Secondary Logo

Journal Logo

Fibrinogen on Admission in Trauma score

Early prediction of low plasma fibrinogen concentrations in trauma patients

Gauss, Tobias; Campion, Sébastien; Kerever, Sébastien; Eurin, Mathilde; Raux, Mathieu; Harrois, Anatole; Paugam-Burtz, Catherine; Hamada, Sophie the Traumabase Group

European Journal of Anaesthesiology (EJA): January 2018 - Volume 35 - Issue 1 - p 25–32
doi: 10.1097/EJA.0000000000000734

BACKGROUND Early recognition of low fibrinogen concentrations in trauma patients is crucial for timely haemostatic treatment and laboratory testing is too slow to inform decision-making.

OBJECTIVE To develop a simple clinical tool to predict low fibrinogen concentrations in trauma patients on arrival.

DESIGN Retrospective cohort study.

SETTING Three designated level 1 trauma centres in the Paris Region, from January 2011 to December 2013.

PATIENTS Patients admitted in accordance with national triage guidelines for major trauma and plasma fibrinogen concentration testing on admission.

INTERVENTION Construction of a clinical score [Fibrinogen on Admission in Trauma (FibAT) score] in a derivation cohort to predict fibrinogen plasma concentration 1.5 g l−1 or less after multiple regressions. One point was given for each predictive factor. The score was the sum of all. Validation was performed in a separate validation cohort.

MAIN OUTCOME MEASURE Predictive accuracy of FibAT score.

RESULTS In total, 2936 patients were included, 2124 in the derivation cohort and 812 in the validation cohort. In the derivation cohort, a multivariate logistic model identified the following predictive factors for plasma fibrinogen concentrations 1.5 g l−1 or less: age less than 33 years, prehospital heart rate more than 100 beats per minute, prehospital SBP less than 100 mmHg, blood lactate concentration on admission more than 2.5 mmol l−1, free intraabdominal fluid on sonography, decrease in haemoglobin concentration from prehospital to admission of more than 2 g dl−1, capillary haemoglobin concentration on admission less than 12 g dl−1 and temperature on admission less than 36°C. The FibAT score had an area under the receiver operating characteristic curve of 0.87 [95% confidence interval (0.86 to 0.91)] in the derivation cohort and of 0.82 (95% confidence interval (0.86 to 0.91)] in the validation cohort to predict a low plasma fibrinogen.

CONCLUSION The FibAT score accurately predicts plasma fibrinogen levels 1.5 g l−1 or less on admission in trauma patients. This easy-to-use score could allow early, goal-directed therapy to trauma patients.

From the Hôpital Beaujon, Assistance Publique Hôpitaux de Paris, Hôpitaux Universitaires Paris Nord Val de Seine, Clichy(TG, SC, ME, CP-B), Service d’Anesthésie- Réanimation, Hôpital Lariboisière, Assistance Publique Hôpitaux de Paris (SK), Université Denis Diderot – Paris VII (SK, CP-B), Département d’Anesthésie Réanimation, Groupe Hospitalier Pitié–Salpêtrière Charles Foix, Assistance Publique Hôpitaux de Paris(MR), Sorbonne Universités, UPMC Université Paris 06, UMRS INSERM 1158, Paris (SC, MR) and Service d’Anesthésie-Réanimation, Hôpital Bicêtre, Groupement Hôpitaux Universitaires Paris Sud, Assistance Publique Hôpitaux de Paris, Le Kremlin Bicêtre, France (AH, SH)

Correspondence to Tobias Gauss, MD, Service Anesthésie et Réanimation, Hôpital Beaujon, Assistance Publique Hôpitaux de Paris, Hôpitaux Universitaire Paris Nord Val de Seine, 100 Bd du Général Leclerc, 92110 Clichy, France Tel: +33 0 1 40 87 52 33; e-mail:

Published online 8 November 2017

Back to Top | Article Outline


Over the last decade, massive transfusion protocols with fixed high plasma to red blood cell (RBC) ratios have become standard in trauma care. However, one recent randomised controlled trial did not demonstrate superiority of this strategy compared with conventional therapy for 30-day mortality.1 In contrast evidence is emerging favouring targeted, goal-directed coagulation management over protocols and fixed ratio-based strategies.2 One central and promising element of targeted and goal-directed management of trauma coagulation disorders is fibrinogen substitution. Low fibrinogen is associated with reduced clot strength and poor outcomes in trauma patients.3,4

