Sunteți pe pagina 1din 6

Official Journal of

the British Blood Transfusion Society

Transfusion Medicine | SHORT COMMUNICATION

The application of a haemorrhage assessment tool in evaluating

control of bleeding in a pilot trauma haemorrhage trial

N. Curry,1,2 C. Foley,3 H. Wong,1,2,4 A. Mora,3 E. Curnow,3 A. Zarankaite,3 R. Hodge,3 V. Hopkins,3 A. Deary,3 J. Ray,5
P. Moss,6 M. J. Reed,7 S. Kellett,8 R. Davenport9 & S. Stanworth1,2,4
Department of Haematology, Oxford University Hospitals NHS Trust, Oxford, UK, 2 NIHR BRC Blood Theme, Oxford University,
Oxford, UK, 3 NHS Blood and Transplant Clinical Trials Unit, NHS Blood and Transplant, Cambridge and Bristol, UK, 4 NHS Blood and
Transplant, John Radcliffe Hospital, Oxford, UK, 5 Department of Emergency Medicine, John Radcliffe Hospital, Oxford, UK, 6 Department
of Emergency Medicine, St. George’s Hospital, London, UK, 7 Emergency Medicine Research Group Edinburgh (EMERGE), Royal
Infirmary of Edinburgh, Edinburgh, UK, 8 Department of Anaesthetics, University Hospital Southampton NHS Foundation Trust,
Southampton, UK, and 9 Centre for Trauma Sciences, Blizard Institute, Queen Mary University of London, London, UK

Received 1 May 2019; accepted for publication 6 October 2019

SUMMARY Major haemorrhage is one of the leading causes of death world-

wide. It affects up to 40% of trauma patients (Stanworth et al.,
Objectives: To determine whether it was feasible to use a 2016) and is one of the most common causes of death in women
haemorrhage assessment tool (HAT) within a trauma trial and at the time of delivery (Say et al., 2014). Often, clinical trials
whether the data obtained could differentiate patients who had evaluating interventions to treat bleeding select transfusion end-
achieved haemostasis. points (such as reduction in red cell usage) (Rahe-Meyer et al.,
Background: Major haemorrhage is one of the leading causes of 2016; Collins et al., 2017) as identified in a recent systematic
death worldwide, affecting 40% of trauma patients. Clinical trials review of pro-haemostatic agents (Fabes et al., 2018), and only
evaluating haemostatic interventions often use transfusion out- a few larger trials have assessed mortality (CRASH-2 trial col-
comes as a primary endpoint. Transfusion is highly dependent laborators, 2010; Holcomb et al., 2015; WOMAN trial Collabo-
on local practice, limiting its reliability as a robust, transferable rators, 2017; Brenner et al., 2018). However, transfusion therapy
endpoint. is highly dependent on local clinical practice, which reduces its
Methods: A 5-point HAT questionnaire was applied to partici- reliability as a robust endpoint, and at best, transfusion require-
pants enrolled into the EFIT-1 trial. This RCT evaluated the fea- ment can be viewed as a surrogate, rather than direct, measure
sibility of administering a 6 g fibrinogen concentrate to patients of bleeding.
with severe trauma haemorrhage. Control of bleeding remains an attractive and relevant end-
Results: Of participants, 98% completed a HAT; 75% partici- point but is a notoriously difficult question to standardise in a
pants had ‘achieved haemostasis’ at the time of tool completion, multicentre trial and may become increasingly relevant as more
as determined by clinical acumen alone. HAT scores were able haemostasis studies are conducted. The PROPPR (Pragmatic,
to differentiate which participants required transfusion after 3 h. Randomized Optimal Platelet and Plasma Ratios) study was
Of participants, 56% were transfused red blood cells when they the first to use ’time to haemostasis’ as a secondary endpoint.
scored 0–2, compared to 17% with HAT scores between 3 and 5. This study evaluated the effect of two blood transfusion ratios
on mortality in trauma haemorrhage (Holcomb et al., 2015).
Conclusion: This study has confirmed the feasibility of using
Importantly, data (adjusted for multiple variables) from this trial
a HAT during the emergency care of patients suffering trauma
demonstrated that shorter times to haemostasis were reported
haemorrhage, and future studies should be conducted to deter-
to be independently associated with reduced 30-day mortality
mine its value as an endpoint in haemostasis studies.
(Holcomb, 2018).
Key words: haemorrhage asssessment tool, major haemor- We report the use of a trial-specific haemorrhage assess-
rhage, randomised controlled trial, trauma. ment tool (HAT), adapted from the tool used in PROPPR (Hol-
comb et al., 2015), within a multicentre UK trauma trial (Curry
Correspondence: Dr. Nicola Curry, Oxford University Hospitals NHS et al., 2018). We piloted the use of this tool, with the record-
Foundation Trust, Oxford Haemophilia & Thrombosis Centre, ing of a HAT score, to (i) determine whether it was feasible
Churchill Hospital, OX3 7LE, UK. to use by evaluating the numbers of HATs completed within
Tel.: 01865 225316; fax: 01865 225608; a trauma RCT, (ii) understand how often a participant had
e-mail: achieved haemostasis and (iii) determine whether the HAT score

