Highlights
-
•
A clinical score was developed to predict massive blood transfusion (MBT).
-
•
Key predictors included injury severity, wound type, major injured body part, airway, fluid, and EMS times.
-
•
The score stratifies patients into low-, moderate-, and high-risk groups.
-
•
High predictive accuracy with AuROC of 0.883.
-
•
Applicable in EMS systems with limited resources.
Keywords: ROC Curve, Calibration, Emergency Medical Services, Blood Transfusions, Hemorrhages
Abstract
Background
Most prehospital trauma deaths result from exsanguination. Developing a scoring system to predict massive blood transfusion (MBT) is essential in urban areas of middle- to lower-income developing countries where traffic accidents are frequent.
Aim
To develop and validate a clinical score for predicting MBT in prehospital trauma patients.
Methods
Data were obtained from emergency medical service (EMS) patient care reports between January 2018 and October 2023. Multiple logistic regression was used to develop the score. Discrimination was assessed using receiver operating characteristics (ROC) curve analysis, and the area under the ROC curve (AuROC) with 95 % confidence interval was reported. Risk groups were defined by positive likelihood ratio (LR + ).Results: Six hundred patients were included. Significant predictors were injury severity score, wound type, serious injury body part, scene distance ≥ 4 km, response time < 8 min, hospital transfer time < 8 min, airway management, and fluid access. The score classified patients into low- (≤3), moderate- (4–9), and high-risk (≥10) groups, with MBT probabilities of 1.6 %, 20.7 %, and 74.2 %, respectively. LR + values were 0.07, 1.08, and 11.87. The AuROC was 0.883.
Conclusion
This study derived a prehospital clinical score that demonstrated good discrimination for predicting MBT. Internal validation using bootstrap resampling confirmed the robustness of the model. Nevertheless, external and prospective validation in diverse trauma populations and EMS settings are required before the score can be considered for clinical application.
Introduction
Injury is one of the main global causes of death and an impactful burden on health service systems, affecting society and the economy. It acts as an obstacle to the prosperity of countries.1 Regarding World Health Organization statistics, Thailand has the highest and second-highest mortality from accidents in Asia and the world, respectively.2 Road accidents are the primary cause of death in developing countries.3 Most patients with prehospital injury die of hemorrhage, which is the main cause of mortality within the first 24 h after injury.4 A previous study found that resuscitation with blood components in a proportion of packed red blood cells (PRBCs) to fresh frozen plasma of 1:1 to 2:1 significantly reduced the mortality rate from exsanguination.4 Multisystem injury is the primary cause of sudden death among patients with injury from accidents, particularly when massive hemorrhage occurs. Patients who require massive blood transfusion (MBT) are defined as those receiving whole blood or PRBCs of at least 10 units or the volume of blood in the body within 24 h.5 The MBT guideline protocol is a predefined protocol between physicians and blood banks with the purpose of immediate blood and blood component transfusion to the injured and reduction of unnecessary blood component transfusion. Generally, physicians will clinically decide based on clinical parameters, such as hypotension and tachycardia.6 A previous study developed an assessment of blood consumption (ABC) score, with an area under the receiver operating characteristic (ROC) curve (AuROC) of 0.842.7 Then, a previous study conducted at a Thai emergency department performed external validation of the ABC score and found an AuROC of 0.587.8 According to a literature review, the MBT score was mostly studied and validated in developed countries, and all studies were conducted in emergency departments9 and other settings within hospitals, such as intensive care units and operation rooms10 where some parameters, including radiographic imaging and laboratory results, are obtained, which cannot be applied in the context of prehospital care by emergency medical service (EMS) staff in developing countries, such as Thailand, with many limitations, including restricted time in prehospital patient care and EMS personnel who are not physicians but paramedics and emergency registered nurses, resulting in difficulty in making the decision whether an injured person will have a chance to receive prehospital MBT. This decision relies on the consideration of the injured individual’s parameters, vital signs, injury mechanism, physical examination, and initial treatment at the scene. We believe that the development of a score will be beneficial for appropriate clinical decisions regarding the selection of proper destined trauma centers with available MBT.
Therefore, this study aimed to develop and validate a risk score using prehospital risk factors for predicting MBT in individuals who were injured by accidents in a middle- to lower-income country.
Objectives
This study aimed to develop and validate a risk score using prehospital risk factors for predicting MBT in individuals who were injured by accidents.
Methods
Study design and settings
This retrospective cohort study was conducted at the Comprehensive Life Support (CLS), Vajira Emergency Medical Service (V-EMS), Faculty of Medicine, Vajira Hospital, Navamindradhiraj University, Bangkok, Thailand. V-EMS is the zone leader of EMS in 1 of 11 areas in the area division of the EMS system in Bangkok.1
The responsible area of V-EMS covers 50 km2, including networking hospitals in emergency medical operation with both public and private hospitals of 6 hospitals and a total population of 500,000. In operation for trauma cases, one V-EMS team includes at least three personnel, including paramedics or emergency nurse practitioners (ENPs) as an operational team leader and at least 2 emergency medical technicians as an ambulance driver and team assistant, with consultation system through telemedicine with emergency physicians for 24 h. Furthermore, every V-EMS personnel should pass specialized injured patient management training, which is a prehospital trauma life support (PHTLS) program.
We used a standard data reporting format according to the Standards for the Reporting of Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) statement.11
Participants
Data of adults who were injured by accidents were obtained from EMS patient care reports, coded using the Thailand Emergency Medical Triage Protocol and Criteria Based Dispatch (TEMTP&CBD) symptom group 21–25 assisted by V-EMS between January 1, 2018, and October 31, 2023.
Eligibility criteria
Adults who were injured by accidents aged at least 18 years, were assisted by V-EMS, and coded with the TEMTP&CBD symptom group 21–25 were included in the study. All injured individuals should be transported to the Emergency Department of Vajira Hospital.
Exclusion criteria
Injured individuals with incomplete or missing data, those who denied treatment, those who died at the scene, and burned and drowning patients were excluded from the study.
