5.

4. Discussion

The current study shows that vital sign criteria alone may not predict short-term outcomes, while acuity scoring tools, including the RTS-hosp and NEWS-prehosp, predict 24-h mortality in hospital and prehospital settings among TBI patients respectively. It is vital for clinicians who work in the emergencyfield to identify critically ill patients based on valid criteria. The results show that clinicians should not rely on vital sign criteria alone in either thefield or the triage room. Even the GCS alone may not be a reliable indicator of 24-h mortality; hemodynamic parameters may need to be added to the GCS to make a rigorous prediction of TBI patients’acuity in thefirst hour after traumatic injury. Therefore, acuity scoring tools may have advantages over vital sign criteria regarding short- term mortality. It is worth mentioning that paramedics commonly relies on GCS and vital sign criteria to triage TBI patients rather than acuity scoring tools. Acuity scoring tools need to be calculated by computers, so their clinical applicability is depend largely on the extent of usage of computers in the prehospital clinical care[17].

Our results, which are supported by Lichtveld et al., indicated that age and triage-RTS (T-RTS) appear to be the greatest causes of death in theﬁrst 24 h in the ED among TBI patients[4]. Age predicts mortality among TBI patients; those of a higher age were associated with an in- crease in the risk of short-term mortality. The fact that older patients are at a much higher risk of mortality in comparison to younger ones has been conﬁrmed by several studies[4,7,10,18–20].

4.1. Glasgow coma scale

The GCS is a cornerstone for assessing level of consciousness (LOC) in TBI patients. LOC had a significant difference from the scene to the ED in both mortality and survival groups in our study; however, the GCS drop from the prehospital to hospital setting did not differ significantly between groups. TBI patients may experience a decrease in LOC due to mild brain damage or hemodynamic compromise; hence, this may have caused a GCS drop in survival group that did not show any significant difference in comparison with mortality group. In addition, GCS at the scene in association with mortality has not been supported by other studies[6,21]. Moazez et al. also indicated that GCS in the triage room is a significantly better predictor of mortality than when it is recorded at the scene[21]. However Majidi et al. indicated prehospital neurologic deterioration is an independent predictor of in- hospital mortality (odds ratio: 2.30), but they didn‘t report any significant association with in-hospital mortality[6].

Although GCS alone did not predict the short-term mortality in our results, GCS is an important indicator of mortality, when combined with other factors, especially in patients with moderate to severe TBI [22]. GCS association with long-term outcomes has been well-

(4)