© 2019 British Blood Transfusion Society doi: 10.1111/tme.12644

2 N. Curry et al.

Fig. 1. Haemorrhage assessment tool, E-FIT-1 trial. Definitions for the HAT: Q4a - this question should be posed to the trauma team leader and should
be based on the patient’s current clinical status (in the last 30 minutes); Q4b- stable BP defined as a systolic BP which is not falling and has variation of
less than 20mmHg during the 30 minutes prior to the HAT; Q4c - stable HR defined as a HR variation of less than 20 bpm when comparing the average
HR at the time of the HAT with the 30 minutes prior to the HAT; Q4d - stable urine output is defined as production of urine of at least 10mL per hour (if
the patient is an uric the answer given should be no); Q4e - decreasing lactate or base excess confirmed if the value for the most recent result is less than
the preceding sample; Q4f - rate of blood transfusion reducing is defined as the number of units (RBC, FFP, platelets and cryoprecipitate) administered
in the last 30 minutes is less than the 30 minutes preceding this time; Q5-6 -clinical opinions (where relevant) should be sought from the surgeon or
radiologist based on the current status of the patient. The HAT assessment score was measured out of 5, where a score of 5 was calculated if ‘yes’ was
given for all answers Q4b - Q4f. For Q4d - if not applicable was returned, the score given for this question was a zero.

Transfusion Medicine, 2019 © 2019 British Blood Transfusion Society

A haemorrhage assessment tool for major traumatic bleeding 3

provided additional information when compared to simple ’clin- 3 h as defined by the surgeon, confirming no further bleeding
ical acumen’. intraoperatively. Only one of the three participants being treated
in interventional radiology had achieved radiological haemosta-
sis at 3 h.
Development of tool Comparing the HAT score to a question on control
The HAT was based on the tool used in the PROPPR study and of bleeding
used five questions aimed at recognising patients in haemor- An additional question in the tool was: ’does the trauma team feel
rhagic shock. Our HAT used the addition of two further criteria that haemostasis has been achieved?’. We compared the results
asking: ’is the base excess or lactate improving?’ and ’is the trans- of this question with the 5-point HAT score. A total of 45 HAT
fusion rate decreasing?’. These additional questions were devel- scores were obtained for this question and these results were
oped by consensus from trauma surgeons and trauma anaes- assessed (Table 2). Thirty-six participants were documented to
thesiologists who were active in trauma haemorrhage research have ’achieved haemostasis’ by the trauma team at the point
across Europe. of HAT completion (18 placebo arm, 18 active arm); within
The HAT was applied to recruited patients at five major this group, 12 participants were in the theatre at the time of
trauma centres in the United Kingdom as part of the E-FIT-1 HAT completion, and 9 participants (75%) were also deemed to
randomised controlled trial, which evaluated the feasibility of have achieved surgical haemostasis. Nine participants had ’not
administering a 6 g fibrinogen concentrate to patients with achieved haemostasis’ at 3 h; five of this group were in theatre,
severe trauma and major bleeding (Curry et al., 2018). Full and all surgical teams agreed that the participants were still
details of the study have been described (Curry et al., 2018). bleeding.
In brief, patients were eligible if they were adults, were actively
bleeding and in haemorrhagic shock and required activation
Relationship to transfusion use
of the major haemorrhage protocol or had already received
a transfusion of emergency red blood cells. The protocol was Table 3 shows the participants’ transfusion requirements when
approved by an independent ethics committee, the NRES Com- grouped by their HAT score (HAT scores of: 0–2, 3 and 4–5)
mittee South Central – Oxford C (15/S3/0316) and MHRA (See Fig. 1 for details on scoring). Of the 47 HAT assessments, 46
(25 224/0003/001–0001). were completed in full (24 intervention arm, 22 placebo). HAT
As part of the study, a HAT was completed 3 h after hos- scores were able to broadly differentiate between participants
pital admission (Fig. 1). Figure 1 describes how to score the requiring transfusion after 3 h. If the HAT score was between
tool, with a maximum score of 5 for participants who had fully 0 and 2, just over half of the participants were transfused with
achieved haemostasis. Face-to-face education was completed RBC compared to fewer than one in five of those with scores
for research staff at each participating site prior to the start between 3 and 5. Similar results were seen for the use of FFP. In
of the study. Three hours was chosen following publication of 27 of the 47 completed HAT questionnaires (57%), urine output
the PROPPR study data, which demonstrated that the major- data were recorded as ’not applicable’ because there were no
ity of deaths from trauma haemorrhage occurred within this clinical means for determining urine output (commonly due to
timeframe. no catheter sited).

