WHAT IS ALREADY KNOWN ON THIS TOPIC

Hypofibrinogenaemia on ED arrival is associated with poor patient outcomes, and early empirical fibrinogen supplementation has no clinical benefit and may cause harm.

Determining who requires fibrinogen supplementation in the prehospital setting could potentially speed up identification and delivery of appropriate blood products on ED arrival.

WHAT THIS STUDY ADDS

Point-of-care international normalised ratio (PoCINR)≥1.2, Fibrinogen on Admission in Trauma ≥4, Coagulopathy of Severe Trauma ≥4 and Trauma Induced Coagulopathy Clinical Score ≥12 all have a specificity ≥85% and may be useful in ruling in hypofibrinogenaemic states.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICYBackground

Trauma-induced coagulopathy (TIC) is a complex pathological process that affects approximately 25% of severely injured patients.1 Brohi et al’s landmark work demonstrated that trauma patients with evidence of coagulopathy were four times as likely to die compared with those without coagulopathy.2 While the exact pathophysiological mechanism of TIC is not fully understood,3 the development of hypofibrinogenaemia is thought to play a significant role.4 Hypofibrinogenaemia is one of the earliest derangements observed, with several studies reporting derangement in fibrinogen in the prehospital setting,5 6 and an Australian study finding that approximately 25% of trauma patients who received a prehospital blood transfusion had a fibrinogen level ≤1.5 g/L on ED arrival.7 Hypofibrinogenaemia on ED arrival and in the prehospital setting has been associated with both increased mortality and increased blood transfusions.5 8

While the provision of fibrinogen concentrate as part of a resuscitation strategy has demonstrated faster reversal of coagulopathy9 and a platelet-sparing effect compared with using plasma alone,10 CRYOSTAT-2 demonstrated that empirical replacement of fibrinogen early in the in-hospital care phase offered no clinical benefit and potentially caused harm in patients with penetrating trauma.11

The prehospital setting may be the ideal time to identify hypofibrinogenaemia and provide targeted fibrinogen supplementation and potentially improve the trauma patient’s clinical trajectory. A recent systematic review examining fibrinogen use recommended further studies to ‘focus on administration as early as possible from the point of injury or point of entry into the trauma system of care’.12 While it is feasible to provide fibrinogen concentrate in the prehospital setting,13 the challenge is to identify patients with hypofibrinogenaemia during the prehospital phase of care.

Two broad risk stratification approaches have been developed to detect TIC, namely point-of-care (PoC) devices and risk stratification tools. PoC devices that measure the international normalised ratio (INR) have been shown to correlate well with laboratory INR when used in the prehospital setting for trauma patients.14 Furthermore, in-hospital studies demonstrate that a laboratory INR correlated with fibrinogen levels of <1.5 g/L with an area under the receiver operating characteristic curve (AUROC) of ≥0.94.15 Several risk stratification tools have been developed to detect TIC, including the Fibrinogen on Admission in Trauma (FibAT), Coagulopathy of Severe Trauma (COAST) score and the Trauma Induced Coagulopathy Clinical Score (TICCS).16–18 Unfortunately, the definition of TIC used for the validation studies of these scores varied, with only the FibAT score being developed and validated to detect hypofibrinogenaemia.16

Given that PoC thromboelastic-guided, haemostatic therapy is used in some trauma centres, hypofibrinogenaemia was defined as the combined outcome of FibTEM A5<10 or fibrinogen level ≤1.5 g/L. The primary objective of this study was to determine the test characteristics of PoCINR and three risk stratification tools to identify hypofibrinogenaemia. The secondary objectives were to determine the test characteristics of PoCINR and three risk stratification tools to identify fibrinogen level ≤1.5 g/L and FibTEM A5<10.

MethodsSetting

The Queensland Ambulance Service (Queensland, Australia) operates a High Acuity Response Unit (HARU) in Brisbane and the Gold Coast which provide a road-based response servicing a population of approximately 2.8 million people. HARU has 24/7 emergency physician oversight and is staffed by critical care paramedics augmented by senior, critical care medicine registrars. The HARU scope of practice includes rapid sequence intubation, blood transfusions and focused abdominal sonogram for trauma (FAST). Each HARU carries two units of O negative packed red blood cells (pRBCs) with the Brisbane HARU also carrying one unit of thawed plasma. Both HARU units have the ability to source additional pRBCs from trauma centres.

Design

This was a prospective study of a convenience sample of adult trauma patients, treated by HARU clinicians between March 2017 and February 2020 and subsequently transported to a trauma centre. Blood was drawn in the prehospital setting prior to administration of tranexamic acid and tested for fibrinogen studies (Rotational Thromboelastometry (ROTEM) FibTEM and fibrinogen level) and PoCINR. Blood was not sampled if there were more urgent clinical priorities.