Data collection
Data on the injured, EMS staff, and EMS operations were collected from EMS patient care reports, which were a record of advanced EMS operation, Bangkok EMS (Erawan Center), and the standard form and unit in the Bangkok Advanced Emergency Operation Unit. This form contained data on EMS operation units, patients, and all treatments by EMS teams, recorded by dispatchers, paramedics, or ENPs operating in the scene. These data were a part of the remuneration for the EMS operation units. All data were filled in and recorded in Microsoft Excel by the principal investigator. Data comprised general characteristics of trauma patients, including sex, age, mechanism of injury, systolic blood pressure (SBP), heart rate, oxygen saturation, Glasgow Coma Scale score (GCS), point-of-care glucose test, shock index (SI), injury severity score (ISS), wound type, bone injury type, hemorrhage type, serious injury body part, distance from the base station to the scene, distance from the scene to the hospital, response time, scene time, transfer to hospital time, hemorrhage control, airway management, fluid access, immobilization, prehospital cardiopulmonary resuscitation (prehospital CPR), and MBT in hospital. Patients who died in hospital before receiving a MBT were classified into the non-MBT group, consistent with the definitions applied in previous studies.12
Sample size determination
The main aim of this study was to develop a risk score for predicting MBT in injured individuals using multiple logistic regression analysis. The authors estimated the sample size using the number of events per variable in the logistic regression analysis.13 The analysis requires the number of samples with studied events of at least 10 per independent variable. According to Kotnarin R,14 there are eight independent variables. Therefore, in the analysis, the number of samples with studied events should be 80 patients receiving MBT, and the proportion of MBT in individuals who were injured by accidents was 14.1 %.14 Thus, the total sample size used in the analyses was at least 568 ((80 × 100)/14.1 = 568). Because of the retrospective nature of the study, an additional 5 % was added to prevent incomplete data. Sample size calculated from the formula (nnew = 568/(1 − 0.05)) was 600.
Statistical analysis
A descriptive analysis was performed to examine the variable distribution. Continuous variables are presented as means ± standard deviations or medians and interquartile ranges (IQRs), whereas categorical variables are presented as frequencies and proportions. When comparing the two groups, differences were evaluated using the independent t-test or Mann–Whitney U test for numeric variables and the chi-square test or Fisher’s exact test for categorical variables.
Development of a risk score predicting MBT in individuals who were injured by accidents uses multivariable analysis through multiple logistic regression analysis, reported with regression coefficient, odds ratio (OR), and 95 % confidence interval (CI) with p-value. Backward elimination is then performed. Variables associated with MBT from the univariable analysis with p-values < 0.2 were considered for the analysis, and non-associated independent variables were removed one by one from the Wald test. The properties of the equation were reported using R2 and the Hosmer–Lemeshow goodness-of-fit test, and the ROC curve was used. The area under the ROC curve (AuROC) and 95 % CI were reported. The regression coefficient was then used to develop the risk score. For validation of the risk score, discrimination was assessed using ROC analysis, with results reported as AuROC and 95 % CI, and stratified into three risk levels—low, medium, and high—according to the positive likelihood ratio (LR + ) with 95 % CI. Internal validation was performed using bootstrap resampling methods. Calibration of the risk score for predicting MBT in trauma patients was evaluated through regression calibration with bootstrap methods and presented using calibration plots.
All statistical tests were considered statistically significant at a p-value ≤ 0.05. Stata (version 17.0; StataCorp College Station, TX, USA) was used for all analyses.
Definition of massive blood transfusion
MBT was defined as any of the following: the transfusion of > 10 units of PRBCs within 24 h from the time of the EMS call, the transfusion of > 4 units of PRBCs in 1 h when an ongoing need was foreseeable, or the replacement of 50 % of the total blood volume or 5 units of PRBCs within 3 h, all measured from the time of the EMS call.14, 15, 16
Ethical statement
This study was conducted according to the tenets of the 1975 Declaration of Helsinki and its revisions in 2000. This study was approved by the Institutional Review Board of the Faculty of Medicine Vajira Hospital, Navamindradhiraj University (COA no. 042/2567). The informed consent requirement was waived because of the retrospective nature of this study and anonymity of all patient data.
Results
Of 735 adult trauma patients managed by V-EMS, 600 met the inclusion criteria and were analyzed: 117 in the MBT group and 483 in the non-MBT group (Fig. 1).
Fig. 1.
Flow chart.
The mean age of patients was 46.1 ± 22.0 years, and 71.5 % were male. Baseline characteristics of the study population are summarized in (Table 1).
Table 1.
Demographic and clinical characteristics of trauma patients (n = 600).
| Characteristics | n | (%) |
|---|---|---|
| Age (years), mean ± SD | 46.05 ± 22.04 | |
| <60 | 429 | (71.5) |
| ≥60 | 171 | (28.5) |
| Sex | ||
| Male | 399 | (66.5) |
| Female | 201 | (33.5) |
| Mechanism of the injury | ||
| Blunt | 454 | (75.7) |
| Penetrating | 146 | (24.3) |
| Systolic blood pressure | ||
| >90 mmHg | 544 | (90.7) |
| ≤90 mmHg | 56 | (9.3) |
| Heart rate | ||
| ≤120 bpm | 509 | (84.8) |
| >120 bpm | 91 | (15.2) |
| Oxygen saturation | ||
| >94 % | 449 | (74.8) |
| ≤94 % | 151 | (25.2) |
| Glasgow coma score, median (IQR) | 15 | (10 – 15) |
| POCT glucose (mg%), median (IQR) | 126 | (106 – 154) |
| Shock Index (SI), median (IQR) | 0.7 | (0.5 – 0.8) |
| Injury severity score (ISS), median (IQR) | 4 | (2 – 9) |
| Type of wounds | ||
| No | 66 | (11.0) |
| Cut/Laceration | 231 | (38.5) |
| Abrasion | 193 | (32.2) |
| Contusion | 104 | (17.3) |
| Burn | 5 | (0.8) |
| Amputation | 1 | (0.2) |
| Type of bone injury | ||
| No | 427 | (71.2) |
| Close fracture | 103 | (17.2) |
| Open fracture | 48 | (8.0) |
| Dislocation | 22 | (3.7) |
| Type of hemorrhage | ||
| No | 299 | (49.8) |
| External stopped | 231 | (38.5) |
| External active | 47 | (7.8) |
| Internal hemorrhage | 23 | (3.8) |
| Serious injury body part | ||
| Abdomen | 38 | (6.3) |
| Chest/clavicle | 55 | (9.2) |
| External body surface | 27 | (4.5) |
| Extremity | 332 | (55.3) |
| Face | 119 | (19.8) |
| Head/neck | 217 | (36.2) |
| Pelvis | 29 | (4.8) |
| Spine | 49 | (8.2) |
| Multiple injury | 28 | (4.7) |
| Distance from base station to scene (km), median (IQR) | 3 | (1–5) |
| Distance from scene to hospital (km), median (IQR) | 4 | (2–6) |
| Response time (min), median (IQR) | 8 | (4–12) |
| Scene time (min), median (IQR) | 14 | (9–23) |
| Transfer to hospital time (min), median (IQR) | 7 | (3–11) |
| Hemorrhage control | ||
| No | 296 | (49.3) |
| Pressure dressing | 163 | (27.2) |
| Dressing | 133 | (22.2) |
| Pelvic binder | 4 | (0.7) |
| Tourniquet application | 4 | (0.7) |
| Airway management | ||
| No | 403 | (67.2) |
| Mask with a bag | 152 | (25.3) |
| Endotracheal tube | 42 | (7.0) |
| Laryngeal mask airway | 3 | (0.5) |
| Fluid access | ||
| No | 263 | (43.8) |
| RLS | 253 | (42.2) |
| NSS | 79 | (13.2) |
| Acetar | 5 | (0.8) |
| Immobilization | 377 | (62.8) |
| prehospital CPR | 26 | (4.3) |
Data are presented as numbers (%), means ± standard deviations, or medians (interquartile ranges).
