The decision tool tells you what to do. This chapter is about why — why the protocol is triggered on the patient’s physiology rather than a number, why blood in a ratio and not clear fluid is what resuscitates the bleeding patient, why tranexamic acid is a life-saver in one bleeding patient and a hazard in the patient in the next bay, and the numbers that read as reassurance and are not. It is meant to be read before the shift, not during it.
NICE NG39, verbatim
BSH British Society for Haematology 2022, with its grade
RCT randomised trial
Teaching standard transfusion teaching, not guidance
Who this is about, and what it is not. This module is the general engine of the major haemorrhage protocol — activation, ratio, targets, TXA-by-setting, the triad, standing down. Your trust’s MHP, agreed with its transfusion laboratory, is the operational document you actually follow; the anatomical decisions (the
pelvis, the
chest) and the
reversal-agent dose are their own modules. Paediatric volumes are weight-based; use your paediatric MHP.
Two cases to hold in mind · illustrative, not real patients
A. A 30-year-old is brought in stabbed in the thigh, pale and tachycardic with a blood pressure of 88 systolic. A colleague hangs a litre of warmed crystalloid and asks the lab for “an urgent Hb before we activate anything.”
B. A 70-year-old man on warfarin is vomiting large volumes of blood. The registrar, reaching for the trauma reflex, draws up tranexamic acid and says “let’s wait for a formal cross-match before we give the O-negative.”
Two patients, two instincts, four errors between them. By the end of this chapter you should be able to name each one — and see that the trauma habits that save patient A are exactly what endanger patient B.
The whole protocol
Why we do what we do, in one table
Every step in the decision tool, and the reason it is there.
Chapter 1
The trigger is physiology, not a number
The protocol exists to be activated early — before the lab can tell you it was needed.
The commonest failure in major haemorrhage is not a wrong drug or a wrong ratio; it is activating too late, or waiting for a laboratory result that is already out of date by the time it returns. NICE is deliberately physiological: “use physiological criteria that include the patient’s haemodynamic status and their response to immediate volume resuscitation to activate the major haemorrhage protocol” NG39 1.5.13, and “do not rely on a haemorrhagic risk tool applied at a single time point.” NG39 1.5.14
BSH frames the same idea as a dynamic definition — the clinical status, the physiology, and the response to resuscitation, “e.g., heart rate >110 beats/min and/or systolic blood pressure <90 mm Hg.” BSH But it adds the warning that catches people out: “these physiological changes may be masked in some patient groups, e.g., the elderly or pregnancy.” The young patient compensates until they collapse; the older patient on a rate-limiting drug never mounts the tachycardia; the pregnant woman has a physiologically expanded volume that hides early loss. The numbers alone are not enough, which is exactly why NICE points to the response as much as the reading.
The first minutes, before the blood arrives
Take a group-and-screen ideally before the first unit, and give emergency group O while you wait — do not wait for a formal cross-match in a shocked bleeding patient. BSH Group O RhD-negative and K-negative red cells are prioritised for females of childbearing potential (aged <50) and where sex is unknown, to protect against future haemolytic disease of the newborn. And every patient wears an ID wristband: wrong-blood-in-tube is the commonest serious near-miss in emergency transfusion, and it kills through ABO-incompatible transfusion. Teaching
Chapter 2
Blood, not saline — and the ratio
Crystalloid fills the tank with fluid that cannot carry oxygen or form a clot.
Resuscitation in active bleeding is restrictive and blood-based. NICE: “for patients with active bleeding use a restrictive approach to volume resuscitation until definitive early control of bleeding has been achieved” NG39 1.5.18, and, bluntly, “in hospital settings do not use crystalloids for patients with active bleeding.” NG39 1.5.23 Crystalloid dilutes the clotting factors and platelets that are already being lost, drops the temperature, and raises the pressure enough to pop a clot that had formed — the “pop the clot” problem behind permissive hypotension.
