Definition and drivers of acute traumatic coagulopathy: clinical and experimental investigations


Karim Brohi, Trauma Clinical Academic Unit, The Royal London Hospital, Whitechapel Road, London E1 1BB, UK.
Tel.: +44 20 7377 7695; fax: +44 20 7377 7044.


Summary. Background: Acute traumatic coagulopathy (ATC) is an impairment of hemostasis that occurs early after injury and is associated with a 4-fold higher mortality, increased transfusion requirements and organ failure. Objectives: The purpose of the present study was to develop a clinically relevant definition of ATC and understand the etiology of this endogenous coagulopathy. Patients/methods: We conducted a retrospective cohort study of trauma patients admitted to five international trauma centers and corroborated our findings in a novel rat model of ATC. Coagulation status on emergency department arrival was correlated with trauma and shock severity, mortality and transfusion requirements. 3646 complete records were available for analysis. Results: Patients arriving with a prothrombin time ratio (PTr) > 1.2 had significantly higher mortality and transfusion requirements than patients with a normal PTr (mortality: 22.7% vs. 7.0%; < 0.001. Packed red blood cells: 3.5 vs. 1.2 units; < 0.001. Fresh frozen plasma: 2.1 vs. 0.8 units; < 0.001). The severity of ATC correlated strongly with the combined degree of injury and shock. The rat model controlled for exogenously induced coagulopathy and mirrored the clinical findings. Significant coagulopathy developed only in animals subjected to both trauma and hemorrhagic shock (PTr: 1.30. APTTr: 1.36; both < 0.001 compared with sham controls). Conclusions: ATC develops endogenously in response to a combination of tissue damage and shock. It is associated with increased mortality and transfusion requirements in a dose-dependent manner. When defined by standard clotting times, a PTr > 1.2 should be adopted as a clinically relevant definition of ATC.


Hemorrhage is responsible for 40% of all trauma deaths and is commonly associated with coagulopathy [1–3]. Acute traumatic coagulopathy (ATC) is an endogenous impairment of hemostasis that occurs early after injury [4]. The presence of ATC is associated with a 4-fold higher mortality, increased transfusion requirements and worse organ failure [5–8]. While there is now a significant body of evidence confirming the existence of ATC, there is no clinically relevant definition and its etiology remains obscure.

To date, identification of ATC has been based on traditional transfusion triggers recommended by generic massive transfusion guidelines [9–11]. Most commonly these are a 50% prolongation of the prothrombin time (PT) or partial thromboplastin time (PTT). However, this threshold is arbitrary and its clinical significance in terms of actual clinical outcomes is unknown. Patients with less severe ATC may also have worse outcomes and potentially benefit from therapeutic intervention. The reported prevalence of ATC varies widely and will depend in part on a robust definition.

Coagulopathy in the trauma patient is often multifactorial and partly induced by therapeutic intervention. However, ATC appears to have an endogenous component as a result of combined shock and tissue damage, and can develop in the absence of exogenous factors such as hemodilution or hypothermia [4,5,7,8,12]. Injury severity is positively associated with the development of ATC [5,7] and hemorrhagic shock has been implicated as an additional etiological factor [8,12]. These two variables frequently co-exist but the relative or synergistic contributions of each are unclear. Understanding the etiology may improve our understanding of the mechanisms of ATC.

The overall purpose of the present study was to develop a clinically relevant definition of ATC and understand the relationship between trauma and shock in its etiology. The first aim was to determine the relationship between admission coagulopathy and outcomes in terms of mortality and transfusion requirements. The second aim was to determine the independent and synergistic effects of injury load and tissue hypoperfusion in the development of ATC. Third, we wished to replicate these findings in an experimental model of trauma hemorrhage, to determine whether ATC can develop through a purely endogenous mechanism. We conducted an international retrospective multicenter cohort study and validated our findings in a new rat model of ATC.

Material and methods

Retrospective multicenter cohort study

With local ethical approvals, five established trauma registries [The Royal London Hospital, The Trauma Registry of the Deutsche Gesellschaft für Unfallchirurgie/German Trauma Society (TR-DGU), Academic Medical Center Amsterdam, Ullevål University Hospital Oslo and San Francisco General Hospital] were retrospectively reviewed for trauma patients admitted between January and December 2007. Datasets included all patients meeting local criteria for trauma team activation, with London excluding patients discharged from the emergency department and the TR-DGU excluding patients with an injury severity score (ISS) ≤ 9. Where information was available, patients taking oral anticoagulants, with pre-existing liver disease or who had arrived as a secondary transfer were excluded from the analysis.