Abbreviations: LRS, lactated Ringer’s solution: NSS, normal saline solution; min, minute; km, kilometer; IQR, interquartile range; cardiopulmonary resuscitation; CPR.
Comparisons between MBT and non-MBT patients demonstrated significant differences in vital signs, GCS, ISS, injury patterns, and prehospital interventions (Table 2). MBT patients more frequently presented with hypotension, hypoxemia, higher severity indices, serious injuries across multiple regions, and received more prehospital procedures including hemorrhage control, airway management, vascular access, and CPR.
Table 2.
Characteristics of trauma patients according to MBT (n = 600).
| Variables | MBT(n = 117) | Non-MBT(n = 483) | P-value | ||
|---|---|---|---|---|---|
| Age (years), Mean ± SD | 44.79 ± 22.61 | 46.35 ± 21.91 | 0.493 | ||
| <60 | 86 | (73.5) | 343 | (71.0) | 0.592 |
| ≥60 | 31 | (26.5) | 140 | (29.0) | |
| Sex | |||||
| Male | 81 | (69.2) | 318 | (65.8) | 0.485 |
| Female | 36 | (30.8) | 165 | (34.2) | |
| Mechanism of the injury | |||||
| Blunt | 93 | (79.5) | 361 | (74.7) | 0.283 |
| Penetrating | 24 | (20.5) | 122 | (25.3) | |
| Systolic blood pressure | |||||
| >90 mmHg | 92 | (78.6) | 452 | (93.6) | <0.001 |
| ≤90 mmHg | 25 | (21.4) | 31 | (6.4) | |
| Heart rate | |||||
| ≤120 bpm | 96 | (82.1) | 413 | (85.5) | 0.350 |
| >120 bpm | 21 | (17.9) | 70 | (14.5) | |
| Oxygen saturation | |||||
| >94 % | 62 | (53.0) | 387 | (80.1) | <0.001 |
| ≤94 % | 55 | (47.0) | 96 | (19.9) | |
| Glasgow coma score, median (IQR) | 12 | (5–15) | 15 | (11–15) | <0.001 |
| POCT glucose (mg%), median (IQR) | 128 | (119–174) | 123 | (105–150) | 0.004 |
| Shock Index (SI), median (IQR) | 0.8 | (0.7–1) | 0.6 | (0.5–0.8) | <0.001 |
| Injury severity score (ISS), median (IQR) | 16 | (9–19) | 4 | (1–6) | <0.001 |
| Type of wounds | |||||
| Abrasion | 10 | (8.5) | 183 | (37.9) | <0.001 |
| Cut/laceration | 66 | (56.4) | 165 | (34.2) | |
| Other | 23 | (19.7) | 87 | (18.0) | |
| No | 18 | (15.4) | 48 | (9.9) | |
| Type of bone injury | |||||
| No | 62 | (53.0) | 365 | (75.6) | <0.001 |
| Close fracture | 27 | (23.1) | 76 | (15.7) | |
| Open fracture | 18 | (15.4) | 30 | (6.2) | |
| Dislocation | 10 | (8.5) | 12 | (2.5) | |
| Type of hemorrhage | |||||
| No | 31 | (26.5) | 268 | (55.5) | <0.001 |
| External stopped | 55 | (47.0) | 176 | (36.4) | |
| External active | 20 | (17.1) | 27 | (5.6) | |
| Internal hemorrhage | 11 | (9.4) | 12 | (2.5) | |
| Serious injury body part | |||||
| Abdomen | 20 | (17.1) | 18 | (3.7) | <0.001 |
| Chest/clavicle | 19 | (16.2) | 36 | (7.5) | 0.003 |
| External body surface | 15 | (12.8) | 12 | (2.5) | <0.001 |
| Extremity | 60 | (51.3) | 272 | (56.3) | 0.326 |
| Face | 31 | (26.5) | 88 | (18.2) | 0.044 |
| Head/neck | 58 | (49.6) | 159 | (32.9) | 0.001 |
| Pelvis | 7 | (6.0) | 22 | (4.6) | 0.518 |
| Spine | 12 | (10.3) | 37 | (7.7) | 0.358 |
| Multiple injury | 12 | (10.3) | 16 | (3.3) | 0.001 |
| Distance from base station to scene (km), median (IQR) | 4 | (2 – 5) | 3 | (1 – 5) | 0.232 |
| <4 | 57 | (48.7) | 280 | (58.0) | 0.070 |
| ≥4 | 60 | (51.3) | 203 | (42.0) | |
| Distance from scene to hospital (km), median (IQR) | 4 | (2 – 5) | 4 | (2 – 6) | 0.829 |
| <4 | 55 | (47.0) | 239 | (49.5) | 0.631 |
| ≥4 | 62 | (53.0) | 244 | (50.5) | |
| Response time (min), median (IQR) | 7 | (5 – 10.5) | 8 | (4 – 12) | 0.485 |
| <8 | 68 | (58.1) | 217 | (44.9) | 0.010 |
| ≥8 | 49 | (41.9) | 266 | (55.1) | |
| Scene time (min), median (IQR) | 15 | (10 – 24) | 14 | (8 – 23) | 0.109 |
| <15 | 58 | (49.6) | 245 | (50.7) | 0.823 |
| ≥15 | 59 | (50.4) | 238 | (49.3) | |
| Transfer to hospital time (min), median (IQR) | 6 | (3 – 9) | 7 | (4 – 12) | 0.063 |
| <8 | 79 | (67.5) | 261 | (54.0) | 0.008 |
| ≥8 | 38 | (32.5) | 222 | (46.0) | |
| Hemorrhage control | |||||
| No | 45 | (38.5) | 251 | (52.0) | <0.001 |
| Pressure dressing | 57 | (48.7) | 106 | (21.9) | |
| Dressing | 9 | (7.7) | 124 | (25.7) | |
| Pelvic binder | 2 | (1.7) | 2 | (0.4) | |
| Tourniquet application | 4 | (3.4) | 0 | (0.0) | |
| Airway management | |||||
| No | 64 | (54.7) | 339 | (70.2) | 0.008 |
| Mask with a Bag | 43 | (36.8) | 109 | (22.6) | |
| Endotracheal tube | 10 | (8.5) | 32 | (6.6) | |
| Laryngeal mask airway | 0 | (0.0) | 3 | (0.6) | |
| Fluid access | |||||
| No | 21 | (17.9) | 242 | (50.1) | <0.001 |
| RLS | 75 | (64.1) | 178 | (36.9) | |
| NSS | 19 | (16.2) | 60 | (12.4) | |
| Acetar | 2 | (1.7) | 3 | (0.6) | |
| Immobilization | 82 | (70.1) | 295 | (61.1) | 0.070 |
| prehospital CPR | 14 | (12.0) | 12 | (2.5) | <0.001 |
Data are presented as numbers (%), means ± standard deviations, or medians (interquartile ranges).