In trauma the empirical ratio is balanced: 1 part plasma to 1 part red cells in adults NG39 1.5.24, which BSH echoes as “1:1 (not >1:2 ratio) with RBCs, until results from coagulation monitoring are available.” BSH Where lab results are not yet available in non-trauma bleeding, BSH sets a floor of at least 1:2 FFP:RBC, then guided by tests aiming to keep the PT ratio <1.5× normal. And the whole approach is a bridge, not a destination: “start with a fixed-ratio protocol… and change to a protocol guided by laboratory coagulation results at the earliest opportunity.” NG39 1.5.27
Two things the trial evidence has taught us to hold loosely. First, the 1:1 ratio comes from observational balanced-resuscitation data and the PROPPR trial, which did not show a mortality difference between 1:1:1 and 1:1:2 — it showed fewer deaths from exsanguination and more patients reaching haemostasis with the balanced ratio. PROPPR Second, once bleeding is controlled the target changes: BSH is explicit that “once bleeding is controlled, there is no indication to restore Hb to physiological levels” — the transfusion threshold is 70 g/l with a 70–90 g/l target range, not a normal haemoglobin.
Why GI bleeding is transfused more restrictively still
In upper GI bleeding a restrictive threshold (transfuse below 70 g/l) improved survival and re-bleeding versus a liberal one, and in variceal bleeding over-transfusion raises portal pressure and can worsen the bleed. Teaching The general restrictive principle is sharpest here — which is the opposite instinct to the trauma bay, and one reason the tool asks for the setting first.
Chapter 3
Tranexamic acid: the same drug, opposite answers
The single most setting-dependent decision in the protocol — and the one the trauma reflex gets wrong.
Tranexamic acid is the headline of the whole guideline, and the headline is conditional. BSH: “tranexamic acid is recommended for patients with presentations of major bleeding due to trauma and PPH, but not gastrointestinal bleeding” (grade 1A). BSH 1A Learn it as a fork:
- Trauma — yes, and early. 1 g IV over 10 minutes then 1 g over 8 hours, as soon as possible after injury and no later than 3 hours. BSH 1A NICE adds: not after 3 hours unless there is evidence of hyperfibrinolysis. NG39 1.5.5 The 3-hour line is not arbitrary — the CRASH-2 timing analysis found benefit within an hour, less between 1 and 3, and harm after 3 hours. CRASH-2
- Post-partum haemorrhage — yes. 1 g IV within 3 hours of onset; a second 1 g if bleeding continues after 30 minutes or restarts within 24 hours. BSH 1A This rests on WOMAN, which reduced death due to bleeding, most clearly when given early. WOMAN
- Surgery — yes, before incision. 1 g before skin incision for in-patient surgery. BSH
- Gastrointestinal bleeding — no. “Tranexamic acid is not recommended for patients with acute gastrointestinal bleeding” (grade 1A). BSH 1A The HALT-IT trial — 12,009 patients — found no reduction in death due to bleeding and a higher rate of venous thromboembolism. HALT-IT
This is the trap the module exists to prevent. “Bleeding badly → give TXA” is drilled into every trauma clinician, and it is right — in trauma. Carried unchanged into the GI bleeder, it is a 1A recommendation against what you are about to do. Before you draw it up, ask what is bleeding. BSH 1A
One more nuance worth carrying: a viscoelastic assay (ROTEM/TEG) that looks normal is not a reason to withhold TXA in trauma. BSH warns these tests are “less sensitive to measuring fibrinolytic activation in trauma and should not be used to withhold the use of TXA.” BSH The trauma indication is empirical and time-driven, not test-gated.
Chapter 4
The targets: fibrinogen, platelets, PT
The components that run out, and the numbers that say replace them.
Once the fixed-ratio phase is running, the protocol becomes goal-directed on repeat testing “every 30–60 min depending on the severity of the haemorrhage.” BSH The three numbers to hold:
- Fibrinogen. Replace below 1.5 g/l in the non-pregnant patient — with cryoprecipitate or fibrinogen concentrate, because FFP cannot raise a fibrinogen below 1 g/l fast enough. BSH In PPH the trigger is higher, <2.0 g/l, because pregnancy runs fibrinogen at 4–6 g/l at term, so a “normal-looking” 1.8 is already low for a mother. BSH
- Platelets. Maintain >50 ×10⁵/l, with a higher target (many protocols use >100) for intracranial or spinal bleeding, or in an actively bleeding patient with a falling count. BSH
- PT ratio. Further FFP aiming to keep the PT ratio <1.5× mean normal. BSH
Use the Clauss fibrinogen assay, not a PT-derived fibrinogen, which can read falsely high and reassure you wrongly. As a dosing rule of thumb, two five-donor pools of cryoprecipitate, or 4–5 g of fibrinogen concentrate, raise an adult’s fibrinogen by roughly 1 g/l. BSH
Why “give more fibrinogen, earlier” is not automatically better
It is tempting to assume that if replacing fibrinogen when it falls helps, giving it empirically and early to everyone must help more. The CRYOSTAT-2 trial tested exactly that — early, empirical, high-dose cryoprecipitate for all trauma haemorrhage — and found no improvement in 28-day mortality over standard care. CRYOSTAT-2 The lesson is goal-directed, not maximal: replace what the tests show is missing, on the targets above, rather than front-loading blindly.
