Substitution of ROTEM FIBTEM A5 for A10 in trauma: an observational study building a case for more rapid analysis of coagulopathy

Rotational thromboelastometry (ROTEM®) allows guided blood product resuscitation to correct trauma-induced coagulopathy in bleeding trauma patients. FIBTEM amplitude at 10 min (A10) has been widely used to identify hypofibrinogenaemia; locally a threshold of < 11 mm has guided fibrinogen replacement. Amplitude at 5 min (A5) carries an inherent time advantage. The primary aim was to explore the relationship between FIBTEM A5 and A10 in a trauma. Secondary aim was to investigate the use of A5 as a surrogate for A10 within a fibrinogen-replacement algorithm. Retrospective observational cohort study of arrival ROTEM results from 1539 consecutive trauma patients at a Level 1 trauma centre in Australia. Consistency of agreement between FIBTEM A5 and A10 was assessed. A new fibrinogen replacement threshold was developed for A5 using the A5–A10 bias; this was clinically compared to the existing A10 threshold. FIBTEM A5 displayed excellent consistency of agreement with A10. Intraclass correlation coefficient = 0.972 (95% confidence interval [CI] 0.969–0.974). Bias of A5 to A10 was − 1.49 (95% CI 1.43–1.56) mm. 19.34% patients met the original local threshold of A10 < 11 mm; 19.28% patients met the new, bias-adjusted threshold of A5 < 10 mm. ROTEM FIBTEM A5 reliably predicts A10 in trauma. This further validates use of the A5 result over A10 allowing faster decision-making in time-critical resuscitation of trauma patients. A modification of -1 to the A10 threshold might be appropriate for use with the A5 value in trauma patients.


Background
Trauma is a major cause of mortality globally and estimated to cause 4.5 million deaths annually [1]. 25-50% of trauma deaths are due to exsanguination [2,3], and swift control of haemorrhage with replacement of blood products underpins the resuscitation of trauma patients. An alternative to standard, fixed-ratio major haemorrhage protocols involves targeted resuscitation guided by point-of-care viscoelastic haemostatic assays (VHAs), such as ROTEM (ROTEM®, Werfen, Barcelona, Spain), which identify deficient and dysfunctional coagulative components. This allows patientspecific administration of blood products to correct coagulopathy [4]. ROTEM FIBTEM assays assess the function of fibrinogen in clotting blood [5]. Low clot amplitude is indicative of hypofibrinogenaemia [6], and predictive of the need for massive transfusion [7]. In guided resuscitation algorithms, this usually triggers fibrinogen replacement using fresh frozen plasma, cryoprecipitate or fibrinogen concentrate. The use of VHAs in trauma is a developing research field: observational data [8] and a single-centre randomised control trial have suggested VHA might increase survival versus fixed-ratio blood product resuscitation [9]. On the contrary, VHA-augmented resuscitation did not improve outcomes when compared to resuscitation augmented by conventional laboratory clotting tests in the recent iTACTIC multicentre randomised control trial [10]. The European Task Force for Advanced Bleeding Care in Trauma advocated for the use of VHAs but did not recommend specific cutoff values to use; "more research is required… physicians should use their own judgement when developing local policies" [11].
Trauma services, including our own, have developed guidelines for fibrinogen replacement using FIBTEM amplitude at 10 min (A10) with a threshold for fibrinogen replacement of < 11 mm. Over time local practice moved to using A5 (amplitude at 5 min) as a surrogate for A10 values to guide resuscitation more rapidly (Additional File 1: current ROTEM-guided transfusion algorithm). This was performed without changing the threshold value of < 11 mm.
The literature demonstrates that FIBTEM A5 may be better at predicting laboratory-based hypofibrinogenaemia than maximum clot formation (the maximum amplitude reached during a ROTEM assay) in large trials [6,12]. A5 thresholds of 9.5 mm and 10 mm, respectively, were shown to predict laboratory-based hypofibrinogenaemia with sensitivities (specificities) of 78% (70%) and 70% (76%). However, these studies did not present predictions of hypofibrinogenaemia based on A10 for comparison. The relationship between the two early FIBTEM variables has been inadequately studied. FIBTEM A5 and A10 have been shown to be highly correlated in multiple surgical settings [13][14][15] and EXTEM A5 and A10 to correlate strongly in the setting of trauma [16]. Yet, data on the reliability, direction and magnitude of agreement between FIBTEM A5 and A10 are lacking. Clinically, this relationship is important to understand for the development of more accurate guided resuscitation algorithms. Given that the clot has had five more minutes to strengthen between the FIBTEM A5 and A10 result, we hypothesised that A5 results reliably predict A10 values that are greater in magnitude. Thus, A5 could be used to safely expedite the detection of and response to hypofibrinogenaemia.