The early identification of trauma patients with low plasma fibrinogen concentrations to enable fibrinogen substitution therapy may reduce haemorrhage and mortality, and has become standard practice in many centres. Moreover, European guidelines on management of bleeding and coagulopathy following major trauma recommend starting fibrinogen substitution when plasma fibrinogen concentrations are below 1.5 to 2 g l−1.5 However, laboratory testing is time consuming. Viscoelastic diagnostics [Thromboelastography (TEG) and thromboelastometry] are not universally available especially in resource limited or conflict settings. We hypothesised that a simple score, derived from data immediately available upon hospital admission, could be used to rapidly identify trauma patients with low plasma fibrinogen concentrations who should have fibrinogen substitution according to the European guidelines to initiate targeted procoagulant therapy as quickly as possible.

The objective of the study was to develop an easy-to-use clinical prediction score [called the Fibrinogen on Admission in Trauma (FibAT) score] to identify trauma patients with a plasma fibrinogen concentration of 1.5 g l−1 or less upon admission.

Back to Top | Article Outline


Study design and population

This study is a retrospective analysis of data derived a from a regional Trauma registry (the TraumaBase). Data for this registry are collected prospectively. This database includes trauma patients admitted to all three level 1 trauma centres in the Paris region: Beaujon University hospital, Bicêtre University hospital and Pitié-Salpêtrière University hospital. The database captures 60% of all trauma admissions from the Paris region of 12 million inhabitants, and collects extensive demographic, clinical, radiological, biological and therapeutic data as well as ICU progress and outcomes. Given the retrospective nature of the study and according to French law, the TraumaBase group obtained approval for this study, including use of data retrospectively, from the Institutional Review Board (Comité de Protection des Personnes, Paris VI, Pitié, president Pr Laurent Lacapelle, Bâtiment de la Force, 47 Boulevard de l”Hôpital, 75651 Paris Cedex 13, 28 November 2012) which is equivalent to an Ethical Committee and from the Advisory Committee for Information Processing in Health Research (Comité consultatif sur le traitement de l’information en matière de recherche dans le domaine de la santé, authorisation 11.305bis) and from the National Commission for Data Protection (Commission Nationale de l’Informatique et des Libertés, authorisation 911461).

All trauma patients triaged according to French out-of-hospital trauma algorithms6 and admitted to one of the three trauma centres above between January 2011 and November 2013 were evaluated for study inclusion. Exclusion criteria were: age less than 15 years, pregnancy, congenital hypofibrinogenaemia, transfer from a primary care hospital and prehospital blood transfusion.

Back to Top | Article Outline

Data collection and selection of variables

Potential predictors of plasma fibrinogen concentrations 1.5 g l−1 or less were selected according to investigators’ consensus and the concept of a simple, easy to calculate score from variables available within 5 min of admission. Selection was refined by reviewing mechanisms and factors that could influence fibrinogen concentrations.7

The following variables were integrated into the predictive model: age, sex, mechanism of injury (penetrating or blunt), highest out of hospital heart rate (HR), lowest out of hospital arterial SBP, out of hospital capillary haemoglobin concentration (Hemocue, Radiometer Medical ApS, Copenhagen, Denmark) and hospital admission capillary haemoglobin concentration (Hemocue, Radiometer Medical ApS), change of capillary haemoglobin concentration between out of hospital and hospital admission, admission body temperature, intraabdominal fluid assessed by sonography on admission (focussed assessment sonography in trauma), suspicion of haemothorax on an admission chest X-ray, pelvic fracture on X-ray, and blood lactate concentration on admission. Continuous variables were converted to categorical variables using clinically relevant cut-offs. Of note, capillary haemoglobin concentration measurement is standard prehospital trauma care in France. To assure standardisation between the measurements, the same method was continued upon admission to hospital.

Other data extracted from the TraumaBase included population characteristics, Injury Severity Score, Simplified Acute Physiology Score II score, Sepsis-related Organ Failure Assessment score score, mortality, coagulation test results, massively transfused patients (defined as a transfusion of four or more RBC units within 6 h or more than 10 RBC units within 24 h of admission) and blood product requirements among transfused patients (RBC units, fresh frozen plasma units, apheresis platelet units, fibrinogen concentrate).

Back to Top | Article Outline

Statistical analyses and score development

The study outcome was defined as a fibrinogen blood concentration 1.5 g l−1 or less (according to the Clauss method) on admission to a trauma centre.