Venous samples were drawn for fibrinogen tests from a newly inserted or pre-existing intravenous cannula or via phlebotomy. PoCINR testing was conducted on venous samples (drawn for the fibrinogen studies) or on a capillary sample taken immediately after the venous sampling. PoCINR samples were tested on a Roche CoaguChek XS device (Millers Point, New South Wales, Australia) by the HARU clinician. The TICCS and COAST score were prospectively calculated by the HARU clinicians at the time of blood sampling. As the FibAT score was published after commencement of the study, a truncated FibAT score incorporating elements that could be obtained by HARU (HR, systolic BP, estimated age, temperature and FAST findings, see table 1) was calculated retrospectively based on the prehospital findings at the time of blood sampling. Fibrinogen study samples were tested within 3 hours of collection.

Table 1

Clinical scoring systems developed to detect trauma-induced coagulopathy

The components and test characteristics of these scores are shown in table 1.

HARU clinicians were not aware of the results of the subsequent fibrinogen studies and laboratory staff were not aware of PoCINR, COAST and TICCS scores.

Prehospital characteristics including prehospital time, interventions and time to blood sampling were collected. Outcome data including mortality and ED transfusion requirements were collated by the participating trauma services retrospectively.

A fibrinogen level ≤1.5 g/L or a FibTEM A5<10 was defined as fibrinogenaemia. These cut-offs are consistent with the Royal Brisbane and Women’s Hospital ROTEM goal-directed therapy targets (see the online supplemental material) and the European guidelines, respectively,19 and reflect the practice in the participating trauma centres, with fibrinogen being replaced if either FibTEM A5 or fibrinogen levels are low. This study was geared to focus on ‘ruling in’ patients with hypofibrinogenaemia as harm has been shown with treating patients unnecessarily,11 and thus a clinically significant specificity threshold of greater than 85% was chosen. This threshold was based on the specificity of commonly used ED PoC tests such as ECGs for ST-segment elevation myocardial infarctions (specificity: 72–89%)20 and bedside urinalysis for urinary tract infections (specificity of nitrites 91–95%, specificity of 3+ leucocytes 72–85%).21

Patient and public involvement

As part of the Queensland Civil Administration Tribunal (QCAT) review for consent being waived, the study protocol was reviewed by laypeople members of the QCAT panel.

Patients were not directly involved in other aspects of this study. The Standards for Reporting of Diagnostic Accuracy Studies was adhered to (see the online supplemental material).

Statistical analysis

Data were collated and processed using Microsoft Office Excel (V.16.14.1, Microsoft) and analysed using Statistical Package for the Social Sciences (SPSS) (Released 2017. IBM SPSS Statistics for Windows, V.25.0. IBM). Continuous variables were reported as medians and IQRs and categorical variables as frequencies and percentages.

Sensitivity and specificity for each score cut-off were calculated. AUROC analyses using PoCINR, COAST, TICCS and FibAT were performed for the detection of hypofibrinogenaemia as defined by a FibTEM A5<10 or fibrinogen level≤1.5 g/L. Receiver operating characteristic (ROC) curves were drawn and the AUROC values of 0.9–0.99, 0.8–0.89, 0.7–0.79, 0.6–0.69 and <0.6 were classified as excellent, very good, good, moderate and poor, respectively. Statistical significance was accepted at p<0.05, with 95th centile CIs (95% CI) reported for sensitivity, specificity and area under the curve. AUROC was considered statistically significant if the 95% CI did not span 0.5.

To allow a comparison between all risk stratification tools, the primary analysis only included patients who had results for all the risk stratification tools (PoCINR, COAST, TICCS and FibAT). Given that a large proportion of recruited patients did not have a PoCINR performed, a complete case, secondary analysis was performed using all valid FibTEM A5 and fibrinogen-level data (regardless if a PoCINR was performed) to examine the test characteristics of COAST, TICCS and FibAT.

Missing data for FibAT were imputed. The imputation model used in the analysis assumed that if a prehospital FAST was not performed, it was negative, and to impute values for temperature, FibAT scores were calculated without temperature, ranked in order, then subranked according to Injury Severity Score. The median temperature of the two temperatures above and below the missing data point was used. Two sensitivity analyses were conducted with an imputation model where missing values were assumed to be negative and an imputation model where missing values were assumed to be positive (see the online supplemental material).

Results

Between March 2017 and February 2020, a total of 222 patients had blood drawn and had a FibTEM and/or fibrinogen level performed. Of this cohort, 152 patients had a PoCINR performed and were included in the primary analysis (figure 1). Nine patients had missing FibAT FAST data, and one patient was missing a FibAT temperature data.