Objective
The primary objective was to explore the relationship between ROTEM FIBTEM A5 and A10 in a large cohort of trauma patients and in subsets of patients with massive haemorrhage and/or major trauma, as these are the patients for whom the benefits of more rapid ROTEM results might be most apparent. The secondary objectives were to assess the clinical relevance of the difference between A5 and A10 results when applied to our current local fibrinogen-replacement algorithm threshold of < 11 mm, and to compare these against a modified threshold for A5 developed using the quantified relationship between FIBTEM A5 and A10 to modify the existing threshold.

Design and setting
This retrospective observational study is reported according to the STROBE Statement for observational studies [17]. The study setting was a 750-bed tertiary health service with a Level 1 Trauma Centre located in Southeast Queensland, Australia. The institution receives over 1500 trauma activations every year with an average of 300 being classed as major trauma (Injury Severity Score (ISS) ≥ 12) [18].

Participants
Participants included patients of all ages that met the trauma call criteria (briefly: injury to more than one body region or injury involving chest or abdomen, and/or significant mechanism of injury; full criteria in Additional File 2) and received a ROTEM on presentation to the emergency department between April 2014 and December 2016. Patients who did not receive a ROTEM on presentation were excluded.

Procedure
Each ROTEM sample was collected as part of routine admission blood tests into a 3.5 ml citrated Vacuette tube (Greiner Bio-One, Kremsmünster, Austria) and transported by hand from the Emergency Department to the Intensive Care Unit for analysis. A ROTEM Delta system was used to analyse samples. This is a semi-manual system that requires blood samples to be pipetted in multiple steps by trained operators. The system uses pin and cup technology with different reagent systems in each of the four channels [5]. In this study, the FIBTEM assay was used (extrinsically activated with tissue factor, the platelet inhibitor cytochalasin D and 300 ul of blood). ROTEM measurements were run for a minimum of ten minutes plus clotting time. This meant the minimum results were A5 and A10.

Data collection
A Trauma Registry was maintained contemporaneously comprising demographic and clinical data including injury mechanism, injury severity score and mortality. This was supplemented by data from the electronic patient records including ROTEM results, baseline laboratory tests and blood product administration in the first 24 h.

Outcomes
The primary outcome was consistency of agreement between pairs of FIBTEM A5 and A10 results in the whole cohort and by subgroup analyses of patients with massive transfusion (≥ 10 units packed red cells in 24 h) and major trauma (ISS ≥ 12). Secondary outcomes included the measured bias (mean difference) between A5 and A10 pairs, and assessment of the clinical relevance of this difference by counting numbers of patients meeting fibrinogen-replacement threshold within our algorithm using the original A10 < 11 mm; current A5 < 11 mm; and a calculated new threshold for A5 (generated by applying the bias to the < 11 mm threshold).

Statistical methods
Data were analysed using STATA (StataCorp, 2017, Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC) and Microsoft Excel (Microsoft 2019, Version 1912 (Build 12,325.20288)). ROTEM data were first screened and biologically implausible results and patients with missing data were excluded from the analysis. Descriptive data analysis was performed using counts and percentages for categorical data. Continuous data were described using median and interquartile range.
For the primary outcome, whole cohort FIBTEM A5 and A10 values were assessed for consistency of agreement using the intra-class correlation coefficient (ICC) based on a twoway mixed-effects model where the A5 and A10 tests were treated as fixed effects and the blood samples as random effects. Agreement was also assessed using a Bland-Altman plot which plots the mean of the two scores to the difference between scores and can identify systematic bias and heterogenous variation.
Subgroup analyses were performed. In the first, participants were grouped by number of red cell transfusions in the first 24 h (nil; one to three units; four to nine units; ten or more units). Ten or more units were selected as the last group as this traditionally represents massive transfusion. In a second analysis, participants were grouped by Injury Severity Score (ISS; zero to seven, eight to eleven, twelve or more). The latter group represented major trauma [18].
For the secondary outcomes, numbers of patients reaching ROTEM-guided threshold for fibrinogen replacement according to local fibrinogen replacement algorithm (FIBTEM amplitude < 11 mm) were counted using the A5 and A10 results to assess clinical relevance of any difference between A5 and A10 results.
Based on the bias (mean difference) between the A5 and A10 results, the threshold for fibrinogen replacement was modified for A5 values and the number of patients reaching this new threshold was counted. Numbers of patients who actually received fibrinogen were counted in each case for comparison.