The prevalence of plasma fibrinogen concentrations 1.5 g l−1 or less in our trauma population was 15.6% (n = 458). Risk factors for hypofibrinogenaemia were identified from the data of two centres that were randomly chosen to serve as the derivation cohort (centres I and II). Patients from the third centre constituted the validation cohort (centre III) for external temporal validation. The proportion of patients in the derivation and validation cohort was 72% and 28%, respectively – approximately following the two-thirds to one-third rule. The rule of 10 events per variable to qualify for logistic regression analysis was respected; 320 events were recorded into the derivation cohort, allowing 32 predictive variables for logistic regression.

To compare patients with and without hypofibrinogaenemia, all categorical variables were analysed using the χ2-square test or the Fisher's exact test, as appropriate, for association with outcome. Univariate odds ratios and 95% confidence intervals (95% CI) were also estimated. After assessment of possible colinearity between categorical variables, the logistic regression model was constructed using backward stepwise selection with hypofibrinogenaemia as the dependent variable. Independent and clinically relevant predictors were entered into the model if a significant association (P < 0.05) was identified on univariate analysis. The score was then created using the predictive factors of low plasma fibrinogen identified after multivariate analysis. It was decided to attribute one point to each predictive factor. The score was the sum of all criteria presented.

To assess external validity, the score was evaluated in the validation cohort and logistic regression model calibration was assessed using the Hosmer–Lemeshow goodness-of-fit statistic (P = 0.6). Sensitivity, specificity, positive and negative predictive values and accuracy (assessed by the proportion of correctly classified patients) were calculated at each cut-off point of the FibAT score.

Missing values never exceeded 3% for any candidate variables. Variables used in the multivariate analysis were imputed in 2% of cases using the mean of the variable. All tests were two sided and a significance threshold of 0.05 was employed for all tests. The results are expressed as median and [interquartile range], or numbers and percentages. Analyses were performed using R open source software 3.1.1 (available online at

Back to Top | Article Outline


Population characteristics

Between January 2011 and December 2013, 3887 patients were admitted in the three centres and screened for inclusion. In total, 2936 patients were included in the analysis: 2124 from centres I and II, assigned to the derivation cohort and 812 from centre III, assigned to the validation cohort (Fig. 1). Patient characteristics of the derivation and validation cohorts are summarised in Table 1.

Fig. 1

Fig. 1

Table 1

Table 1

Back to Top | Article Outline

Derivation cohort

Univariate analysis conducted on data from the derivation cohort between patients with plasma fibrinogen concentrations 1.5 g l−1 or less (n = 320) and plasma fibrinogen concentrations more than 1.5 g l−1 (n = 1804), identified the following predictive variables (Table 2): age less than 33 years, out of hospital HR more than 100 beats per minute, out of hospital SBP less than 100 mmHg, out of hospital capillary haemoglobin concentration less than 12 g dl−1, decrease of capillary haemoglobin concentration between out-of-hospital and hospital admission of more than 2 g dl−1, admission body temperature less than 36°C, free intraabdominal fluid on sonography, suspicion of haemothorax on chest X-ray, pelvic fracture, open fracture and an admission blood lactate concentration of more than 2.5 mmol l−1.

Table 2

Table 2

A multivariate logistic regression model identified the following independent predictive factors for plasma fibrinogen concentrations 1.5 g l−1 or less on admission (Table 3): age less than 33 years, out-of-hospital HR more than 100 beats per minute, out-of-hospital SBP less than 100 mmHg, capillary haemoglobin concentration on admission less than 12 g dl−1, DHb (Out hospital – admission) >2g dl-1,temperature less than 36°C, free intraabdominal fluid on sonography and blood lactate level on admission of more than 2.5 mmol l−1. One point was given for each predictive item: these were summed to provide the FibAT score, ranging from 0 to 8 points. The area under the curve for the derivation cohort was 0.87 [95% CI (0.86 to 0.91)].

Table 3

Table 3

Back to Top | Article Outline

Validation cohort

The area under the receiver operating characteristic curve was 0.82 [95% CI (0.86 to 0.91)] (Fig. 2). Determination of the FibAT score predictive performance on the validation cohort showed that the most appropriate cut-off score to identify trauma patients with plasma fibrinogen concentrations 1.5 g l−1 or less on admission was 5 points, with a specificity of 98%, a negative predictive value of 87% and an accuracy of 86% (Table 4). Using a 5-point score cut-off, the positive likelihood ratio was 23 and negative likelihood ratio was 0.55. This is a favourable result given the overall incidence of plasma fibrinogen concentrations of 1.5 g l−1 or less at just 17% in the validation cohort.