Table1 Comparisonofbaselinecharacteristicsbasedonsurvivalandmortalitygroupsinbothprehospitalandhospitalsettings. CharacteristicsAll(n=185)Survival(n=160)Mortality(n=25)PvalueCharacteristicsAll(n=185)Survival(n=160)Mortality(n=25)Pvalue Demographic Ageyear(mean±SD)39.01±18.438.1±17.744.2±21.90.127Transfertimeminute(mean±SD)39.01±18.438±20.833.3±13.30.211 Sexmale(%)141(76)123(76.9)18(72)0.616Sexfemale(%)44(24)37(23.1)7(28)0.616 BluntTrauma(%)170(92)147(91.9)23(92.0)1.000PenetratingTrauma(%)15(8)13(8.1)2(8.0)1.000 PrehospitalHospital GCS-Eye3.33±0.913.47±0.762.44±1.20.001GCS-Eye2.65±0.962.81±0.861.6±0.910.001 GCS-Verbal4.10±1.194.30±0.982.84±1.50.001GCS-Verbal3.24±1.353.64±1.881.88±1.230.001 GCS-Motor5.15±1.315.41±0.963.52±1.90.001GCS-Motor4.38±1.64.72±1.332.24±1.50.001 GCS12.59±3.3013.1±2.58.8±4.70.001GCS10.29±3.7511.0±3.25.7±3.60.001 Right-Pupillarynormalsize(%)155(83.8)140(87.5%)15(60)0.002Right-Pupillarynormalsize(%)144(77.8)135(84.4)9(36)0.001 Right-Pupillarynormalreaction(%)167(90.3)149(93.1)18(72)0.004Right-Pupillarynormalreaction(%)158(85.4)145(91.8)13(8.2)0.001 Left-Pupillarynormalsize(%)160(86.5)144(90)16(64)0.002Left-Pupillarynormalsize(%)158(85.4)145(91.8)13(8.2)0.001 Left-Pupillarynormalreaction(%)171(92.4)153(95.6)18(72)0.001Left-Pupillarynormalreaction(%)158(85.4)145(90.6)13(52)0.001 Respiratoryrate(permin)15.87±3.216.3±2.413.1±5.30.001Respiratoryrate16.24±4.416.9±3.011.9±8.30.001 SpO2(%)91.61±17.494.6±7.172.0±39.20.001SpO292.8±20.696.5±9.769.2±44.20.001 Heartrate(bpm)74.5±16.677.0±9.859.0±34.60.001Heartrate76.93±17.880.7±8.552.6±35.40.001 SBP(mmHg)108.97±30.4113.6±23.479.2±49.20.001SBP113.95±33119.8±23.976.3±53.50.001 DBP(mmHg)67.64±1970.4±14.549.7±31.30.001DBP70.4±22.2574.1±16.346.4±36.60.001 MAP(mmHg)81.42±22.284.8±16.759.5±36.90.001MAP84.92±22.289.3±17.656.4±41.60.001 SI0.7±0.120.6±0.10.5±0.30.007SI0.14±0.040.1±0.00.1±0.080.010 MSI0.91±0.290.9±0.20.7±0.40.017MSI0.93±0.440.9±0.30.7±0.70.088 RTS7.11±1.57.4±0.85.1±2.90.001RTS6.45±1.66.8±0.93.7±2.70.001 TRISS(%)81.95±23.185.6±17.158.4±38.40.001TRISS(%)75.66±24.880.4±19.145.2±34.60.001 NEWS5.89±3.65.2±2.59.8±6.20.001NEWS7.74±2.67.2±1.510.7±5.30.001

(5)

established in many studies; however, its association with short-term outcomes is unclear. The motor component of the GCS has been shown to be equally as sensitive as the GCS in many situations[10,21]. The GCS motor score was also associated with 2-week outcomes[10]as well as in-hospital mortality[23]and 6-month outcomes[2,6,14]. The fact that the GCS, in addition to other clinical parameters, may result in better predictive value has been reported by several scholars[19]. It

has also been reported that pupil reactivity and the GCS together have the highest ability to predict death[14,23]. Gomes et al. reported that the combination of the GCS, CT scanning results and hypotension may improve the prediction of 48-h mortality[24]. There is also the probability that the GCS’s validity can be inﬂuenced by ethanol, neurode- pressants and advanced age in TBI patients[25]. A recent study has shown that elderly TBI patients might have a higher GCS than younger Table 2

Univariate analysis of variables in the equation according to ENTER method (Ref = Survival).

Variable Odds Ratio 95% CI PValue Variable Odds Ratio 95% CI PValue

Lower Upper Lower Upper

Demographic

Age 0.984 0.963 1.005 0.984 Gender (female) 0.774 0.300 1.995 0.595

Prehospital Hospital

GCS-Eye 2.765 1.800 4.245 0.001 GCS-Eye 4.191 2.420 7.257 0.001

GCS-Verbal 2.353 1.673 3.309 0.001 GCS-Verbal 2.518 1.726 3.675 0.001

GCS-Motor 2.361 1.712 3.255 0.001 GCS-Motor 2.708 1.927 3.806 0.001

GCS 1.386 1.223 1.571 0.001 GCS 1.513 1.300 1.760 0.001

Right- Pupillary normal size 4.667 1.846 11.795 0.001 Right-Eye normal size 9.600 3.820 24.124 0.001

Right-Pupillary normal reactivity 5.268 1.813 15.302 0.002 Right-Eye normal reaction 8.923 3.459 23.020 0.001

Left-Pupillary normal size 5.062 1.926 13.305 0.001 Left-Eye normal size 6.139 2.504 15.048 0.001

Left-Pupillary normal reaction 8.500 2.676 27.003 0.001 Left-Eye normal reaction 8.923 3.459 23.020 0.001

Respiratory rate 1.284 1.135 1.453 0.001 Respiratory rate 1.233 1.117 1.361 0.001