Data were collected on patient demographics, mechanism of injury, transfer to hospital time, ISS, admission base deficit (BD), admission PT or Quick Values (QV) or international normalized ratio (INR), packed red blood cell (PRBC) and fresh frozen plasma (FFP) units received in the first 24 h of admission, and hospital mortality.

In order to compare PT results across centers using different assay methods and reagents a prothrombin time ratio (PTr) was calculated for each patient. Automated coagulation analyzers were in use at all centres with reagents demonstrating similar sensitivities [International Sensitivity Index (ISI): 1.03–1.09]. INR was converted to PT by raising it to the power of 1/ISI for the specific thromboplastin used, then multiplying this by the mean normal PT for that batch of reagents. QV was converted to PT by dividing 100 by the QV and multiplying by the mean normal PT for the reagent. PTr was calculated by dividing PT by the appropriate mean normal PT.

Rodent model of ATC

Forty male Wistar rats (Charles River Ltd, Margate, UK) weighing 230 g to 360 g were randomly assigned to either sham control (S), trauma (T), hemorrhagic shock (H) or trauma and hemorrhagic shock groups (TH). They received a standard diet and water ad libitum, and were cared for in accordance with the UK Home Office Guidance in the Operation of the Animals (Scientific Procedures) Act 1986. General anesthesia was induced and maintained with intra-peritoneal sodium thiopentone (Intraval Sodium, 120 mg kg−1) and body temperature was maintained at 37 ± 1 °C. A tracheotomy was performed and a small section of polyethylene tube (Internal Diameter 1.67 mm; Portex, Hythe, UK) inserted to aid respiration. The left carotid artery was catheterized with polyethylene tubing (Internal Diameter 0.58 mm; Portex) connected to a pressure transducer (Capto SP 844; AD instruments, Chalgrove, UK) and a syringe pump (PHD 22/2000, Harvard Apparatus, Tonbridge, UK). Sodium chloride (0.9%) was continuously infused at a rate of 200 μL h−1 with supplementary flushes in order to maintain patency of the carotid catheter.

Sham control animals were anesthetized, tracheotomized and monitored. Rats in trauma groups received a 6-cm paramedian laparotomy, closed in one layer with 7-mm surgical clips (Harvard Apparatus) and bilateral mid-shaft closed tibia and fibula fractures. The experimental period commenced immediately after traumatic injury or 7 min after completion of carotid catheterization (the mean time taken to complete traumatization).

Animals in hemorrhagic shock groups were bled from an average mean arterial pressure of 125 mmHg (MAP) to a target of 40–50 mmHg over a period of 10 min and this pressure was maintained by further withdrawals as necessary.

After a 70-min experimental period, 0.5 mL of blood was aspirated from the carotid catheter and discarded. Next, 100 μL was then aspirated into a heparin-lined syringe (Radiometer, Crawley, UK) for arterial blood gas analysis. A further 2 mL was aspirated into a 5-mL syringe pre-filled with 200 μL of 3.2% sodium citrate (Sigma-Aldrich, Gillingham, UK). Lactate measurement (Accutrend, Roche, Welwyn Garden City, UK) was performed with 20 μL of this sample and the remainder was centrifuged at 3500 ×g for 15 min at room temperature. Plasma supernatant was immediately snap frozen in liquid nitrogen and stored at −80 °C. All rats (10 per group) survived to the end of the experimental period and were included for data analysis. Animals were subsequently euthanized by excision of their heart.

Rat coagulation analysis was performed on a single channel thrombotrack coagulometer (Axis-Shield, Kimbolton, UK) using modified sample volumes. PT was measured by incubating 100 μL of Neoplastine CI reagent (Stago, Reading, UK) for 3 min followed by the addition of 50 μL of thawed plasma. APTT was performed by incubating 50 μL of Automate PTT reagent (Stago) with 50 μL of thawed plasma for 3 min followed by addition of 50 μL of 0.025 mm calcium chloride.