Fig. 2

Fig. 2

Table 4

Table 4

Back to Top | Article Outline


The FibAT score allows prediction of plasma fibrinogen concentrations of 1.5 g l−1 or less on admission in trauma patients. The score performed well: a FibAT score of 5 points or more indicates a high probability of a fibrinogen concentration below 1.5 g dl−1; when the FibAT score is less than 5 points, the probability of hypofibrinogenaemia is low. This study uses the current European trauma guideline recommendations as its starting point and should not be considered as evaluation of the therapeutic benefit of fibrinogen replacement.

Trauma-associated coagulopathy occurs very early following injury. Floccard et al.8 showed that 46% of trauma patients had coagulopathy at the trauma scene and that 60% had abnormal coagulation status on admission. This strongly suggests that haemostatic resuscitation should be started as soon as possible, a rationale supported by the results from the Crash-2 trial.9 Despite the lack of randomised controlled trials examining early fibrinogen therapy in trauma haemorrhage, observational data following major trauma10 and one randomised trial after aortic surgery11 both suggest a strong benefit, and only large doses of fibrinogen in the form of cryoprecipitate or fibrinogen concentrate seem to correct trauma-associated coagulopathy.12 Thus, the European guidelines on bleeding management after major trauma recommend fibrinogen substitution in the form of concentrate or cryoprecipitate in uncontrolled bleeding in trauma patients whose plasma fibrinogen concentration is less than 1.5 to 2 g l−1.5 Evidence suggests that initiation of haemostatic treatment is still often delayed, especially fibrinogen substitution.13 The aim of the FibAT score is to provide a tool to identify those patients with a fibrinogen concentration below the recommended level of 1.5 g l−1 as early as possible, to avoid the delay inherent in classical laboratory testing. In this respect the FibAT score differs from established or more recently developed prediction tools14,15 which indicate the overall probability of massive transfusion or coagulopathy and, therefore, trigger a global ‘all in’ haemostatic therapy. The FibAT score might accelerate targeted blood component therapy based on simple clinical criteria.

Early recognition of patients with low plasma fibrinogen concentrations is desirable, as substitution therapy should be initiated as soon as possible. The standard Clauss method of laboratory testing takes 40 to 45 min on average. Viscoelastic methods are faster – maximal clot firmness (TEG) or amplitude (Rotational thromboelastometry) take around 15 min and the FIBTEM test at 5 min correlates well with the fibrinogen concentration once the test has been launched.16 In comparison all variables of the FibAT score are available to the trauma team within the first 10 min of arrival. Viscoelastic methods may not be available because of the lack of personnel, software, consumables or reagents. The FibAT score allows targeted fibrinogen substitution when viscoelastic testing is not feasible such as in nonspecialised centres, in mass casualty events or in resource-limited situations such as military and humanitarian operations.

Hyprofibrinogenaemia is a key part of trauma coagulopathy which explains an overlap between the FibAT score with existing scores which were designed to predict massive transfusion.14,15,17 However, not all patients with low fibrinogen concentrations will require massive transfusion (16% of patients had fibrinogen concentrations ≤ 1.5 g l−1, but only 6% of patients required a massive transfusion). Early goal-directed coagulopathy therapy based on early fibrinogen replacement may prevent a yet to be determined proportion of massive transfusions. Plasma is an insufficient substitute for fibrinogen and concentrate or cryoprecipitate is often needed to achieve useful concentrations.12 Furthermore, both plasma and cryoprecipitate require thawing, so the anticipation of requirements before laboratory test results are available is useful. Some patients present with low fibrinogen concentrations without a full clinical picture that may trigger massive transfusion protocols. The FibAT score may facilitate the early administration of fibrinogen under the conditions of the European guidelines. This potential impact requires exploration in a new prospective controlled study that integrates the FibAT score into a comprehensive early goal-directed and targeted trauma coagulation management bundle.

Our study has a number of limitations. First, 11% of data were removed from the initial register because of incomplete or duplicated data. However, this percentage is relatively low compared with other studies.14

Second, the derivation and validation cohorts differed even for items that were part of the score. This approach has been recommended recently18,19 to allow external temporal validation to an initial derivation cohort. In the authors’ opinions, the discrepancies actually emphasise the external validity of the score, as it performs well, even in an independent and significantly different cohort.20,21

Third, the FibAT score was derived from a civilian cohort with a predominance of blunt injuries (91%). This may limit its validity in trauma populations with higher incidences of penetrating trauma. However, the validation cohort had a higher incidence of penetrating trauma (19%) than the derivation cohort (5%) and the score still performed well.