SpO2 1.051 1.022 1.080 0.001 SpO2 1.041 1.021 1.061 0.001

Heart rate 1.044 1.021 1.067 0.001 Heart rate 1.065 1.043 1.150 0.001

SBP 1.034 1.018 1.051 0.001 SBP 1.039 1.022 1.057 0.001

DBP 1.045 1.024 1.067 0.001 DBP 1.047 1.027 1.067 0.001

MAP 1.043 1.023 1.063 0.001 MAP 1.046 1.027 1.065 0.001

SI normal 2.213 0.876 5.589 0.093 SI normal 1.608 0.671 3.853 0.286

MSI normal 5.437 2.052 14.407 0.001 MSI normal 6.110 2.462 15.163 0.001

RTS 2.167 1.486 3.160 0.001 RTS 2.898 1.920 4.375 0.001

TRISS 1.037 1.021 1.053 0.001 TRISS 1.047 1.030 1.065 0.001

NEWS 0.758 0.671 0.856 0.001 NEWS 0.682 0.567 0.821 0.001

Table 3

Multivariate analysis of variables according to hierarchical forward stepwise LR method (Ref = Survival).

Variable B S.E. Wald df Sig. Exp(B) 95% C.I. for EXP(B)

Lower Upper

Age −0.032 0.016 4.041 1 0.044 0.968 0.938 0.999

RTS (hospital) 1.147 0.234 24.018 1 0.000 3.149 1.990 4.981

Constant −3.567 1.323 7.269 1 0.007 0.028

Recoded phase

NEWS≤7 (prehospital) 1.256 0.536 5.48 1 0.019 3.511 1.227 10.043

RTS≥4 (hospital) 4.095 0.536 5.485 1 0.000 60.046 6.840 527.117

Constant −2.716 1.111 5.972 1 0.012 0.066

Table 4

Accuracy of acuity scoring tools (Ref = Survival).

Variable Area S.E. Sig. 95% C.I. Variable Area S.E. Sig. 95% C.I.

Lower Upper Lower Upper

RTS 0.793 0.054 0.001 0.687 0.898 RTS 0.870 0.048 0.001 0.776 0.964

RTS≤4 0.617 0.068 0.060 0.483 0.751 RTS≤4 0.677 0.068 0.004 0.543 0.811

TRISS 0.752 0.053 0.001 0.648 0.856 TRISS 0.826 0.041 0.001 0.745 0.907

TRISS≤50 0.626 0.067 0.044 0.494 0.757 TRISS≤50 0.690 0.065 0.002 0.563 0.817

MAP 0.625 0.078 0.091 0.471 0.778 MAP 0.617 0.078 0.112 0.464 0.771

MAP (7–10) 0.599 0.065 0.113 0.471 0.727 MAP (7–10) 0.668 0.062 0.007 0.546 0.789

SI 0.363 0.082 0.065 0.202 0.525 SI 0.473 0.083 0.718 0.311 0.636

SI (0.5–0.7) 0.591 0.059 0.143 0.476 0.707 SI (0.5–0.7) 0.599 0.060 0.110 0.483 0.716

MSI 0.343 0.087 0.659 0.296 0.639 MSI 0.467 0.087 0.659 0.296 0.639

MSI (0.7–1.3) 0.653 0.066 0.014 0.523 0.783 MSI (0.7–1.3) 0.674 0.065 0.005 0.548 0.801

NEWS 0.434 0.078 0.374 0.281 0.588 NEWS 0.434 0.078 0.374 0.281 0.588

NEWS≤7 0.679 0.063 0.004 0.555 0.803 NEWS≤7 0.592 0.060 0.140 0.474 0.710

ISS-Hospital 0.506 0.060 0.926 0.388 0.623 ISS≤50 0.458 0.059 0.495 0.342 0.573

(6)

TBI patients[26]. Overall, it may be concluded that GCS in prehospital setting might not have clinical applicability as same as those in hospital setting.

4.2. Shock index & modiﬁed shock index

The limited ability of vital signs to predict 24-h mortality has also been emphasized by others [15]. Some authors have observed that, even though individual vital signs cannot predict the severity of bleeding, the SI seems to provide a clinically useful tool for predicting shock in trauma patients[15,27,28]. Another study also indicated that SI is associated with a much higher odds ratio of SI (9.47) than the odds ratio of HR (1.06) in trauma patients[27]. Our results did not indicate that either the SI or the MSI predicted mortality with any signiﬁcance.