Statistical analysis

Data analysis was performed using Microsoft Excel software and GraphPad Prism 5 statistical package. Parametric data were expressed as mean ± standard error with multigroup analysis using one-way anova and Bonferroni’s multiple comparison tests. Non-parametric data are expressed as median and interquartile range (IQR) with two-group analysis performed using Fisher’s exact test or χ2, as appropriate. Non-parametric multigroup analysis was performed using Kruskal–Wallis and Dunn’s multiple comparison tests.


During 2007, 5693 trauma patients were admitted to centers participating in this study. Three thousand six hundred and forty-six patients had data sets complete for ISS, admission BD and PT/INR/QV. Demographics, injury mechanism and severity, initial base deficit and coagulation screen results are shown in Table 1.

Table 1.   Clinical characteristics of trauma patients
 All CentresLondonAmsterdamOsloSan FranciscoTR-DGU
  1. ED, emergency department; ISS, injury severity score; BD, base deficit; PTr, prothrombin time ratio; NA, not available. Values are number (%) or median (inter-quartile range).

Number of patients36466284776172361688
Age39 (25–54)33 (24–39)37 (24–52)38 (22–55)38 (25–54)41 (27–58)
Male (%)2716 (75%)499 (79%)347 (73%)460 (75%)185 (78%)1225 (73%)
Time to ED60 (42–83)62 (48–81)N/A62 (32–103)27 (21–34)63 (48–35)
ISS22 (10–33)17 (9–29)9 (4–16.5)14 (5–26)21 (13–30)27 (13–38)
ISS > 152425 (67%)356 (57%)146 (31%)300 (49%)168 (71%)1455 (86%)
Blunt injury3276 (90%)510 (81%)448 (94%)566 (92%)163 (69%)1589 (94%)
BD (mmol L−1)2.7 (0.6–5.8)2.7 (0.6–6.2)1.3 (−04–3.4)14 (−0.6–3.7)5.6 (3.1–9.3)3.4 (1.2–6.3)
BD > 6 (mmol L−1)871 (24%)170 (27%)47 (10%)81 (13%)111 (47%)462 (27%)
PTr1.1 (1.1–1.4)1.0 (1.0–1.1)1.1 (1.1–1.2)1.1 (1.l–1.3)1.0 (1.0–1.1)1.3 (1.1–1.8)
PTr > 1.21323 (36%)84 (13%)86 (13%)174 (23%)31 (13%)948 (56%)
PTr > 1.5700 (19%)33 (5%)21 (4%)57 (9%)9 (0%)580 (34%)

Relationship between admission coagulopathy and outcomes

Figure 1A displays the trend of mortality rates by admission PTr. Patients with a PTr of 1.2 or less had a mortality equivalent to those with a normal PTr (7.0%). There was a significant increase in mortality with PTr > 1.2. Mortality for trauma patients admitted with a PTr of 1.3 was almost twice that of patients with PTr ≤ 1.2 (12.2% vs. 7.0%; < 0.001). Overall, patients with a PTr > 1.2 had a mortality of 22.7% (< 0.001 vs. PTr ≤ 1.2). Patients with a PTr > 1.5 had a mortality of 27.8% (< 0.001, vs. PTr ≤ 1.2), four times greater than normal and consistent with previous studies.

Figure 1.

 Relationships between acute traumatic coagulopathy (ATC) and clinical outcomes. (A) Increasing mortality with increasing prolongations of the prothrombin time (PT). *< 0.001 compared with prothrombin time ratio (PTr) = 1. (B) Increasing 24-h administration of transfusion products with increasing prolongations of the PT. *< 0.001 compared with PTr = 1. +< 0.001 compared with PTr = 1. (C) The prevalence of prothrombin ratios in the emergency department.

A similar pattern was seen with blood product requirements (Fig. 1B). 38.2% of patients received at least one unit of PRBC and 12.4% received a massive transfusion (≥ 10 PRBCs) in the first 24 h of their admission. 30.3% of patients received at least one unit of FFP. Patients with a PTr > 1.2 were significantly more likely to receive blood products (PRBC 64% vs. 21%, < 0.001; FFP 57% vs. 11.7%, < 0.001). There was no difference in volume of blood products required with a PTr of 1.2 or less. Patients arriving with a PTr > 1.2 received significantly more PRBCs and FFP than patients arriving with a normal PT (PRBC: 3.5 vs. 1.2 units; < 0.001. FFP: 2.1 vs. 0.8 units; < 0.001).