Figure 1

Patient flow diagram. COAST, Coagulopathy of Severe Trauma score; FibAT, Fibrinogen on Admission in Trauma score; HARU, High Acuity Response Unit; PoCINR, point-of-care international normalised ratio; TICCS, Trauma Induced Coagulopathy Clinical Score.

Patient characteristics for the 152 patients with a PoCINR are shown in table 2 and in the online supplemental material. The majority of patients were male and there was a predominance of vehicle-related trauma, with penetrating trauma accounting for approximately 18% of the cohort (figure 1). The median time after emergency medical service (EMS) call to blood sampling time was 53.1 min and the median EMS arrival to blood sampling time was 42 min (29–56 min). The median EMS call to hospital arrival time was 75.4 min. Prior to blood sampling, the median volume of crystalloid administered was 0 mL (0–250 mL), with six patients (4.1%) having one unit of pRBCs commenced and one patient (0.7%) having commenced a second unit of pRBCs.

Table 2

Characteristics and outcomes of included patients, n=152

Hypofibrinogenaemia was determined to be present in 48 patients (31.6%). Table 3 shows the test characteristics of the risk stratification tools for hypofibrinogenaemia, fibrinogen level ≤1.5 g/L and FibTEM A5<10.

Table 3

Test characteristics for the risk stratification tools for hypofibrinogenaemia, fibrinogen level ≤1.5 g/L and FibTEM A5<10

Figure 2A–D demonstrate the sensitivity and specificity of the risk stratification tests and proportion of patients with hypofibrinogenaemia at each cut-off value of the risk stratification tool.

Figure 2

Test characteristics of risk stratification tools versus hypofibrinogenaemia. (A) Point-of-care international normalised ratio (PoCINR). (B) Coagulopathy of Severe Trauma (COAST) score. (C) Trauma Induced Coagulopathy Clinical Score (TICCS). (D) Fibrinogen on Admission in Trauma (FibAT) score.

ROC curves for hypofibrinogenaemia are shown in figure 3. PoCINR was the only tool that had statistically significant AUROC, demonstrating good performance detection of fibrinogen level ≤1.5 g/L and moderate performance detecting clinically significant hypofibrinogenaemia and FibTEM A5<10 (figure 3 and table 3).

Figure 3

Receiver operating characteristic curve comparison for hypofibrinogenaemia. AUC, area under the curve; COAST, Coagulopathy of Severe Trauma score; FibAT, Fibrinogen on Admission in Trauma score; PoCINR, point-of-care international normalised ratio; TICCS, Trauma Induced Coagulopathy Clinical Score.

The complete case, secondary analysis demonstrated FibAT≥4, COAST≥4 and TICCS≥12 yielded test characteristics similar to the primary analysis in detecting hypofibrinogenaemia, fibrinogen level ≤1.5 g/L and FibTEM A5<10 (see the online supplemental material). A FibAT≥4 demonstrated a specificity exceeding 85% for all the sensitivity analyses performed (see the online supplemental material).

Discussion

In this study, we demonstrated that PoCINR was the only risk stratification tool that had statistically significant, moderate to good discrimination for detecting hypofibrinogenaemia, fibrinogen level ≤1.5 g/L and FibTEM A5<10. The other risk stratification tools demonstrated non-statistically significant AUROC.

The risk stratification tools were able to demonstrate a specificity ≥85% for hypofibrinogenaemia at the following cut-offs: PoCINR≥1.2, FibAT≥4, COAST≥4 and TICCS≥12. Only the truncated FibAT had a similar specificity cut-off (FibAT≥4) to its derivation paper,16 with COAST≥3 and TICCS≥10 (the cut-offs recommended in the derivation papers) demonstrating specificities of 62.9% and 75%, respectively, for hypofibrinogenaemia.

PoCINR also has other advantages over clinical risk scores. It is an objective measure; it can be repeated and may detect ongoing hypofibrinogenaemia requiring targeted supplementation despite improvement in physiological parameters (eg, BP, HR) from haemostatic resuscitation.7 There are some limitations with PoCINR use in the prehospital setting. First, the device needs to be readily available and maintained. Second, as demonstrated in this study, it was not always possible to get readings due to equipment issues, a finding similar to a French EMS study.22 In light of these limitations, device manufacturers should look to making PoC devices that are robust and suitable for use in the prehospital setting.

Of the clinical scores, the truncated FibAT is superior in terms of its logistic and clinical characteristics in detecting hypofibrinogenaemia. The truncated FibAT is objective, incorporates the use of FAST, which with repeated use allows dynamic assessment for ongoing bleeding with concurrent resuscitation. Unlike COAST and TICCS, the truncated FibAT had a similar threshold specificity cut-off as its derivation study which is likely due to FibAT being the only scoring system to specifically evaluate for hypofibrinogenaemia. The clinical limitations of the COAST and TICCS scores have been summarised elsewhere,23 24 and include the subjectivity of their variables, but these scores are relatively easy to use and do not require any equipment to calculate.