That result is unsurprising because head injuries were the main ones sustained by our patients, and these can be accompanied by less severe internal bleeding than what is seen in multiple trauma patients in general. Therefore, SBP and HR, by compromising the major value of SI and MSI, may cause mild changes in response to haemorrhage among TBI patients, resulting in nonsigniﬁcant association of SI and MSI with 24-h mortality.

4.3. Revised trauma score

The RTS, especially≤4, predicted 24-h mortality in the emergency hospital setting among TBI patients in our results. The RTS is composed of the GCS, SBP and RR; therefore, it represents the function of all three main organs including the brain, the heart and the lungs. The RTS has a greater predictive value of mortality in comparison to its subsets. It has good discriminatory power to identify injuries in three main organs as well as the severity of injuries[4]. An RTS≤4 suggests the need to transfer a patient to a Level 1 trauma centre. Thisfinding is supported by other studies, which indicate that the RTS significantly predicts mortality in traumatic patients[29]. The RTS in the hospital was the best predictor of mortality in the current study, which may have been due to the GCS, RR and HR alternations needing time to change the RTS among TBI patients. Additionally, our patients were victims of traffic accidents in urban areas, requiring consideration of prompt response in thefield as well. Therefore, it is expected that RTS changes in critically ill patients have been remarkable in the ED, resulting in a good AUC (0.870, 95%CI: 0.776–0.964) and OR (3.149, 95% CI: 1.990–4.981).

Although RTS’ have shown good validity in trauma patients, some studies have questioned their usefulness[8]. It was not a valid measure of patients using Helicopter Emergency Medical Services, resulting in a high rate of erroneous triage of traumatic patients. Overall, the RTS is a good tool for triaging severe TBI patients in hospital emergency care;

hence, patients with an RTS of≤4 should be considered critically ill.

4.4. National early warning score

The NEWS, especially when≥7, adequately predicts mortality in a prehospital setting among TBI patients. All three main systems, including the brain, the heart and the lungs, contribute to theﬁnal score;

however, the respiratory system has the greatest share (eight out of twenty). Our results indicate that the NEWS predicts mortality in the emergency prehospital setting but not the emergency hospital setting.

This may be because paramedics perform invasive interventions, such as extensive respiratory interventions, including airway stabilization, suctioning, oxygenation, intubation and positive pressure ventilation, to stabilize patients. Therefore, certain signs and symptoms of critical conditions may have been alleviated prior to their arrival at the hospital. Interventions artiﬁcially decrease the NEWS and weaken associations between the NEWS in the hospital and mortality. The NEWS has shown various thresholds in diﬀerent studies with heterogeneous sample sizes. Hodgson et al. indicated that a NEWS≥5 predicts mortality in patients with acute exacerbation of COPD[16]. Alam et al. also indicated that a NEWS≥7 has an AUC of 0.768 (0.618–0.919) for mortality in the ED[30]. The results supported the usefulness of the NEWS in relation to identifying critically ill patients. Overall, Abbott et al. concluded that the NEWS should be reviewed to produce an up- dated threshold for identifying patients with critically ill conditions [31]. We believe that our results can be very useful because it suggests that paramedics may use respiratory distress as a valid criteria to identify short-term mortality risk in TBI patients.

4.5. Injury severity score & the trauma and injury severity score

The TRISS has shown a good diagnostic validity, with an AUC of 0.826 (95% CI: 0.745–0.907) in the ED among TBI patients; thus, it is used to improve outcome predictions after trauma[32,33]. Some studies have questioned the usefulness of the TRISS, indicating that it overestimates mortality in patients with high a TRISS[34]. Conversely, our results do not indicate that the TRISS predicts mortality, but it does show good diagnostic validity. This disparity may be related to the fact that the ISS is part of the TRISS; however, the ISS for TBI patients does not signiﬁcantly predict mortality because it treats all injuries rather equally, while brain injury may have greater association with mortality than other injuries. For example, patients with a low ISS, such as those with an isolated TBI, may have a high mortality rate. This hypothesis may be conﬁrmed because the ISS has poor diagnostic validity for TBI patients (AUC 0.506, 95% CI: 0.388–0.623). Notably, the TRISS is not useful in the prehospital setting because the ISS, as part of the TRISS, needs to be calculated based on diagnostic examinations in the hospital.