Figure 1C shows the prevalence of abnormal PT ratios in the study cohort. 36.2% had a PTr > 1.2 compared with 19.2% with a PTr > 1.5. The classic definition of coagulopathy as a PT prolonged over 1.5-times normal would miss 17% of this trauma cohort with a PT ratio of 1.3–1.5, who have significantly increased transfusion requirements (four times more units of PRBCs and FFP) and 3-fold higher mortality than patients with a normal PTr.

Human responses to shock and trauma

Consistent with previous studies, the PTr showed a strong correlation with both injury severity (ISS: r = 0.38, < 0.001) and hypoperfusion (BD: r = 0.31, < 0.001). Figure 2A shows a multivariate plot of PTr against both ISS and BD. In the absence of hypoperfusion the median admission PTr did not exceeded 1.2, regardless of the injury severity. Similarly, without a significant injury load (ISS > 16) the median PTr did not exceed 1.2, regardless of the severity of hypoperfusion. With increasing combined injury severity and hypoperfusion there was a stepwise increase in the PTr. The worst coagulopathies were seen in patients admitted with the combination of the most severe injuries and highest base deficits (ISS > 35 and BD > 12, median PT ratio = 2.0). A similar pattern of co-association was seen when mortality is plotted as the dependent variable (Fig. 2B). ATC appears dependent upon the combined effects of both tissue trauma and systemic hypoperfusion and does not occur to a clinically significant degree without both (Table 2).

Figure 2.

 The relationship between injury severity and shock. (A) Median prothrombin ratios of patients grouped according to injury severity score (ISS) and base deficit (BD). *< 0.001 compared with ISS < 16, BD ≤ 0. (B) Mortality of patients grouped according to ISS and BD. *< 0.001 compared with ISS < 16, BD ≤ 0.

Table 2.   The relationship between injury severity and shock
 ISS < 16ISS 16 – 24ISS 25 –35ISS > 3 5
  1. BD, base deficit (mmol/); ISS, injury severity score; N, % of patients in subgroup; PTr, prothrombin time ratio (median).

BD = 011%1.11%4%1.14%4%1.115%1%1.220%
BD 0.1–619%1.12%13%1.14%15%1.225%9%1.426%
BD 6.1–123%1.14%3%1.29%6%1.329%6%1.549%
BD > 121%1.221%1%1.540%2%1.560%3%270%

Rodent responses to shock and trauma

Animals in all four groups received a mean of 11–13% estimated circulating volume of normal saline. This resulted in a similar drop in hemoglobin (Hb) and hematocrit (Hct) for both H and TH groups (Hb: H 8.0 vs. TH 9.5g dL−1, = ns; Hct: H 25 vs. TH 29%, = ns). Arterial lactate concentration was increased in hemorrhagic shock compared with sham controls (H 6.7 vs. S 2.3mmol L−1; < 0.001), but not significantly increased by trauma (TH 8.6 vs. H 6.7mmol L−1; = ns) (Fig. 3A). An increase in lactate was associated with mild metabolic acidosis (pH: H 7.30 vs. S 7.40, < 0.001), but not significantly increased by trauma (TH 7.25 vs. H 7.30, = ns).

Figure 3.

 Physiology and coagulopathy of the acute traumatic coagulopathy (ATC) Rodent Model. (A) Sham and trauma groups display similar arterial lactate concentrations. Animals subjected to hemorrhagic shock, with or without trauma, develop a similar increase in arterial lactate concentrations. *< 0.001 compared with Sham. (B) Sham and trauma groups display similar prothrombin time (PT) and activated partial thromboplastin times (APTT) times (PT: T 18.01s vs. S 18.00s, = ns; APTT: T 15.22s vs. S 15.39s, = ns). Animals subjected to hemorrhagic shock display a trend towards prolonged PT and APTT times (PT: H 20.37s vs. S 18.00s, = ns; APTT: H 17.77s vs. S 15.39s, = ns). Animals subjected to both trauma and hemorrhagic shock develop significantly prolonged PT and APTT times (PT: TH 23.47s vs. 18.00s, < 0.001; APTT: TH 20.89s vs. S 15.39s, < 0.001). *< 0.01 compared with hemorrhage alone. +< 0.01 compared with hemorrhage alone.