Fourth, the FibAT score was developed using a French cohort. French emergency medical services (EMS) may differ from other prehospital care systems, as a medical doctor is part of the standard EMS team. This is also the case in several other European countries such as Germany, Norway, Switzerland, and Austria. Capillary haemoglobin concentration measurement may not be part of standard prehospital trauma patient assessment as it is in France, but easy to adopt. Reduction in capillary haemoglobin concentration values from prehospital to admission is one of the strongest criteria in the FibAT score, perhaps reinforcing the role of prehospital capillary haemoglobin concentration testing in trauma patients. It is perhaps worth repeating at this point that capillary haemoglobin concentrations are as easy to determine as blood glucose concentrations, despite their lack of ubiquity across EMS.

Fifth, some factors have not been included in the model. Examples include fluid resuscitation, the use of colloids and time from injury.22 These factors can all impact upon plasma fibrinogen concentrations but are operator and system dependent.23 The study group deliberately decided to omit fluid challenge as a variable. The estimation of the amount of fluid given in the resuscitation phase is notoriously imprecise and as such a potential source of error. Low amounts of fluid given may actually not influence coagulation factors. Finally fluid strategies vary between countries and systems. Inclusion into the model may have decreased its external validity.

Sixth, we decided not to weigh any individual factors so as to keep the score easy to use. As Table 3 demonstrates, all retained variables’ derived odds ratios were relatively closely spread, and weighting would have little effect on the model. Furthermore, experience shows that most scores are infrequently used, probably because they are too complicated to calculate.

Finally the therapeutic impact of the score was not evaluated. This aspect was beyond the scope of the study and requires a separate prospective study.

In conclusion, the FibAT score is an easy-to-use bedside score that predicts plasma fibrinogen concentrations of 1.5 g l−1 or less in trauma patients upon admission. Identification of hypofibrinogenaemia is important in trauma management. This easy to use score could allow early goal-directed therapy to trauma patients. Further prospective studies are required to validate this approach and evaluate the therapeutic impact.

Back to Top | Article Outline

Acknowledgements relating to this article

Assistance with the study: We would like to thank Beny Charbit, MD, PhD, for counselling at the launch of the project. TG has participated as an instructor for a serious game (educational game) on major trauma care for LFB (Laboratoire Français du Biomédicament) without financial compensation.

The TraumaBase Group: Jacques Duranteau, MD, PhD (Service d’Anesthésie-Réanimation, Hôpital Bicêtre, Groupement Hôpitaux Universitaires Paris Sud, AP-HP, Kremlin Bicêtre, France); Olivier Langeron, MD, PhD (Sorbonne Universités, UPMC Université Paris 06, and Département d’Anesthésie Réanimation, Groupe Hospitalier Pitié-Salpêtrière Charles Foix, AP-HP, Paris, France); Jean Mantz, MD, PhD (Université Denis Diderot - Paris VII, Département d’anesthésie-réanimation des Hôpitaux Universitaires Paris Nord Val de Seine); Bruno Riou, MD, PhD (Sorbonne Universités, UPMC Université Paris 06, UMRS INSERM 1166, IHU ICAN, and Service d’accueil des urgences, Groupe Hospitalier Pitié-Salpêtrière Charles Foix, AP-HP, Paris, France); Bernard Vigué, MD (Service d’Anesthésie-Réanimation, Hôpital Bicêtre, Groupement Hôpitaux Universitaires Paris Sud, AP-HP, Kremlin Bicêtre, France.

Financial support: The trauma registry TraumaBase receives funding from the Regional Health Authority (Agence Régionale de Santé, ARS).

Conflicts of interest: AH had a consultancy activity for LFB (Laboratoire Français du Biomédicament) in the last 18 months. He advised LFB for the script of a serious game (educational game) that deals with trauma patient care. CP-B has received payments for industry supplied lectures by LFB.

Presentation: Preliminary results of the study have been presented at the SFAR and ESICM meeting in September and October 2014 in Paris.