Ethics
Ethics approval was obtained from the Gold Coast Hospital and Health Service Human Research Ethics Committee (HREC number blinded for peer review). Waived consent was granted as section 2.3 of the National Statement on Ethical Conduct in Human Research.

Participants
Over the three-year period between 20/1/14 and 30/12/16, 4495 patients met the trauma call criteria and 1539 patients had ROTEM FIBTEM performed. One patient's data were removed due to implausible ROTEM results and 13 patients did not have an A10 value recorded. This left 1525 patients for the primary analysis (Fig. 1).
Of the 1525 patients used for primary analysis, 1124 (73.70%) were male. All continuous variables displayed nonnormal distributions. Median age was 37 years (interquartile range [IQR] 25-54). Blunt trauma comprised the main mechanism of injury, present in 1391 (91.21%) cases. 474 patients (31.08%) were classified as major trauma (ISS ≥ 12) and 123 (8.07%) received packed red cell transfusions in the first 24 h from arrival. Further descriptive statistics can be found in Table 1.

Primary outcomes
In the whole cohort, A5 and A10 values had an ICC of 0.972 (95% CI 0.969-0.975). ICC values above 0.90 are considered to represent an excellent level of agreement [19].
Bland-Altman plots were also constructed to indicate agreement and plots for the whole cohort and the subgroups of massive haemorrhage and major trauma (Fig. 2).

Secondary outcomes
From the whole cohort Bland-Altman plot, the calculated bias (mean difference) between A10 and A5 values was observed to be 1.49 mm (95% CI 1.43-1.56 mm), indicating the average A10 value was 1.49 mm greater than its counterpart A5 value. 295 (19.34%) patients met the original local ROTEM threshold (A10 < 11 mm) for fibrinogen replacement. 424 (27.80%) patients met the current ROTEM threshold (A5 < 11 mm) for fibrinogen replacement (see Additional File 1). A minority of patients in both groups actually received fibrinogen replacement (indicating that they had met the second clinical criterion of active bleeding): 45 (2.95% of total) and 51 (3.34% of total) patients, respectively.
The calculated new trial threshold for A5 (after subtracting the bias from the original A10 threshold of < 11 mm) was A5 < 10 mm (rounded to the nearest integer due to the precision of the ROTEM machine readouts   ROTEM criterion for fibrinogen replacement. Numbers of patients meeting these criteria are displayed in Table 4.