Rats subjected to trauma alone did not have a different coagulation function compared with sham controls, while hemorrhage caused a small, but statistically insignificant, prolongation of both PT and APTT (Fig. 3B). Significant coagulopathy was seen only in rats subjected to a combination of trauma and hemorrhagic shock (PTr = 1.3, APTTr = 1.36). Comparing all rats with a significant lactate rise (> 6 mmol L−1) regardless of group, only injured rats developed a statistically significant coagulopathy (PTr: 1.28 vs. 1.21, < 0.01; APTTr: 1.55 vs. 1.31, < 0.05). The rat model of combined trauma hemorrhage produced an endogenous coagulopathy consistent with ATC observed in the clinical population.


We have shown that ATC is associated with worse outcomes at a PT ratio > 1.2. This may be a more appropriate definition for future use and one that significantly increases the reported incidence of ATC. These studies also demonstrate the interplay between injury and shock as drivers of ATC. Although coagulopathy is multifactorial, our experimental model reproduced an early coagulopathy consistent with the clinical features of ATC in the absence of significant dilution, acidemia or hypothermia.

ATC has variable definitions in the literature that are based on relatively weak evidence from studies in non-trauma patient populations. Some trauma studies have defined coagulopathy as clotting times outside the normal range for healthy subjects [6,7]. Most have used an INR > 1.5, adopted from generic international guidelines for initiating FFP administration [9–11]. This threshold was derived from correlations between clotting times and the incidence of dilutional microvascular bleeding in patients undergoing massive transfusion [13–16]. However, the pathophysiology of ATC differs substantially from dilution-induced coagulopathy and this may be reflected in hemostatic parameters used to monitor it [4,5,8].

The threshold we have identified is lower than previous definitions of ATC and reveals the clinical implications of more subtle early abnormalities of coagulation. It identifies an additional 17% of all trauma patients who are at risk of worse outcomes and may benefit from hemostatic therapies. As PTr increases beyond 1.2 there is a continuous rise in all cause mortality and transfusion requirements. A positive dose-response relationship between coagulation parameters and mortality has been identified previously [17]. Assimilating coagulation and mortality data from multiple institutions enabled us to capture a larger trauma cohort and improve the resolution of this relationship. When using standard laboratory tests of coagulation, ATC should be defined as an admission PTr > 1.2.

We have demonstrated in both clinical and experimental models that a combination of injury and shock is associated with the development of clinically relevant ATC. A positive correlation between injury severity and coagulopathy is consistent with all previous studies of ATC [4–8]. Our group has previously published a prospective study of 208 patients in which increasing systemic hypoperfusion was associated with a progressive prolongation of the PT and APTT [8]. This mirrored findings in a military study conducted over 30 years earlier during the Vietnam War [12]. Given that trauma-induced hemorrhage often co-exists with severe injury, shock may be the primary driver of ATC. However, in this larger clinical study we have differentiated a subgroup of patients who are not coagulopathic despite having severe system hypoperfusion (ISS < 16, BD > 12mmol L−1). The experimental model was consistent and rats subjected to hemorrhagic shock did not develop a significant coagulopathy without the addition of a significant tissue injury. This is consistent with a previous study in rabbits where coagulopathy was not seen in hemorrhagic shock without the additional intravenous administration of thrombin [18]. Our studies demonstrate that ATC is associated with the co-existence of both tissue damage and systemic hypoperfusion. Clinically ATC should be anticipated in patients with significant trauma and an abnormal base deficit.

In the experimental model we controlled for the exogenous drivers of coagulopathy such as hypothermia and hemodilution. All rats were kept normothermic, received minimal fluid administration (equivalent to < 750 mL of crystalloid in humans) and had only a mild reduction in their hemoglobin and hematocrit. Global dilution of coagulation factors to 50% of normal levels is required for the development of coagulopathy [14,19,20]. Although acidemia causes coagulation protease dysfunction, the levels of metabolic acidosis in the present study were not severe enough to explain the clinical coagulopathy. Exogenous acid administered to dogs and swine only induces a moderate prolongation of clotting times with a pH below 7.2 and 7.1, respectively [21–23].