4.6. Limitations

Acuity scoring tools, such as the RTS, NEWS, SI, MSI and TRISS, Table 5

Diagnostic validity of acuity scoring tools based on mortality.

Variable Sensitivity (%) Speciﬁcity (%) PPV (%) NPV (%) Variable Sensitivity (%) Speciﬁcity (%) PPV (%) NPV (%)

RTS≤4 85.71 (42.13–99.64)

89.33 (83.83–93.45)

24.00 (1.79–34.73)

99.37 (96.28–99.90)

RTS≤4 90.00 (55.50–99.75)

90.86 (85.58–94.68)

36.00 (25.24–48.38)

99.38 (96.12–99.90) MAP (7–10) 25.71

(12.49–43.26) 89.33 (83.26–93.78)

36.00 (21.34–53.84)

83.75 (80.80–86.32)

MAP (7–10) 28.00 (16.23–42.49)

91.85 (85.89–95.86)

56.00 (38.25–72.33)

77.50 (74.21–80.48) SI (0.5–0.7) 17.31

(10.59–25.96) 91.36 (83.00–96.45)

72.00 (53.02–85.42)

46.25 (43.52–49.01)

SI (0.5–0.7) 18.09 (10.90–27.37)

91.21 (83.41–96.13)

68.00 (49.11–82.39)

51.87 (49.02–54.72) MSI (0.7–1.3) 40.00

(21.13–61.33) 90.62 (85.01–94.66)

25.24 (25.24–56.82)

90.62

(87.49–93.04) MSI (0.7–1.3) 36.36 (20.40–54.88)

91.45 (85.82–95.37)

48.00 (31.68–64.76)

86.88 (83.58–89.59) NEWS≤7 33.33

(19.09–50.22) 91.78 (86.08–95.68)

52.00 (34.97–68.58)

83.75 (80.42–86.61)

NEWS≤7 17.98 (10.64–27.55)

90.62 (82.95–95.62)

64.00 (45.30–79.24)

54.37 (51.47–57.25) TRISS≤50 42.11

(20.25–66.5) 89.76 (84.11–93.92)

32.00 (19.04–48.49)

93.12 (90.20–95.23)

ISS≤50 4.76 (0.12–23.82)

85.37 (79.01–90.39)

4.00 (0.59–22.62)

87.50 (86.19–88.70)

(7)

have been derived from outcomes in extensive trauma data banks in developed countries; therefore, it is important for these tools to be examined in different settings. Our study was limited to one hospital trauma centre in a densely-populated area of a metropolitan city. We could not extend the study period further because strict supervision was required to collect reliable data. Therefore, other studies may want to replicate this study with larger sample sizes. Another limitation of the present study was the fact that the paramedics collected data in the field and we used the documented data in the ED. However, paramedics are well trained and educated as well, but some differences may be attributed to inter-observer disagreement. The differences may not be significant because no statistically significant differences were observed between advanced providers (i.e. nurses and paramedics) in this regard [35].

5. Conclusions

The RTS and the NEWS may be used to triage TBI patients in hospital and prehospital emergency care respectively because of their signiﬁcant validity; therefore, traditional vital sign criteria may be of limited use for the triage of TBI patients especially in prehospital setting. Acuity scoring tools help clinicians stratify patients as well as al- locate resources. Accurate triage is critical, especially for TBI patients, for ﬁnding the most relevant trauma centre. It is therefore recommended that clinicians in both prehospital and emergency department use acuity scoring tools including RTS and NEWS in routine practice.

Author contributions

A.M. and Z.N. designed the study. Z.N., H.Z. and A.M. collected data. A.M. analysed the data which Z.N., H.Z. and A.M. interpreted the result. Z.N., H.Z. and A.M. did perform the literature review and wrote the manuscript. All authors proofread the manuscript and revised it critically. All authors have read and approved theﬁnal manuscript.