Abbreviations: LRS, lactated Ringer’s solution: NSS, normal saline solution; min, minute; km, kilometer; IQR, interquartile range; cardiopulmonary resuscitation; CPR.
Univariable logistic regression identified several covariates associated with MBT, including abnormal vital signs, injury severity, specific injury types, and prehospital interventions (Table 3). In multivariable analysis, independent predictors of MBT were higher ISS, wound type, serious injury to the external body surface, distance from base to scene ≥ 4 km, shorter response and transfer times, airway management, and fluid access (Table 3).
Table 3.
Univariable and multivariable analyses of prehospital factors for predicting massive blood transfusion in trauma patients.
| Variables |
Univariable analysis |
Multivariable analysis |
||||||
|---|---|---|---|---|---|---|---|---|
| Coef. | OR (95 %CI)† | P-value | Coef. | ORadj (95 %CI)‡ | P-value | |||
| Age (years) | −0.003 | 1.00 | (0.99–1.01) | 0.493 | ||||
| <60 | 0.124 | 1.13 | (0.72–1.79) | 0.593 | ||||
| ≥60 | 1.00 | Reference | ||||||
| Sex | ||||||||
| Male | 0.155 | 1.17 | (0.76–1.8) | 0.486 | ||||
| Female | 1.00 | Reference | ||||||
| Mechanism of the injury | ||||||||
| Blunt | 0.270 | 1.31 | (0.80–2.15) | 0.284 | ||||
| Penetrating | 1.00 | Reference | ||||||
| Systolic blood pressure | ||||||||
| >90 mmHg | 1.00 | Reference | ||||||
| ≤90 mmHg | 1.377 | 3.96 | (2.24–7.02) | <0.001 | ||||
| Heart rate | ||||||||
| ≤120 bpm | 1.00 | Reference | ||||||
| >120 bpm | 0.255 | 1.29 | (0.76–2.21) | 0.351 | ||||
| Oxygen saturation | ||||||||
| >94 % | 1.00 | Reference | ||||||
| ≤94 % | 1.274 | 3.58 | (2.34–5.48) | <0.001 | ||||
| Glasgow coma score | −0.112 | 0.89 | (0.86–0.93) | <0.001 | ||||
| POCT glucose (mg%) | 0.003 | 1.00 | (1.00–1.01) | 0.092 | ||||
| Shock Index (SI) | 1.224 | 3.40 | (1.94–5.97) | <0.001 | ||||
| Injury severity score (ISS) | 0.159 | 1.17 | (1.14–1.21) | <0.001 | 0.137 | 1.15 | (1.10–1.19) | <0.001 |
| Type of wounds | ||||||||
| Abrasion | 1.00 | Reference | 1.00 | Reference | ||||
| Cut/laceration | 1.991 | 7.32 | (3.64–14.71) | <0.001 | 1.216 | 3.37 | (1.46–7.80) | 0.004 |
| Other | 1.576 | 4.84 | (2.21–10.61) | <0.001 | 1.482 | 4.40 | (1.71–11.33) | 0.002 |
| No | 1.926 | 6.86 | (2.98–15.83) | <0.001 | 1.822 | 6.18 | (2.31–16.55) | <0.001 |
| Bone injury | 1.009 | 2.74 | (1.81–4.17) | <0.001 | ||||
| Type of bone injury | ||||||||
| No | 1.00 | Reference | ||||||
| Close fracture | 0.738 | 2.09 | (1.25–3.50) | 0.005 | ||||
| Open fracture | 1.262 | 3.53 | (1.86–6.72) | <0.001 | ||||
| Dislocation | 1.590 | 4.91 | (2.03–11.84) | <0.001 | ||||
| Type of hemorrhage | ||||||||
| No | 1.00 | Reference | ||||||
| External stopped | 0.994 | 2.70 | (1.67–4.36) | <0.001 | ||||
| External active | 1.857 | 6.40 | (3.22–12.74) | <0.001 | ||||
| Internal hemorrhage | 2.070 | 7.93 | (3.23–19.47) | <0.001 | ||||
| Serious injury body part | ||||||||
| Abdomen | 1.673 | 5.33 | (2.72–10.44) | <0.001 | ||||
| Chest/Clavicle | 0.879 | 2.41 | (1.33–4.37) | 0.004 | ||||
| External Body Surface | 1.753 | 5.77 | (2.62–12.7) | <0.001 | 1.978 | 7.23 | (2.23–23.40) | 0.001 |
| Extremity | −0.203 | 0.82 | (0.55–1.22) | 0.326 | ||||
| Face | 0.481 | 1.62 | (1.01–2.59) | 0.045 | ||||
| Head/Neck | 0.695 | 2.00 | (1.33–3.02) | 0.001 | ||||
| Pelvis | 0.288 | 1.33 | (0.56–3.2) | 0.519 | ||||
| Spine | 0.320 | 1.38 | (0.69–2.73) | 0.359 | ||||
| Multiple injury | 1.205 | 3.34 | (1.53–7.26) | 0.002 | ||||
| Distance from the base station to the scene (km) | 0.009 | 1.01 | (0.96–1.07) | 0.754 | ||||
| <4 | 1.00 | Reference | 1.00 | Reference | ||||
| ≥4 | 0.373 | 1.45 | (0.97–2.18) | 0.071 | 0.935 | 2.55 | (1.34–4.86) | 0.005 |
| Distance from the scene to the hospital (km) | −0.027 | 0.97 | (0.93–1.02) | 0.302 | ||||
| <4 | 1.00 | Reference | ||||||
| ≥4 | 0.099 | 1.10 | (0.74–1.65) | 0.631 | ||||
| Response time (min) | −0.024 | 0.98 | (0.95–1.01) | 0.144 | ||||
| <8 | 0.531 | 1.70 | (1.13–2.56) | 0.011 | 0.959 | 2.61 | (1.43–4.75) | 0.002 |