An endogenous mechanism for the pathogenesis of ATC is consistent with our hypothesis of systemic activation of the Protein C pathway [8]. Thrombin is produced by activation of coagulation after tissue damage, which in turn would lead to fibrin generation. However, if hypoperfusion causes an increase in thrombomodulin, this could switch thrombin to its anticoagulant function via the generation of activated protein (aPC). In bleeding patients this may represent a harmful side-effect of an organ protective anti-inflammatory response. Antibodies to anticoagulant domains of aPC improve coagulation function in a mouse model of ATC [24]. APC has yet to be measured in human ATC and it is unclear whether this is the dominant mechanism for the coagulopathy.

There are several limitations to our studies. Our clinical study was retrospective and based on available registry data. There were differences in registry inclusion criteria between institutions. Full datasets were not available on a large number of patients, and it may be that base deficits and prothrombin times are only performed on more severely injured patients. This may explain the high massive transfusion rate. The amount of pre-hospital fluid received and the admission body temperature were not included in all datasets. However, given the short transport times and the pre-hospital care practices (low volume fluid resuscitation, active environmental warming) it is unlikely that pre-hospital dilution and hypothermia were severe enough to make a significant contribution to the coagulopathy. Patient medication and pre-morbid coagulation status were not always known and these may be confounders, although the majority of trauma patients are young adults. The APTT or other tests of hemostasis were also not available.

Prothrombin ratios were calculated from recorded PT/QV/INR results, which may vary between institutions depending upon the test methods used. With the threshold for ATC reduced to a PTr of 1.2, there is now a narrow window between normal and abnormal PT readings, and the effects of differences in reagents may become more important. As much care as possible was taken in normalizing these variables to PTr, and the ISIs for the reagents were similar across centers. Definitions suitable for international research and clinical guidelines have to accept a degree of variability for the purposes of pragmatic utility.

There are more fundamental problems with the use of PT for the diagnosis of ATC. While cheap and readily available, it was not developed for the diagnosis of traumatic coagulopathy and assesses only a small component of clot formation. Laboratory PT results may not be available in a reasonable timeframe and point-of-care PT devices have not been evaluated. Other diagnostic devices such as thromboelastometry are available and require further evaluation for their utility in ATC [25,26].

For the experimental model, there are species differences between the coagulation systems of humans and rats [27]. Furthermore, the reproducibility of tissue injury in our experimental model has not been assessed by radiological or histological methods. However, we believe any variations to be relatively minor in terms of clinical effect and to have had minimal impact on the degree of coagulopathy elicited.


This is the first multinational study of the prevalence, etiology and clinical implications of ATC. The study included a large number of trauma patients receiving contemporary management in a wide geographic area, improving the validity and transportability of our findings compared with previous single-center studies. We have shown that ATC develops in response to the combination of tissue damage and systemic hypoperfusion. Using a novel experimental model we have confirmed that these drivers are sufficient for the development of coagulopathy. ATC on hospital admission is associated with increased mortality and transfusion requirements in a dose-dependent manner. When defined by standard clotting times, a PTr > 1.2 should be adopted as a clinically relevant definition of ATC.


D. Frith designed and performed the experimental model, data analysis and drafted the article. J.C. Goslings contributed to study conception, clinical data acquisition and revised the intellectual content. C. Gaarder contributed to study conception, clinical data acquisition and revised the intellectual content. M. Maegele contributed to study conception, clinical data acquisition and revised the intellectual content. M.J. Cohen contributed to study conception, clinical data acquisition and revised the intellectual content. S. Allard contributed to study conception and revised the intellectual content. P.I. Johansson contributed to study conception, clinical data acquisition and revised the intellectual content. S. Stanworth contributed to study conception and revised the intellectual content. C. Thiemermann contributed to experimental model design and revised the intellectual content. K. Brohi contributed to clinical and experimental study conception and design, data analysis and critical revision. All authors approved the final version for publication.


The authors would like to thank T. P. Saltzherr (Research Fellow, Trauma Unit Amsterdam Medical Center, Amsterdam), N. O. Skaga and M. Hestnes (Trauma Registry, Oslo University Hospital, Ulleval) for their assistance in collecting the data used in this study.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.