References

[1] Weber KT, Guimarães VA, Pontes Neto OM, Leite JP, Takayanagui OM, Santos- Pontelli TE. Predictors of quality of life after moderate to severe traumatic brain injury. Arq. Neuropsiquiatr. 2016;74:409–15.

[2] Husson E, Ribbers G, Willemse-van Son A, Verhagen A, Stam H. Prognosis of six- month functioning after moderate to severe traumatic brain injury: a systematic review of prospective cohort studies. J. Rehabil. Med. 2010;42:425–36.

[3] McIntyre A, Aubut J, Dijkers M, Teasell R. Mortality among older adults after a traumatic brain injury: a meta-analysis. Brain Inj. 2013;27:31–40.

[4] Lichtveld R, Spijkers A, Hoogendoorn J, Panhuizen I, van der Werken C. Triage revised trauma score change betweenﬁrst assessment and arrival at the hospital to predict mortality. Int. J. Emergency Med. 2008;1:21–6.

[5] Mirhaghi A, Mirhaghi M, Oshio A, Sarabian S. Systematic review of the personality proﬁle of paramedics: bringing evidence into emergency medical personnel re- cruitment policy. Eurasian J. Emergency Med. 2016;15:144–9.

[6] Majidi S, Siddiq F, Qureshi A. Prehospital neurologic deterioration is independent predictor of outcome in traumatic brain injury: analysis from national trauma data bank. Am. J. Emerg. Med. 2013;31:1215–9.

[7] Steyerberg E, Mushkudiani N, Perel P, Butcher I, Lu J, McHugh G, et al. Predicting outcome after traumatic brain injury: development and international validation of prognostic scores based on admission characteristics. PLoS Med. 2008;5:e165. discussion e.

[8] Zafonte R, Hammond F, Mann N, Wood D, Millis S, Black K. Revised trauma score:

an additive predictor of disability following traumatic brain injury? Am. J. Phys.

Med. Rehabil. 1996;75:456–61.

[9] Malekzadeh J, Shafaee H, Behnam H, Mirhaghi A. The eﬀect of Cincinnati prehospital stroke scale on telephone triage of stroke patients: evidence-based practice in emergency medical services. Int. J. Evidence Based Healthcare 2015;13:87–92.

[10] Timmons S, Bee T, Webb S, Diaz-Arrastia R, Hesdorﬀer D. Using the abbreviated injury severity and glasgow coma scale scores to predict 2-week mortality after traumatic brain injury. J. Trauma 2011;71:1172–8.

[11] Kamal V, Agrawal D, Pandey R. Prognostic models for prediction of outcomes after traumatic brain injury based on patients admission characteristics. Brain Inj.

2016;22:1–14.

[12] Majdan M, Brazinova A, Rusnak M, Leitgeb J. Outcome prediction after traumatic brain injury: comparison of the performance of routinely used severity scores and multivariable prognostic models. J. Neurosci. Rural Pract. 2017;8:20–9.

[13] Majdan M, Mauritz W, Brazinova A, Rusnak M, Leitgeb J, Janciak I, et al. Severity and outcome of traumatic brain injuries (TBI) with diﬀerent causes of injury. Brain Inj. 2011;25:797–805.

[14] Majdan M, Steyerberg E, Nieboer D, Mauritz W, Rusnak M, Lingsma H. Glasgow coma scale motor score and pupillary reaction to predict six-month mortality in patients with traumatic brain injury: comparison ofﬁeld and admission assessment.

J. Neurotrauma 2015;32:101–8.

[15] Liu Y, Liu J, Fang Z, Shan G, Xu J, Qi Z, et al. Modiﬁed shock index and mortality rate of emergency patients. World J. Emerg. Med. 2012;3:114–7.

[16] Hodgson L, Dimitrov B, Congleton J, Venn R, Forni L, Roderick P. A validation of the national early warning score to predict outcome in patients with COPD exacerbation. Thorax 2016;71:208–36.

[17] Fitzgerald M, Cameron P, Mackenzie C, Farrow N, Scicluna P, Gocentas R, et al.

Trauma resuscitation errors and computer-assisted decision support. Arch. Surg.

2011 Feb;146:218–25.

[18] Sugerman D, Xu L, Pearson W, Faul M. Patients with severe traumatic brain injury transferred to a Level I or II trauma center: United States, 2007 to 2009. J. Trauma Acute Care Surg. 2012;73:1491–9.