| ≥8 | 1.00 | Reference | 1.00 | Reference | ||||
| Scene time (min) | 0.014 | 1.01 | (1.00–1.03) | 0.125 | ||||
| <15 | 1.00 | Reference | ||||||
| ≥15 | 0.046 | 1.05 | (0.70–1.57) | 0.823 | ||||
| Transfer to hospital time (min) | −0.026 | 0.97 | (0.95–1.00) | 0.088 | ||||
| <8 | 0.570 | 1.77 | (1.15–2.71) | 0.009 | 0.846 | 2.33 | (1.31–4.15) | 0.004 |
| ≥8 | 1.00 | Reference | 1.00 | Reference | ||||
| Hemorrhage control | 0.549 | 1.73 | (1.15–2.62) | 0.009 | ||||
| Airway management | 0.668 | 1.95 | (1.29–2.95) | 0.002 | 0.791 | 2.21 | (1.21–4.03) | 0.010 |
| Fluid access | 1.524 | 4.59 | (2.77–7.6) | <0.001 | 1.230 | 3.42 | (1.81–6.47) | <0.001 |
| Immobilization | 0.401 | 1.49 | (0.97–2.31) | 0.072 | ||||
| prehospital CPR | 1.674 | 5.34 | (2.4–11.87) | <0.001 | ||||
Abbreviations: Coef., regression coefficient;OR, odds ratio; ORadj, adjusted odds ratio; CI, confident interval; NA, data not applicable; min, minute; km, kilometer; cardiopulmonary resuscitation; CPR.Variables with p-values < 0.2 in the univariable analysis were included in the multivariable model. †Crude odds ratio estimated by the logistic regression model. ‡Adjusted odds ratio estimated by the logistic regression model. Model summary: −2 Log likelihood = − 385.92, Pseudo R2 = 0.350; Hosmer and Lemeshow test: chi-square = 9.37, df = 8, p = 0.312; constant = −6.728.
The Hosmer–Lemeshow test indicated a good model fit (χ2 = 9.37; df = 8; p = 0.312). No evidence of multicollinearity was observed, and the model had an R2 of 0.35, showing that the eight independent variables could explain a substantial proportion of the variance in predicting MBT. The detailed regression equation and variable coding are provided in the Supplementary Material.
Validation using ROC analysis showed that the regression equation predicted MBT with an AuROC of 0.894 (95 % CI: 0.863–0.925) (Fig. 2). To facilitate application, the regression coefficients were transformed into a risk score (Table 4). The mean risk score was 5.39 ± 2.82 (range 1–20.5), with MBT patients scoring significantly higher than non-MBT patients (8.59 ± 3.63 vs. 4.62 ± 1.91, p < 0.001). The risk score demonstrated good predictive ability, with an AuROC of 0.883 (95 % CI: 0.850–0.916) (Fig. 2).
Fig. 2.
Area under the received operating characteristic curve (AuROC) of the clinical prediction score for MBT.
Table 4.
Derivation of the scores from multivariable logistic regression analysis.
| Variables | Coef. | ORadj | (95 %CI) | P-value | Score |
|---|---|---|---|---|---|
| Injury severity score | 0.137 | 1.15 | (1.10 – 1.19) | <0.001 | 0.2/1 score (ISS) |
| Type of wounds | |||||
| Abrasion | 1.00 | Reference | 0 | ||
| Cut/laceration | 1.216 | 3.37 | (1.46–7.80) | 0.004 | 1.5 |
| Other | 1.482 | 4.40 | (1.71–11.33) | 0.002 | 2 |
| No | 1.822 | 6.18 | (2.31–16.55) | <0.001 | 2 |
| External body surface injury | 1.978 | 7.23 | (2.23–23.40) | 0.001 | 2.5 |
| Distance from the base station to scene ≥ 4 km | 0.935 | 2.55 | (1.34–4.86) | 0.005 | 1 |
| Response time < 8 min | 0.959 | 2.61 | (1.43–4.75) | 0.002 | 1 |
| Transfer to hospital time < 8 min | 0.846 | 2.33 | (1.31–4.15) | 0.004 | 1 |
| Non-airway management | 0.791 | 2.21 | (1.21–4.03) | 0.010 | 1 |
| Fluid access | 1.230 | 3.42 | (1.81–6.47) | <0.001 | 1.5 |
Abbreviations: Coef., regression coefficient;OR, odds ratio; ORadj, adjusted odds ratio; CI, confident interval; min, minute. Variables with p < 0.2 in the univariable analysis were included in the multivariable model. Model summary: −2 Log likelihood = −385.92, Pseudo R2 = 0.350; Hosmer and Lemeshow test: chi-square = 9.37, df = 8, p = 0.312; constant = −6.728.
The score calibration plot exhibited good agreement between the risk score and MBT in accident patients (Fig. 3a, Fig. 3b).
Fig. 3a.
Calibration plot between the score predicted probability of MBT and the observed probability of MBT.
Fig. 3b.
Internal validation of the MBT score using 500 bootstrap resamples.