Back to Top | Article Outline


1. Holcomb JB, Tilley BC, Baraniuk S, et al. PROPPR Study Group. 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; 313:471–482.
2. Gonzalez E, Moore EE, Moore HB, et al. Goal-directed hemostatic resuscitation of trauma-induced coagulopathy: a pragmatic randomized clinical trial comparing a viscoelastic assay to conventional coagulation assays. Ann Surg 2016; 263:1051–1059.
3. Nystrup KB, Windeløv NA, Thomsen AB, et al. Reduced clot strength upon admission, evaluated by thrombelastography (TEG), in trauma patients is independently associated with increased 30-day mortality. Scand J Trauma Resusc Emerg Med 2011; 19:52.
4. Hagemo JS, Stanworth S, Juffermans NP, et al. Prevalence, predictors and outcome of hypofibrinogenaemia in trauma: a multicentre observational study. Crit Care 2014; 18:R52.
5. Rossaint R, Bouillon B, Cerny V, et al. The European guideline on management of major bleeding and coagulopathy following trauma: fourth edition. Crit Care 2016; 20:100.
6. Hamada SR, Gauss T, Duchateau F-X, et al. Evaluation of the performance of French physician-staffed emergency medical service in the triage of major trauma patients. J Trauma Acute Care Surg 2014; 76:1476–1483.
7. Dobson GP, Letson HL, Sharma R, et al. Mechanisms of early trauma-induced coagulopathy: the clot thickens or not? J Trauma Acute Care Surg 2015; 79:301–309.
8. Floccard B, Rugeri L, Faure A, et al. Early coagulopathy in trauma patients: an on-scene and hospital admission study. Injury 2012; 43:26–32.
9. Shakur H, Roberts I, Bautista R, et al. CRASH-2 trial collaborators. Effects of tranexamic acid on death, vascular occlusive events, and blood transfusion in trauma patients with significant haemorrhage (CRASH-2): a randomised, placebo-controlled trial. Lancet 2010; 376:23–32.
10. Inaba K, Karamanos E, Lustenberger T, et al. Impact of fibrinogen levels on outcomes after acute injury in patients requiring a massive transfusion. J Am Coll Surg 2013; 216:290–297.
11. Rahe-Meyer N, Solomon C, Hanke A, et al. Effects of fibrinogen concentrate as first-line therapy during major aortic replacement surgery: a randomized, placebo-controlled trial. Anesthesiology 2013; 118:40–50.
12. Khan S, Davenport R, Raza I, et al. Damage control resuscitation using blood component therapy in standard doses has a limited effect on coagulopathy during trauma hemorrhage. Intensive Care Med 2015; 41:239–247.
13. Stanworth SJ, Davenport R, Curry N, et al. Mortality from trauma haemorrhage and opportunities for improvement in transfusion practice: mortality from trauma haemorrhage. Br J Surg 2016; 103:357–365.
14. Brockamp T, Nienaber U, Mutschler M, et al. TraumaRegister DGU. 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:R129.
15. Peltan ID, Rowhani-Rahbar A, Vande Vusse LK, et al. Development and validation of a prehospital prediction model for acute traumatic coagulopathy. Crit Care 2016; 20:371.
16. Reed MJ, Nimmo AF, McGee D, et al. Rotational thrombolelastometry produces potentially clinical useful results within 10 min in bleeding emergency department patients: the DEUCE study. Eur J Emerg Med 2013; 20:160–166.
17. Cantle PM, Cotton BA. Prediction of massive transfusion in trauma. Crit Care Clin 2017; 33:71–84.
18. Moons KG, Altman DG, Reitsma JB, et al. Transparent Reporting of a Multivariate Prediction Model for Individual Prognosis or Development Initiative. New guideline for the reporting of studies developing, validating, or updating a multivariable clinical prediction model: the TRIPOD Statement. Adv Anat Pathol 2015; 22:303–305.
19. Steyerberg EW, Moons KG, van der Windt DA, et al. PROGRESS Group. Prognosis research strategy (PROGRESS) 3: prognostic model research. PLoS Med 2013; 10:e1001381.
20. Steyerberg EW, Pencina MJ, Lingsma HF, et al. Assessing the incremental value of diagnostic and prognostic markers: a review and illustration. Eur J Clin Invest 2012; 42:216–228.
21. Steyerberg EW, Bleeker SE, Moll HA, et al. Internal and external validation of predictive models: a simulation study of bias and precision in small samples. J Clin Epidemiol 2003; 56:441–447.
22. Hagemo JS, Christiaans SC, Stanworth SJ, et al. Detection of acute traumatic coagulopathy and massive transfusion requirements by means of rotational thromboelastometry: an international prospective validation study. Crit Care 2015; 19:97.
23. Cade JA, Truesdale M. Preferences of critical care registrars in fluid resuscitation of major trauma patients: concordance with current guidelines. Anaesth Intensive Care 2011; 39:262–267.
© 2018 European Society of Anaesthesiology