In light of the poor performance of the clinical tools in relation to their AUROC characteristics, it appears preferable that robust PoC devices that measure fibrinogen rapidly from capillary or venous blood should be developed and validated for use in the prehospital environment.

We chose to evaluate the risk stratification tools based on specificity and ‘ruling in’ clinically significant hypofibrinogenaemia given the opportunity costs and the financial expense25 of providing empirical fibrinogen to patients who may not benefit and may potentially be harmed from fibrinogen supplementation as shown in CRYOSTAT-2.11 The use of high-specificity tools may be incorporated into clinical guidelines for early fibrinogen supplementation in trauma, and also be used as inclusion criteria for future clinical studies examining patients with hypofibrinogenaemia in the context of trauma.

Our study has several limitations. The cohort was a convenience sample due to clinical and logistical limitations, resulting in spectrum bias. One effect of this was that a more injured cohort may not have been recruited or had blood sampling delayed due to competing clinical needs. Given the pragmatic nature of this study, blood was not sampled until emergent clinical care was completed, as the blood sampling did not contribute to the clinical care (ie, the receiving hospitals did not use these results to assist in patient management). In addition, HARU dispatch is geared towards a more severely injured cohort, and thus less injured patients may not have been included. While these would have influenced the prevalence of hypofibrinogenaemia in our cohort, the cohort’s characteristics were similar to a European study demonstrating the feasibility of prehospital administration of fibrinogen concentrate.13

The sample size was relatively small. Although the prevalence of hypofibrinogenaemia in this cohort was approximately 30%, the total number of patients was low, thereby reducing the precision of our results, especially the sensitivity of the risk assessment tools. To mitigate against this, we conducted a secondary, complete case analysis involving all the cases with either FibTEM or fibrinogen level measured, which increased the sample size by 50, and produced similar results for the non-PoCINR risk assessment tools. We also had a cohort of patients who had an incomplete truncated FibAT. We conducted several additional analyses and found that the test characteristics were similar to the imputation model used in the primary analysis (see the online supplemental material). Of the cohort studied in the primary analysis, nine of the 152 did not have a FAST performed and it was assumed that if a FAST was not performed it was negative. This aligns with HARU practice, where FAST is used if there is suspicion of abdominal or thoracic trauma.26 While unlikely, it is possible that a patient with intra-abdominal bleeding without clinically apparent truncal trauma did not receive a FAST and was misclassified to have a lower FibAT score.

We did not collect data regarding the preinjury, anticoagulation status of the patient cohort. This is unlikely to have affected the results for two reasons. First, the use of anticoagulants is unlikely to have affected the fibrinogen level,27 and second, trauma-based studies estimate preinjury anticoagulant use to be approximately 3%, with the prevalence increasing with increasing age,28 and given the relatively young age in this study, makes the use of anticoagulants less likely in this cohort.

Finally, there was heterogeneity about the assay used for fibrinogen testing (Clauss vs prothrombin time derived). While we observed a very strong correlation between fibrinogen level and FibTEM A5, which was comparable to studies using the Clauss method, and the area under curve for FibTEM A5 to detect fibrinogen ≤1.5 g/L was >0.9 (see the online supplemental material), which was also comparable to studies using the Clauss method,29 we acknowledge that at low fibrinogen levels, the derived value may be higher than the Clauss value.30 This may have resulted in misclassifying patients who had the derived measurement performed, thereby slightly altering the test characteristics.

This study demonstrates that it is possible to identify trauma patients with hypofibrinogenaemia with high specificity with both a PoCINR device and clinical scores. These findings support the implementation of risk stratification tools to identify hypofibrinogenaemia which would allow targeted fibrinogen replacement early in the clinical course of trauma patients. Future work should examine the feasibility of PoC devices to test for hypofibrinogenaemia in the prehospital setting and if the use of PoC devices and/or clinical risk stratification tools would result in more rapid fibrinogen replacement with resultant improvement in coagulation dysfunction and potentially improved mortality.

Data availability statement

Data are available upon reasonable request. Data are available upon reasonable request to the corresponding author.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by the Royal Brisbane and Women’s Hospital Ethics Committee (HREC/15/QRBW/186). QCAT waived the need for patient consent (QCATCRL015-15).

Acknowledgments

QAS High Acuity Response Unit Clinicians; Mr Lachlan Parker, QAS; Royal Brisbane and Women’s Hospital Trauma Service; Princess Alexandra Trauma Service; Ms Liz Wake Gold Coast University Hospital (GCUH); Pathology Queensland Central Transfusion Laboratory; Pathology Queensland Laboratory GCUH.