The risk score stratified patients into low- (≤3), moderate- (4–9), and high-risk (≥10) groups, with MBT probabilities of 1.6 %, 20.7 %, and 74.2 %, respectively. LR + values were 0.07 for the low-risk group, 1.08 for the moderate-risk group, and 11.87 for the high-risk group, indicating strong predictive performance (Table 5, Fig. 4). Risk curve analysis showed that higher scores correlated with greater MBT probability, while decision curve analysis demonstrated a high net clinical benefit (Fig. 5).
Table 5.
Score categorization probability groups, likelihood ratio, and 95 % confidence interval of MBT.
| Score categorization probability groups | Score |
MBT(n = 117) |
Non-MBT(n = 483) |
LR+(95 %CI) | P-value | |||
|---|---|---|---|---|---|---|---|---|
| n | (%) | n | (%) | |||||
| Low risk | ≤3 | 2 | (1.6) | 123 | (98.4) | 0.07 | (0.02–0.27) | <0.001 |
| Intermediate risk | 4–9 | 92 | (20.7) | 352 | (79.3) | 1.08 | (0.97–1.20) | 0.240 |
| High risk | ≥10 | 23 | (74.2) | 8 | (25.8) | 11.87 | (5.45–25.86) | <0.001 |
| Mean ± SD | − | 8.59 | ± 3.63 | 4.62 | ± 1.91 | <0.001 | ||
Abbreviations: CI, confidence interval; LR, likelihood ratio; SD, standard deviations.
Fig. 4.
Risk curve analysis: observed risk of MBT (hollow circles) and predicted risk of MBT according to scores (solid line). The size of the circles represents the relative number of patients in each score.
Fig. 5.
Decision curve plotting the net benefit against the threshold probability.
Discussion
Incidence of MBT in the injured delivered by the EMS team
This study found that the incidence of MBT at the trauma center in the injured transported to the emergency department by the EMS team in Bangkok, Thailand, was 19.5 %, which was very high, compared with that observed in similar studies in the United States of America (12.3 %),14 China (9 %),15 Canada (2 %–4%),17 and Germany (approximately 3 %–4%).18
A possible explanation for this markedly high incidence of MBT was that most patients in this study had blunt trauma, which was experienced by 79.5 % of patients receiving MBT, which is consistent with the findings of a previous study reporting that blunt trauma was a risk factor for MBT19 and another study reporting that the injured with blunt trauma had a higher tendency to receive MBT than those with penetrating trauma, particularly blood vessel injuries.20
Another possible explanation was that the study area was a hospital, which is a trauma center, which has a more sensitive screening format and criteria in MBT activation, considering blood loss data at the scene through data communication and the rapid assessment from positive FAST results by the EMS team. These data affect physicians’ decisions in MBT more, particularly in the emergency department. The authors believed that the design of the rapid MBT activation system focused on blood transfusion in the early phase to reduce the mortality of the injured, which may be different from those observed in some countries that still wait for additional laboratory results or clinical signs before considering MBT. Furthermore, in the study area, there are still resources and readiness in rapid MBT. There is systematic blood component storage and massive transfusion protocol evaluation team preparation.
Prehospital potential predictors associated with MBT
Our study identified eight potential prehospital predictors associated with MBT.
(1) ISS. Previous evidence consistently shows that higher ISS is associated with increased MBT requirements. Rau CS et al. reported that patients receiving MBT had a median ISS of 26 compared with 13 in non-MBT patients (p < 0.001), and MBT was administered in 30.3 % of patients with ISS ≥ 25 compared with 2.7 % with ISS < 25.21 Similarly, Stanworth SJ et al. demonstrated that including ISS improved prediction model accuracy.22 These findings highlight ISS as a reflection of injury severity and blood loss, which necessitate rapid transfusion. We believe that in the EMS context, ISS can serve as a tool to predict MBT need and assist trauma teams and blood banks in early preparation.
(2) Wound type and (3) external body surface injury. Cuts/lacerations and external injuries were associated with higher odds of MBT. Previous studies also reported that large wounds, particularly in vital areas such as the chest or abdomen, were linked to MBT.23 Notably, in our study, “no wound type” was also predictive, possibly reflecting internal hemorrhage, consistent with prior findings that patients without visible external wounds often had severe internal bleeding requiring urgent MBT.5
(4) Distance from base station to the scene. Distances ≥ 4 km increased MBT risk, likely due to longer response times and greater blood loss at the scene. Evidence from the UK has similarly shown that longer EMS distances are associated with higher injury severity, mortality, and transfusion requirements.24 In our study area, V-EMS covers approximately 50 km2 in Zone 1 of Bangkok, representing a wide service area.
(5) Response time < 8 min. Patients with response times under 8 min had an MBT rate of 58.1 %. This may be related to Thai EMS standards, which use < 8 min as a national response indicator.25 A U.S. study found that shorter response times in penetrating trauma were linked with MBT need, while prolonged times increased mortality.26
(6) Transfer to hospital time < 8 min. This was a predictor of MBT, although previous evidence is lacking. We suggest this may reflect institutional regulations, including trauma fast track protocols, whereby severely injured patients are prioritized for rapid transfer to Vajira Hospital. In this study, 67.5 % of patients transferred within 8 min received MBT.
(7) Non-airway management. Our study found that patients without airway management had 2.21 times greater odds of receiving MBT compared with those who received airway intervention. This contrasts with Kotnarin et al., who reported airway management as a strong predictor of MBT.13 In our study, 36.8 % of MBT patients received oxygen via mask with bag, and 8.5 % required intubation. According to PHTLS® principles, airway management remains a cornerstone of trauma care to prevent hypoxia and secondary ischemic injury.27
(8) Fluid access. Patients with fluid access had 3.42 times greater odds of receiving MBT. This aligns with Larson N et al., who reported prehospital IV fluid therapy as a direct risk factor for MBT.28 However, other studies, including a retrospective cohort, reported no difference in MBT or mortality.29 A systematic review of 14 studies suggested that prehospital crystalloid resuscitation may increase transfusion requirements, while prehospital blood products may reduce mortality, though evidence quality remains low to moderate.30
Prehospital clinical prediction score for MBT
The risk score classified injured patients into three groups: low- (≤3), moderate- (4–9), and high-risk (≥10), with MBT rates of 1.6 %, 20.7 %, and 74.2 %, respectively. The LR + values were 0.07 (95 % CI: 0.02–0.27, p < 0.001) for the low-risk group, 1.08 (95 % CI: 0.97–1.20, p = 0.240) for the moderate-risk group, and 11.87 (95 % CI: 5.45–25.86, p < 0.001) for the high-risk group. These findings indicate that the tool has strong predictive performance.