RESULTS Patients who failed to achieve bleeding control

Patients More than twice as many participants deemed to have ongoing

bleeding, according to the stand-alone question, continued
A total of 48 participants were enrolled in the E-FIT-1 study, to require blood components in the 3 h after the assessment
with 24 participants each in the intervention and comparator (between 3 and 6 h from admission). The majority of partic-
arms. Forty-seven HATs were completed (98%) at the 3-h time ipants did not require additional RBC treatment in either
point (±60 min), 24 in the active arm and 23 in the placebo arm study group after 3 h of admission, but for those with a
(baseline characteristics Table 1). There was one missing data HAT question assessment recording ongoing bleeding, the
point in one of the placebo arm questionnaires (lactate result not participants who received blood (n = 4) required 13 units
completed). The median time of completion of the tool was 2 h between 3 and 6 h (IQR: 5–21) compared to 2 units (IQR:
40 min (IQR: 2 h 15 to 3 h 03) from admission. 2–3) in the group (n = 7) where haemostasis was reported as
HAT score results
Of the 47 completed questionnaires, 17 participants were in the
theatre at the time of completion of the HAT, and of this group, This study has shown the feasibility of using a HAT during the
9 (53%) were deemed to have achieved anatomic haemostasis at emergency care of trauma patients suffering with significant

© 2019 British Blood Transfusion Society Transfusion Medicine, 2019

4 N. Curry et al.

Table 1. Haemorrhage assessment results according to study arm

Active arm Placebo arm Overall

(n = 24/24) (n = 23/24)2 (n = 48)

Bleeding controlled 18 18 36
Blood pressure stable or increasing 16 18 34
Heart rate stable or maintained within normal range 18 18 36
Urine output stable or >0·5 mL kg h−11 9 9 18
Lactate or base excess level decreasing 19 17 36
Rate of blood transfusion reducing 19 20 39
Participant in theatre 10 7 17
Surgeon confirmed anatomic haemostasis 5 4 9
Participant in interventional radiology 2 1 3
Resolution of a radiological blush 1 0 1

1 Questionwas reported as not applicable/not available for 27 participants.

2 Only23 participants completed the HAT for the placebo arm.
Data are numbers.

Table 2. Characteristics of the groups according to whether the trauma team deemed them to be bleeding or not (n = 45)

Haemostasis Haemostasis not

achieved (n = 36) achieved (n = 9)
Number (%) Number (%)

Number receiving blood components at time of assessment 14 (39) 8 (89)

Blood pressure stable or increasing 30 (83) 2 (22)
Heart rate stable or maintained within normal range 32 (89) 2 (22)
Urine output stable or > 0·5 mL kg h−11 17/18 (94) 1/1 (100)
Lactate or base excess level decreasing 31 (86) 4 (44)
Rate of blood transfusion reducing 33 (92) 5 (56)
Median score (of 5) for HAT 4 0
Participant in theatre 12 5
Surgeon confirmed anatomic haemostasis 9 (75) 0 (0)
Participant in interventional radiology 2 1
Resolution of a radiological blush 0 (0) 1 (100)
Transfusion data
Number receiving RBC between 3 and 6 h 7 (19) 4 (44)
Median number of RBC units received between 3 and 6 h for whole group 0 (IQR:0–0) 0 (IQR: 0–6)
Median number of FFP units received between 3 and 6 h 0 (IQR:0–0) 2 (IQR: 0–6)