Chapter 5
Reversal, and the agents that do not help
Reverse the anticoagulated bleeder — and resist the reflexes that do harm.
The rule is bleeding-gated. “Rapidly reverse anticoagulation in patients who have major trauma with haemorrhage” NG39 1.5.6, but “do not reverse anticoagulation in patients who do not have active or suspected bleeding.” NG39 1.5.12 Reversing a non-bleeding patient trades a theoretical benefit for a real thrombotic risk.
- Warfarin (VKA), adult, bleeding: prothrombin complex concentrate immediately — not plasma. NG39 1.5.8–1.5.9
- Any other anticoagulant (a DOAC), or any anticoagulant in a child: consult a haematologist immediately. NG39 1.5.10–1.5.11 The specific agents (idarucizumab, andexanet) and the PCC dose are in the reversal module.
Two agents to withhold, not reach for:
- Platelets to “cover” an antiplatelet in an intracranial bleed. The PATCH trial found platelet transfusion increased death or dependence in spontaneous intracerebral haemorrhage on antiplatelet therapy. PATCH BSH notes it is unclear how far this applies to traumatic ICH — so this is a reason for caution and a haematology/neurosurgery discussion, not a reflex transfusion.
- Desmopressin, recombinant factor VIIa, aprotinin. “The use of desmopressin, rVIIa or aprotinin is not recommended in the management of major haemorrhage unless as part of a clinical trial” (grade 1B). BSH 1B
Chapter 6
The lethal triad: warmth, acid, calcium
Three things that stop blood clotting, all preventable in the resus room.
BSH names the enemy plainly: “hypothermia, acidosis and hypocalcaemia will further worsen coagulation.” BSH Each has a resus-room answer.
- Warmth. “Minimise ongoing heat loss in patients with major trauma.” NG39 1.6.1 Transfuse through a blood warmer; the exposed, resuscitated patient is the one who gets cold, and cold blood does not clot.
- Calcium. Citrate in stored blood chelates calcium, so massive transfusion drives ionised calcium down and impairs clot formation and cardiac contractility. Monitor and replace. Note what BSH does not do: it sets no numeric calcium threshold — it says calcium should be “monitored and supplemented as appropriate”, so keep ionised calcium in the normal range per your local protocol and check it early and often. Do not invent a number and attribute it to BSH. BSH
- Acidosis. Largely corrected by restoring perfusion — stopping the bleeding and transfusing — not by giving bicarbonate. Teaching
Chapter 7
Standing it down
The protocol is not finished until it has been switched off.
Deactivation is a recommendation in its own right, not an afterthought: “deactivation of the MHP is also important, as delays in standing down the MHP will lead to blood wastage and prevent resumption of other laboratory services.” Frontline staff should “know when to activate and deactivate the local MHP” (grade 2B). BSH 2B
So when bleeding is controlled: stand down explicitly, tell the transfusion laboratory, return unused components, and convert the patient’s emergency or major-incident identifier back to their routine hospital identifier. And remember what the whole protocol was buying time for — source control by the surgeon, interventional radiologist, endoscopist or obstetrician. The transfusion is supportive; the definitive treatment is the one that stops the bleeding. Make sure it is happening.
Pitfalls
Ten ways the major haemorrhage protocol goes wrong
Each of these happens in good departments, usually under pressure or out of habit.
01Waiting for a lab result to activate
- Why it happens
- A number feels more defensible than a judgement.
- Instead
- Activate on physiology and response to volume; a single-timepoint value can be normal in a patient who is exsanguinating. NG39 1.5.13–1.5.14
02Crystalloid to chase the blood pressure
- Why it happens
- Saline is on the shelf and blood takes a phone call.