This prehospital clinical prediction score was simple, convenient, and clinically applicable for injured patient management. The score achieved an AuROC of 0.894, which is considered excellent, and is therefore recommended in clinical epidemiology research.31, 32
Furthermore, the tool outperformed existing models developed in similar settings, such as the ABC score (AuROC = 0.842)8 and the ABC score with blood lactate (ABC + L; AuROC = 0.749).9 Although the ABC and ABC + L scores are easy to apply, they lack several EMS-relevant factors, such as response time, distance, prehospital interventions, and airway management. Incorporating these variables into the current score enhances both its efficiency and accuracy in predicting MBT in accident patients.
Strengths and limitations
The strengths of this study lie in the development and validation of a simple risk score for predicting MBT in prehospital trauma patients using fast and easily obtainable parameters. This tool is well-suited to the context of prehospital care and can assist paramedics in making timely decisions regarding patient transfer to appropriate trauma centers.
However, several important limitations should be acknowledged. First, this was a retrospective observational study conducted at a single EMS center (CLS, V-EMS). Therefore, the findings may not be generalizable to other EMS settings, and external validation or prospective studies are needed before applying the tool to real-world practice. Second, potential confounding factors exist, as most patients had blunt trauma (79.5 %), which may have rendered certain predictors, such as direct pressure or tourniquet application, insignificant despite their importance in other studies. Additionally, the study did not collect detailed injury mechanism data (e.g., car accidents, motorcycle crashes, or falls), which future research should incorporate. Third, all data were derived from retrospective EMS patient care reports, raising the possibility of selection bias despite efforts to minimize it.
Fourth, in our setting, the ISS scoring system is routinely applied to all trauma patients by the V-EMS team. While this allowed its inclusion in our model, the complexity of ISS may limit the generalizability and practicality of the score in other EMS contexts. Fifth, external or prospective validation in diverse EMS systems is necessary to confirm the robustness and clinical utility of the tool. Sixth, patients who died before receiving MBT were classified as non-MBT, which may have introduced bias due to competing risks, and this should be considered when interpreting the results. Seventh, continuous variables were dichotomized to improve interpretability and usability for EMS decision-making; however, this simplification may have reduced statistical power and precision. Eighth, the wide definition of MBT used in this study included both immediate transfusion needs within 1 h, which are highly relevant to EMS decision-making, and transfusions initiated up to 24 h after the EMS call, which may be less directly related to prehospital care. This broad outcome definition may have introduced heterogeneity in the clinical scenarios captured, and future studies should explore outcome measures more specifically aligned with EMS-relevant timeframes.
Finally, all injured patients in this study were transported exclusively to Vajira Hospital, although in real-world practice, over 35 % should have been transferred to other trauma centers depending on distance, patient preference, or health scheme. Thus, prospective studies incorporating these additional contextual factors are warranted.
Conclusion
In this study, we developed a prehospital clinical score for predicting MBT using routinely available EMS parameters. The score demonstrated good predictive performance with internal validation by bootstrap resampling. However, it has not yet undergone external or prospective validation. Therefore, the score should be regarded as a preliminary tool that requires further validation in diverse trauma populations and EMS systems before adoption into clinical practice.
Data availability
The datasets generated and/or analyzed during this study are available from the corresponding author upon reasonable request.
Funding sources
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
CRediT authorship contribution statement
Thongpitak Huabbangyang: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Gawin Tiyawat: Visualization, Conceptualization. Chantappapa Poking: Investigation, Conceptualization. Chattarat Chaimungkalayon: Investigation, Conceptualization. Natnicha Wiriyakitsoontorn: Investigation, Conceptualization. Thanrada Khottharin: Investigation, Conceptualization. Salintip Phomkunthong: Investigation, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
We also thank the V-EMS, Faculty of Medicine Vajira Hospital, Navamindradhiraj University, for facilitating data access and collection.
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.resplu.2025.101128.
Appendix A. Supplementary material
The following are the Supplementary data to this article:
References
- 1.Huabbangyang T., Klaiaungthong R., Jansanga D., et al. Survival rates and factors related to the survival of traffic accident patients transported by emergency medical services. Open Access Emerg Med. 2021;13:575–586. doi: 10.2147/OAEM.S344705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.World Health Organization. Global situation report on road safety 2015. 2015. [Internet]. [cited 2023 Nov 12]. Available from: https://www.afro.who.int/sites/default/files/2017-06/summary%20thailand.pdf.
- 3.Huabbangyang T., Sangketchon C., Ittiphisit S., Uoun K., Saumok C. Predictive factors of outcome in cases of out-ofhospital cardiac arrest due to traffic accident injuries in Thailand; a national database study. Arch Acad Emerg Med. 2022;10(1):e64. doi: 10.2147/OAEM.S344705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kauvar D.S., Wade C.E. The epidemiology and modern management of traumatic hemorrhage: US and international perspectives. Crit Care. 2005;9 Suppl 5(Suppl 5):S1–S9. doi: 10.1186/cc3779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Holcomb J.B., Tilley B.C., Baraniuk S., et al. Transfusion of plasma, platelets, and red blood cells in a 1:1:1 vs a 1:1:2 ratio and mortality in patients with severe trauma: the PROPPR randomized clinical trial. JAMA. 2015;313(5):471–482. doi: 10.1001/jama.2015.12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Jennings LK, Watson S. Massive Transfusion. StatPearls. Treasure Island (FL): StatPearls Publishing Copyright © 2024, StatPearls Publishing LLC.; 2024.