1 Question was reported as not applicable/not available for 27 participants.

trauma haemorrhage. Very high rates of completion were better utilised to find risk of ongoing bleeding as it appears
observed for participants who were alive at the 3-h time point. to differentiate those patients requiring an ongoing transfusion
Three quarters of all participants had haemorrhage control need rather than determining whether a patient has achieved
achieved by 3 h. The results of the study-specific HAT appeared haemostasis.
to be able to distinguish between those still actively bleeding It has not been possible to determine whether any of the single
and those no longer bleeding. Patients with a score of 0–2 were parameters within the HAT are sensitive to detecting cessation
three times as likely to receive a transfusion of either RBC or of bleeding due to the small sample size. The question regarding
FFP in the 3 h following the assessment. urine output, which was included in our HAT tool, was not
The HAT, when compared to clinical acumen alone – i.e. ask- answered in 27 of 47 cases. This was due to the emergency nature
ing a simple question (does the trauma team feel the patient of the study and reflected the short timeframe within which
has achieved haemostasis?) – provided very similar results. The the HAT was completed. Future adaptations of this HAT would
HAT score and the trauma team leader’s assessment concurred benefit from reviewing this parameter.
in more than three quarters of all cases, suggesting that clin- When compared to the PROPPR study, the findings from
ical judgement in this small group appears to be reliable as our HAT appear consistent. In the PROPPR study, the median
a scoring method. These data suggest that the HAT may be time to haemostasis was 105 min (IQR, 64–179 min) in the

Transfusion Medicine, 2019 © 2019 British Blood Transfusion Society

A haemorrhage assessment tool for major traumatic bleeding 5

Table 3. Haemorrhage assessment tool scores and transfusion requirement (n = 46)

RBC units from RBC units FFP units from FFP units
HAT score admission to 3 h between 3 and 6 h admission to 3 h between 3 and 6 h

0–2 (n = 9) 12 (6–23) 3–18 (100%) 2 (0–6) 0–24 (56%) 8 (6–16) 2–15 (100%) 0 (0–6) 0–12 (44%)
3 (n = 8) 2 (1–3) 0–8 (87·5%) 0 (0–0) 0–2 (12·5%) 2 (1·5–4·5) 0–8 (87·5%) 0 (0–0) 0–1 (12·5%)
4 and 5 (n = 29) 3 (2–5) 0–12 (86%) 0 (0–0) 0–4 (17%) 3 (1–6) 0–13 (76%) 0 (0–0) 0–4 (10%)

Data are presented as median and IQR for units of RBC or FFP, followed by the range of units.
The data in % represent the proportion of the group in receipt of a transfusion.

intervention arm compared to 100 min (IQR, 56–181 min) patients considered to have an inherited bleeding disorder (i.e.
in the comparator arm, (P = 0·44). There was a signifi- International Society for Thrombosis and Hemostasis Bleed-
cantly greater number of participants in the 1:1:1 arm who ing Assessment Tool (ISTH BAT) (Rodeghiero et al., 2010))
achieved anatomic haemostasis (86·1%) compared to the 1:1:2 or to see how much bleeding has already occurred following
group (78·1%, P = 0·006). The results from our single-point a clinical bleeding episode (World Health Organization bleed-
study showed a similar result, such that 75% of patients had ing assessment tool (Kaufman et al., 2015)). No tool has been
achieved haemostasis at 3 h. Time to haemostasis would be validated and used contemporaneously to assess ’cessation of
a more ideal data point to collect but does require significant bleeding’.
research resources, which was undertaken as a continuous The results of our study and the PROPRR trial findings sug-
one-to-one follow-up in the US study for every participant gest that HATs can be completed within highly complex major
in that study, allowing a definitive time to haemostasis to be haemorrhage RCTs, which may have value as endpoints in
recorded. haemostasis clinical studies.
There are limitations to this study, most notably the small
sample size. Furthermore, the majority of the participants in this
study were deemed to have achieved haemostasis by the 3-h time ACKNOWLEDGMENTS
point, limiting the comparisons that could be made between
The trial was funded by CSL Behring.
actively and non-actively bleeding patients. Future studies would
N. C. and S. S. designed the study, wrote the protocol, analysed
need to apply the HAT tool at a much earlier time point, perhaps
data and wrote the manuscript. C. F. led the CTU running of the
at 1 h after admission to hospital. The tool itself appeared to
trial, analysed data and wrote the manuscript. H. W., A. M., R.
perform no better than clinical acumen, likely reflecting the
H., V. H. and A. D. conducted the study; conducted CTU trial
automatic use of many of the HAT parameters by clinicians
work; and read and approved the final manuscript. E. C. and A.
within their decision-making. As such, a HAT may only ever
Z. designed and conducted the statistical analysis for the trial and
be used in the clinical trial setting. This study did not set out
wrote and approved the final manuscript. J. R., P. M., M. J. R., S. Z.
to evaluate the relationship between the HAT score and other
and R. D. led recruitment at sites and read and approved the final
clinical outcome data (i.e. ICU admission, length of stay or
manuscript. All authors read and approved the final manuscript.
mortality), and these relationships would be worth exploring in
larger studies.
Validated bleeding assessment tools are available that have CONFLICT OF INTEREST
been designed to look at either the baseline bleeding risk for
The authors have no competing interests.