- Instead
- No crystalloid for active bleeding in hospital; restrictive, blood-based resuscitation. NG39 1.5.18, 1.5.23
03Giving TXA in a GI bleed
- Why it happens
- The trauma reflex fires whenever someone is bleeding heavily.
- Instead
- TXA is not recommended in GI bleeding — no benefit, more VTE. Ask what is bleeding first. BSH 1A HALT-IT
04Waiting for a cross-match in a shocked patient
- Why it happens
- Group-specific blood feels safer than emergency O.
- Instead
- Give emergency group O now; take the group-and-screen ideally before the first unit but do not wait on it. BSH
05Treating a “normal” fibrinogen in a mother as fine
- Why it happens
- 1.8 g/l looks reassuring against the usual 1.5 trigger.
- Instead
- Pregnancy runs 4–6 g/l; replace below 2.0 in PPH. BSH
06Transfusing platelets to cover an antiplatelet ICH
- Why it happens
- It feels like correcting the defect.
- Instead
- PATCH found platelets worsened outcome in spontaneous ICH on antiplatelets; discuss, do not reflex. PATCH
07Forgetting the calcium
- Why it happens
- It is not in the pack and there is no single BSH number to prompt it.
- Instead
- Citrate drives ionised calcium down; monitor and replace to your local target, early and often. BSH
08Letting the patient get cold
- Why it happens
- Exposure for assessment, cold fluids, a long stay in resus.
- Instead
- Warm the blood and the patient; hypothermia is a preventable arm of the lethal triad. NG39 1.6.1
09Reversing a raised INR that is not bleeding
- Why it happens
- An abnormal number invites correction.
- Instead
- Do not reverse a patient without active or suspected bleeding. NG39 1.5.12
10Never standing it down
- Why it happens
- The crisis passes, the team disperses, and no one closes the loop.
- Instead
- Deactivate explicitly, tell the lab, return components — and confirm source control is under way. BSH 2B
Red herrings
Seven numbers that look reassuring and are not
None of these is wrong information. Each answers a different question from the one you are asking.
Looks reassuring
“The blood pressure is 110.”
Young patients hold their pressure until they crash, and the older or pregnant patient can be masked. Judge the response to volume, not one reading. NG39 1.5.13
Looks reassuring
“The first haemoglobin is 130.”
Haemoglobin lags acute blood loss — it has not had time to fall before dilution. A normal early Hb does not exclude major haemorrhage. Teaching
Looks reassuring
“The ROTEM looks normal, so no TXA.”
Viscoelastic tests are insensitive to fibrinolysis in trauma and “should not be used to withhold the use of TXA.” The trauma indication is empirical. BSH
Looks reassuring
“Fibrinogen 1.8 — above the 1.5 trigger.”
In a post-partum mother the target is 2.0, and normal pregnancy runs 4–6. 1.8 is already low for her. BSH
Looks reassuring
“We gave a big empirical dose of cryo early.”
CRYOSTAT-2 showed early empirical high-dose cryoprecipitate did not improve mortality. Replace to target, not maximally. CRYOSTAT-2
Looks reassuring
“Calcium was normal on the gas.”
Ionised calcium falls as citrate load accumulates through the transfusion. One early value does not cover the next ten units. BSH
Looks reassuring
“Bleeding’s slowing — the protocol worked.”
Transfusion buys time; it does not fix the source. Slowing is not control until the surgeon, IR, endoscopist or obstetrician has acted. BSH
The cases, revisited
The four errors, named
- Case A — the crystalloid. A litre of saline in an actively bleeding patient dilutes clotting factors, cools the patient and can pop a forming clot. In hospital, NICE says no crystalloid for active bleeding — give blood, restrictively. (Chapter 2)
- Case A — “an Hb before we activate.” Activation is on physiology and response to volume, not a laboratory value at one moment; the shocked, tachycardic stab is an activation now, and the Hb may be normal anyway. (Chapter 1)
- Case B — the tranexamic acid. This is the trauma reflex in the wrong patient: TXA is a 1A recommendation against use in GI bleeding. Do not give it. (Chapter 3)
- Case B — waiting for a cross-match. The exsanguinating patient gets emergency group O now; the formal cross-match is not a gate. (Chapter 1)
The thread through all four: the same habits that resuscitate a trauma patient correctly — press on, give TXA, act fast — are, applied without asking what is bleeding, exactly how the GI bleeder is harmed. The setting is the first question for a reason.