Discussion
The study has demonstrated an excellent consistency of agreement of ROTEM FIBTEM A5 and A10, with an ICC (95% CI) of 0.972 (0.969-0.975) across a large cohort of trauma patients. This level of consistency of agreement was maintained within all subgroups analysed and importantly in the groups of major trauma (ISS ≥ 12) and massive transfusion (≥ 10 units PRCs in 24 h). These sub-groups who are critically unwell are most likely to derive a treatment benefit from quicker identification and treatment of hypofibrinogenaemia during resuscitation and in which demonstrating reliability of FIBTEM A5 as a surrogate for A10 is most useful. Based on ROTEM results, locally published data demonstrate that teams can expect to deliver fibrinogen concentrate to such patients within 30 min or cryoprecipitate within 60 min [20,21]. Early ROTEM values (by virtue of being point-of-care) have shorter turnaround times than urgent laboratory coagulation studies; FIBTEM A5 has been measured to have a turnaround time of 12 min versus 37 min for Clauss fibrinogen [15]. The authors of this study are not aware of a study directly measuring turnaround time of A5 versus A10 results, but theoretically a five-minute difference would be expected. In severely bleeding patients, this potentially allows for identification and correction of coagulopathy earlier. It should be noted that multiple assays are often performed over the course of a resuscitation and this time benefit could be present at each stage. This study further supports the use of FIBTEM A5 over A10 as a rapid decision-making tool for targeted resuscitation of coagulopathy in trauma.
To the authors' knowledge, the consistency of agreement and magnitude of differences between FIBTEM A5 and A10 has not been reported previously in the field of trauma. Strong correlations (albeit inappropriate Pearson or Spearman correlation [22]) have been demonstrated in elective surgery (cardiac and non-cardiac) studies [13][14][15]. Similarly, strong correlations have been observed in trauma between other related ROTEM values: FIBTEM A5 and MCF; A10 and MCF [23] and EXTEM A5 and A10 and MCF [16]. The intra-class correlation was used in this study (versus Pearson correlation) as it validly provides a more robust measure of consistency of agreement for the comparison of two diagnostic tests measuring the same clinical concept [24,25].
The findings demonstrate FIBTEM A10 had a bias of 1.49 mm from A5 (A10 result was on average 1.49 mm greater than its corresponding A5 result). This was in line with our hypothesis: a positive difference is to be expected as the blood clot has had five more minutes to strengthen between the A5 and A10 result. This bias was used to modify our existing fibrinogen replacement threshold with cautious success. Our institution's protocol for fibrinogen   a small minority of these extra patients (six) received fibrinogen replacement. The most likely reason for this is that for the majority there was no clinical suspicion of bleeding. Nonetheless, use of the current A5-based criterion might predispose to increased replacement of fibrinogen. Unnecessary replacement of fibrinogen with fibrinogen concentrate, cryoprecipitate or fresh frozen plasma is undesirable. There is a relatively unknown degree of risk of iatrogenic harm from the products themselves [26,27]; misinterpretation of a 'step-wise' algorithm could lead to failure to correct other abnormal variables; and all methods of replacement are financially costly [28]. Our modification of the A5 threshold resulted in 19.28% meeting the replacement criterion. This is noted to be almost exactly the same proportion as using the original threshold of A10 < 11 mm. In similar cohorts of trauma patients, Rourke et al. [6] and Baksaas Aasen et al. [12] found that FIBTEM A5 thresholds of 9.5 mm and 10 mm, respectively, were optimal for detecting laboratory Clauss hypofibrinogenaemia of < 1.5 g/l and < 2.0 g/l with sensitivities (specificities) of 78% (70%) and 70% (76%). The current study did not include receiver operating curve analysis comparing FIBTEM A5 to Clauss fibrinogen and so we cannot provide direct validity to these thresholds. These studies demonstrate that there is no consensus as to what level of hypofibrinogenaemia should trigger transfusion. However, it is likely that our threshold of FIBTEM A5 < 10 mm corresponds to Clauss hypofibrinogenaemia < 1.5-2.0 g/l. Further work is required to develop a consensus ROTEM FIBTEM threshold for replacement of fibrinogen in the context of trauma [11].
The authors acknowledge several limitations to the current study. Whilst it includes every patient who had a ROTEM on admission, not all trauma patients had a ROTEM. However, the likely bias in the study in this case would be towards including more severe trauma and this is the cohort of most clinical relevance. The particular viscoelastic haemostatic device used was ROTEM Delta. This is a multi-step, partially manual assay that was performed by clinicians in an emergency trauma setting. Whilst all users had training there are multiple stages where errors could occur. The fully automated ROTEM Sigma assay was not studied. Finally, the analyses of numbers of patients meeting different FIBTEM criteria for fibrinogen replacement were retrospective and theoretical: throughout the study period the clinical criterion used was the 'current' threshold of A5 < 11 mm.

Conclusion
This study provides further evidence to validate the use of ROTEM FIBTEM A5 in place of A10 in trauma. In a large cohort of trauma patients FIBTEM A5 has excellent agreement with A10 and this is preserved in the subgroups of major trauma and massive transfusion. There is a small difference in magnitude between A5 and A10 results in the same patient and interchangeable use of values is inadvisable as it might identify quite different numbers of patients as deficient in fibrinogen. Our findings back previous work suggesting a FIBTEM A5 threshold of < 10 mm may be appropriate for targeted fibrinogen replacement when used in the context of traumatic haemorrhage. Useful future work might involve receiver operating curve analyses of ROTEM FIBTEM A5 versus A10 in the same cohort of patients and similar work with EXTEM values to guide platelet transfusion.