REFERENCES OBS2. British Journal of Anaesthesia, 119, results from a UK multi-centre, ran-
411–421. domised, double blind, placebo-controlled
Brenner, A., Arribas, M., Cuzick, J., Jairath,
V., Stanworth, S., Ker, K., Shakur-Still, H. CRASH-2 trial collaborators (2010) Effects pilot trial. Critical Care, 18, 164.
& Roberts, I. (2018) Outcome measures in of tranexamic acid on death, vascular Fabes, J., Brunskill, S.J., Curry, N., Doree,
clinical trials of treatments for acute severe occlusive events, and blood transfu- C. & Stanworth, S.J. (2018) Pro-coagulant
haemorrhage. Trials, 19, 533–537. https:// sion in trauma patients with significant haemostatic factors for the prevention and haemorrhage (CRASH-2): a randomised, treatment of bleeding in people without
Collins, P.W., Cannings-John, R., Bruynseels, placebo-controlled trial. Lancet, 376, haemophilia. Cochrane Database Systems
D. et al. (2017) Viscoelastometric-guided 23–32. Review, 12, CD010649.
early fibrinogen concentrate replacement Curry, N., Foley, C., Wong, H. et al. (2018) .1002/14651858
during postpartum haemorrhage: a dou- Early fibrinogen concentrate therapy for Holcomb, J. (2018) Transport time and preop-
ble blind randomised controlled trial: major haemorrhage in trauma (E-FIT-1): erating room haemostatic interventions are

© 2019 British Blood Transfusion Society Transfusion Medicine, 2019

6 N. Curry et al.

important: improving outcomes after severe Rahe-Meyer, N., Levy, J.H., Mazer, C.D. et al. systematic analysis. Lancet Global Health, 2,
truncal injury. Critical Care Medicine, 46, (2016) Randomized evaluation of fibrino- e323–e333.
447–453. gen vs placebo in complex cardiovascular Stanworth, S.J., Davenport, R., Curry, N. et al.
surgery (REPLACE): a double-blind phase (2016) Mortality from trauma haemor-
Holcomb, J., Tilley, B.C., Baraniuk, S. et al.
III study of haemostatic therapy. British rhage and opportunities for improvement
PROPPR Study Group. (2015) Transfusion
Journal of Anaesthesia, 117, 41–51. in transfusion practice. British Journal of
of plasma, platelets, and red blood cells in a Rodeghiero, F., Tosetto, A., Abshire, T. et al. Surgery, 103, 357–365.
1:1:1 vs a 1:1:2 ratio and mortality in patients (2010) ISTH/SSC bleeding assessment tool: WOMAN Trial Collaborators (2017) Effect
with severe trauma. The PROPPR random- a standardized questionnaire and a proposal of early tranexamic acid administration on
ized clinical trial. JAMA American Medical for a new bleeding score for inherited bleed- mortality, hysterectomy, and other mor-
Association, 313, 471–482. ing disorders. Journal of Thrombosis and bidities in women with post-partum
Kaufman, R., Djulbegovic, B., Gernsheimer, T. Haemostasis, 8, 2063–2065. Supplementary haemorrhage (WOMAN): an inter-
material to the official communication of national, randomised, double-blind,
et al. (2015) Platelet transfusion: a clinical
the SSC (last revision: 19 July 2011). URL. placebo-controlled trial. Lancet, 389,
practice guideline from the AABB. Annals of 2105–2116.
Internal Medicine, 162, 205–213. https://doi Say, L., Chou, D., Gemmill, A. et al. (2014)
.org/10.7326/M14-1589 Global causes of maternal death: a WHO

Transfusion Medicine, 2019 © 2019 British Blood Transfusion Society