[19] Kamal V, Agrawal D, Pandey R. Epidemiology, clinical characteristics and outcomes of traumatic brain injury: evidences from integrated level 1 trauma center in India.

J. Neurosci. Rural Pract. 2016;7:515–25.

[20] Chamoun R, Robertson C, Gopinath S. Outcome in patients with blunt head trauma and a glasgow coma scale score of 3 at presentation. J. Neurosurg. 2009;111:683–7.

[21] Moazzez Lesko MJT, Brien S, Childs C. Comparing model performance for survival prediction using total GCS and its components in traumatic brain injury. J.

Neurotrauma 2012;1–23.

[22] Baltas I, Gerogiannis N, Sakellariou P, Matamis D, Prassas A, Fylaktakis M. Outcome in severely head injured patients with and without multiple trauma. J. Neurosurg.

Sci. 1998;42:85–8.

[23] Hoﬀmann M, Lefering R, Rueger J, Kolb J, Izbicki J, Ruecker A, et al. Pupil evaluation in addition to glasgow coma scale components in prediction of traumatic brain injury and mortality. Br. J. Surg. 2012;99:122–30.

[24] Gómez PA, de-la-Cruz J, Lora D, Jiménez-Roldán L, Rodríguez-Boto G, Sarabia R, et al. Validation of a prognostic score for early mortality in severe head injury cases:

clinical article. J. Neurosurg. 2014;121:1314–22.

[25] Barmparas G, Liou D, Lamb A, Gangi A, Chin M, Ley E, et al. Prehospital hy- pertension is predictive of traumatic brain injury and is associated with higher mortality. J. Trauma Acute Care Surg. 2014;77:592–8.

[26] Kehoe A, Rennie S, Smith J. Glasgow coma scale is unreliable for the prediction of severe head injury in elderly trauma patients. Emerg. Med. J. 2015;32:613–5.

[27] Rau C, Wu S, Kuo S, Pao-Jen K, Shiun-Yuan H, Chen Y, et al. Prediction of massive transfusion in trauma patients with shock index, modiﬁed shock index, and age shock index. Int. J. Environ. Res. Public Health 2016;13. pii: E683.

[28] Mutschler M, Nienaber U, Münzberg M, WölﬂC, Schoechl H, Paﬀrath T, et al. The shock index revisited - a fast guide to transfusion requirement? A retrospective analysis on 21,853 patients derived from the TraumaRegister DG. Crit. Care 2013;17:R172.

[29] Heydari Khayat N, ShariﬁPoor H, Rezaei M, Mohammadinia N, Darban F.

Correlation of revised trauma score with mortality rate of traumatic patients within theﬁrst 24 h of hospitalization. Zahedan J. Res. Med. Sci. 2014;16:33–6.

[30] Alam N, Vegting I, Houben E, van Berkel B, Vaughan L, Kramer M, et al. Exploring the performance of the national early warning score (NEWS) in a European emergency department. Resuscitation 2015;90:111–5.

[31] Abbott T, Vaid N, Ip D, Cron N, Wells M, Torrance H, et al. A single-centre observational cohort study of admission national early warning score (NEWS).

Resuscitation 2015;92:89–93.

[32] Chawda M, Hildebrand F, Pape H, Giannoudis P. Predicting outcome after multiple trauma: which scoring system? Injury 2004;35:347–58.

[33] Darbandsar Mazandarani P, Heydari K, Hatamabadi H, Kashani P, Jamali Danesh Y.

Acute physiology and chronic health evaluation (APACHE) III score compared to trauma-injury severity score (TRISS) in predicting mortality of trauma patients.

Emergency (Tehran) 2016;4:88–91.

[34] Vassar M, Wilkerson C, Duran P, Perry C, Holcroft J. Comparison of APACHE II, TRISS, and a proposed 24-hour ICU point system for prediction of outcome in ICU trauma patients. J. Trauma Acute Care Surg. 1992;32:490–500.

[35] Báez AA, Giráldez EM, De Peña JM. Precision and reliability of the glasgow coma scale score among a cohort of Latin American prehospital emergency care providers. Prehosp. Disaster Med. 2007;22:230–2.