- 7.Abuzeid A.M., O'Keeffe T. Review of massive transfusion protocols in the injured, bleeding patient. Curr Opin Crit Care. 2019;25(6):661–667. doi: 10.1097/MCC.0000000000000668. [DOI] [PubMed] [Google Scholar]
- 8.Nunez T.C., Voskresensky I.V., Dossett L.A., Shinall R., Dutton W.D., Cotton B.A. Early prediction of massive transfusion in trauma: simple as ABC (assessment of blood consumption)? J Trauma. 2009;66(2):346–352. doi: 10.1097/TA.0b013e3181961c35. [DOI] [PubMed] [Google Scholar]
- 9.Chaochankit W., Akaraborworn O., Sangthong B., Thongkhao K. Combination of blood lactate level with assessment of blood consumption (ABC) scoring system: a more accurate predictor of massive transfusion requirement. Chin J Traumatol. 2018;21(2):96–99. doi: 10.1016/j.cjtee.2017.12.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Saviano A., Perotti C., Zanza C., et al. Blood transfusion for major trauma in emergency department. Diagnostics (Basel) 2024;14(7) doi: 10.3390/diagnostics14070708. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Collins G.S., Reitsma J.B., Altman D.G., Moons K.G. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ. 2015;350 doi: 10.1136/bmj.g7594. [DOI] [PubMed] [Google Scholar]
- 12.Brockamp T., Nienaber U., Mutschler M., et al. Predicting on-going hemorrhage and transfusion requirement after severe trauma: a validation of six scoring systems and algorithms on the TraumaRegister DGU. Crit Care. 2012;16(4):R129. doi: 10.1186/cc11432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Peduzzi P., Concato J., Kemper E., Holford T.R., Feinstein A.R. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996;49(12):1373–1379. doi: 10.1016/s0895-4356(96)00236-3. [DOI] [PubMed] [Google Scholar]
- 14.Kotnarin R. Development of prehospital factors for predicting massive blood transfusion in trauma patients. J Med Assoc Thai. 2023;106:771–777. doi: 10.35755/jmedassocthai.2023.08.13874. [DOI] [Google Scholar]
- 15.Kovar A., Carmichael H., McIntyre R.C., Jr., et al. The Extremity/Mechanism/Shock Index/GCS (EMS-G) score: a novel pre-hospital scoring system for early and appropriate MTP activation. Am J Surg. 2019;218(6):1195–1200. doi: 10.1016/j.amjsurg.2019.08.019. [DOI] [PubMed] [Google Scholar]
- 16.Li R., Han W., Lu J., Sun X., Tang T. The predictive value of four traumatic hemorrhage scores for early massive blood transfusion in trauma patients in the pre-hospital setting. Eur J Trauma Emerg Surg. 2024;50(3):967–973. doi: 10.1007/s00068-023-02412-8. [DOI] [PubMed] [Google Scholar]
- 17.Torres C.M., Kent A., Scantling D., Joseph B., Haut E.R., Sakran J.V. Association of whole blood with survival among patients presenting with severe hemorrhage in US and Canadian adult civilian trauma centers. JAMA Surg. 2023;158(5):532–540. doi: 10.1001/jamasurg.2022.6978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Jänig C., Willms C., Schwietring J., et al. Patients at risk for transfusion-a six-year multicentre analysis of more than 320,000 helicopter emergency medical service missions. J Clin Med. 2023;12(23) doi: 10.3390/jcm12237310. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Yeates E.O., Grigorian A., Inaba K., et al. Blunt trauma massive transfusion (B-MaT) score: a novel scoring tool. J Surg Res. 2022;270:321–326. doi: 10.1016/j.jss.2021.09.034. [DOI] [PubMed] [Google Scholar]
- 20.Kojima M., Endo A., Shiraishi A., Shoko T., Otomo Y., Coimbra R. Association between the plasma-to-red blood cell ratio and survival in geriatric and non-geriatric trauma patients undergoing massive transfusion: a retrospective cohort study. J Intensive Care. 2022;10(1):2. doi: 10.1186/s40560-022-00595-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Rau C.S., Wu S.C., Kuo S.C., 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(7) doi: 10.3390/ijerph13070683. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Stanworth S.J., Morris T.P., Gaarder C., et al. Reappraising the concept of massive transfusion in trauma. Crit Care. 2010;14(6):R239. doi: 10.1186/cc9394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Moore E.E., Moore H.B., Kornblith L.Z., et al. Trauma-induced coagulopathy. Nat Rev Dis Primers. 2021;7(1):30. doi: 10.1038/s41572-021-00264-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ong M.E., Chiam T.F., Ng F.S., et al. Reducing ambulance response times using geospatial-time analysis of ambulance deployment. Acad Emerg Med. 2010;17(9):951–957. doi: 10.1111/j.1553-2712.2010.00860.x. [DOI] [PubMed] [Google Scholar]
- 25.Huabbangyang T., Sangketchon C., Piewthamai K., et al. Perception and satisfaction of patients' relatives regarding emergency medical service response times: a cross-sectional study. Open Access Emerg Med. 2022;14:155–163. doi: 10.2147/OAEM.S360114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Nasser A.A.H., Nederpelt C., El Hechi M., et al. Every minute counts: the impact of pre-hospital response time and scene time on mortality of penetrating trauma patients. Am J Surg. 2020;220(1):240–244. doi: 10.1016/j.amjsurg.2019.11.018. [DOI] [PubMed] [Google Scholar]
- 27.Hussmann B., Lendemans S. Pre-hospital and early in-hospital management of severe injuries: changes and trends. Injury. 2014;45(Suppl 3):S39–S42. doi: 10.1016/j.injury.2014.08.016. [DOI] [PubMed] [Google Scholar]
- 28.Larson N., Yarbrough C., Khang P.N., Rogers F., Blondeau B., Dries D. Fluid resuscitation in the prehospital setting. Intl J Paramedicine. 2025;9:43–62. doi: 10.56068/HBEA1061. [DOI] [Google Scholar]
- 29.Bores S.A., Pajerowski W., Carr B.G., et al. The association of prehospital intravenous fluids and mortality in patients with penetrating trauma. J Emerg Med. 2018;54(4):487–499.e6. doi: 10.1016/j.jemermed.2017.12.046. [DOI] [PubMed] [Google Scholar]
- 30.Gianola S., Castellini G., Biffi A., et al. Volume replacement in the resuscitation of trauma patients with acute hemorrhage: an umbrella review. Int J Emerg Med. 2023;16(1):87. doi: 10.1186/s12245-023-00563-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Hosmer D.W., Lemeshow S., Sturdivant R.X. 3rd ed. Wiley; Hoboken (NJ): 2013. Applied Logistic Regression; pp. 177–180. [Google Scholar]
- 32.Hajian-Tilaki K. Receiver Operating Characteristic (ROC) curve analysis for medical diagnostic test evaluation. Caspian J Intern Med. 2013;4(2):627–635. [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated and/or analyzed during this study are available from the corresponding author upon reasonable request.