New-Onset Bleeding in the Critically Ill: A Clinician's Masterclass

A Review Article for Postgraduate Trainees and Practicing Consultants in Internal Medicine and Critical Care

Dr Neeraj Manikath , claude.ai

Keywords: coagulopathy, ICU bleeding, disseminated intravascular coagulation, thrombocytopenia, haemostasis, transfusion

Abstract

New-onset bleeding in the intensive care unit (ICU) is simultaneously one of the most common and most treacherous clinical problems encountered by the intensivist. It carries a mortality premium that ranges from 10% in minor mucocutaneous haemorrhage to over 50% in the setting of massive transfusion. Yet, the majority of preventable deaths from ICU haemorrhage arise not from a failure of pharmacological rescue, but from a failure of systematic thinking — misidentifying the mechanism, misjudging the urgency, or mistreating a coagulopathy that was never correctly characterised. This review distils three decades of clinical evidence, bedside wisdom, and state-of-the-art haemostatic science into an actionable, mechanisms-based framework for the clinician at the bedside.

1. Introduction: The Patient Who Should Not Have Bled

Case Vignette: A 58-year-old woman with septic shock secondary to community-acquired pneumonia is on Day 4 of her ICU admission. She is mechanically ventilated, receiving noradrenaline at 0.2 mcg/kg/min, broad-spectrum antibiotics, and prophylactic low-molecular-weight heparin. The nursing staff call you at 0200 h because blood is oozing from her endotracheal tube, her arterial line site, and — most alarmingly — the edges of her peripheral IV. Her platelet count yesterday was 110 × 10⁹/L. Today it is 44 × 10⁹/L. Her PT is 22 seconds. Her fibrinogen is 0.9 g/L. She is not on any anticoagulation beyond prophylaxis. What killed her coagulation system overnight?

This vignette — encountered in every busy ICU — represents the quintessential haemostatic catastrophe: consumption coagulopathy driven by uncontrolled sepsis, layered atop a baseline of hepatic dysfunction, nutritional depletion, and iatrogenic factor dilution. It is a story about trajectory, not snapshot. The clinician who sees only today's numbers without understanding yesterday's direction will always be one step behind.

Haemorrhage in the ICU is not a single disease. It is a final common pathway reached by at least a dozen distinct pathophysiological routes: platelet dysfunction from uraemia, factor depletion from dilution, fibrinolysis from hypoperfusion, vitamin K deficiency from malnutrition and antibiotics, heparin accumulation in renal failure, or the rare but devastating heparin-induced thrombocytopaenia with thrombosis (HITT) that paradoxically causes both clotting and bleeding. The clinician who approaches all ICU bleeding with the same empirical formula — "give FFP and platelets" — is practising haematology by reflex rather than by reason.

Epidemiologically, the stakes are unambiguous. Clinically important bleeding — defined as bleeding that causes haemodynamic compromise, requires unplanned transfusion, or leads to procedural interruption — occurs in approximately 5–10% of all ICU admissions. Among patients with sepsis-associated coagulopathy (SAC) or overt DIC, this figure rises to 15–25%. The mortality attributable to haemorrhagic complications, independent of the underlying illness, adds a relative risk of 1.3–1.7 across most ICU cohorts. Yet, haemostatic interventions are among the most frequently misapplied therapies in critical care — transfusion thresholds are too liberal, factor concentrates are underused, and point-of-care viscoelastic testing (VET) remains strikingly underutilised outside of cardiac surgery and trauma centres.

This review will guide you through the architecture of ICU coagulopathy: its mechanisms, its mimics, its mastery.

2. Pathophysiology: The Haemostatic System Under Siege

Understanding bleeding in the ICU requires abandoning the antiquated "cascade model" of coagulation and embracing the cell-based model of haemostasis — a framework far more relevant to what actually happens at the endothelial surface during critical illness.

2.1 The Normal Haemostatic Architecture

Haemostasis proceeds in three overlapping phases:

Primary haemostasis — Platelet adhesion, activation, and aggregation at the site of vascular injury, forming the primary platelet plug. Governed by von Willebrand factor (vWF), glycoprotein Ib-IX-V, and GPIIb/IIIa.
Secondary haemostasis — The coagulation cascade amplifies thrombin generation on the platelet surface, converting fibrinogen to fibrin and stabilising the plug. Tissue factor (TF) on activated monocytes and subendothelial cells initiates extrinsic pathway activation.
Fibrinolysis — Plasmin, generated from plasminogen by tPA, dissolves the clot once healing has begun. Regulated by PAI-1 and α2-antiplasmin.

2.2 How Critical Illness Disrupts Each Phase

Phase	ICU Disruption	Clinical Consequence
Primary haemostasis	Uraemic platelet dysfunction, thrombocytopaenia, vWF cleavage by ADAMTS13	Mucocutaneous bleeding, ooze from puncture sites
Secondary haemostasis	Factor dilution, consumption (DIC), liver failure, vitamin K depletion	Prolonged PT/APTT, deep tissue haemorrhage
Fibrinolysis	Hyperfibrinolysis in trauma/liver disease, hypofibrinolysis in sepsis	Uncontrollable surgical bleeding OR paradoxical thrombosis
Endothelium	Loss of thrombomodulin, TF overexpression, glycocalyx shedding	Microvascular thrombosis + haemorrhage simultaneously

2.3 The Sepsis–Coagulation Axis: A Bidirectional Catastrophe

In sepsis, the haemostatic system fails in both directions simultaneously — this is the mechanistic heart of why sepsis-associated coagulopathy (SAC) is so lethal. Tumour necrosis factor-α (TNF-α) and interleukin-6 (IL-6) drive TF expression on monocytes and endothelial cells → thrombin burst → fibrin deposition in microvasculature → organ failure. Simultaneously, natural anticoagulants — protein C, antithrombin, TFPI — are consumed, downregulated, or cleaved by neutrophil elastase. The result is a system simultaneously "all-on and all-out."

🔑 Key Mechanistic Insight: In sepsis, fibrinogen is consumed BEFORE platelets and factors. A falling fibrinogen in a septic patient is therefore an early warning sign, often preceding overt DIC by 12–24 hours. Serial fibrinogen monitoring is more sensitive than the PT or platelet count for detecting early consumption coagulopathy.

2.4 Dilutional Coagulopathy: The Resuscitation Tax

Every litre of crystalloid administered to a bleeding or shocked patient dilutes clotting factors and platelets. After 1.5 litres of normal saline, the PT begins to prolong. After 2–3 litres, clinically significant coagulopathy sets in. This is not an emergency complication; it is the expected physiological consequence of volume resuscitation. The intensivist who does not anticipate this is perpetually surprised by it.

Modern damage control resuscitation protocols use 1:1:1 ratios of packed red cells: fresh frozen plasma: platelets — precisely to combat dilutional coagulopathy. Yet this paradigm, developed in trauma, is frequently not applied to medical ICU patients with GI bleeding, coagulopathy of liver disease, or post-cardiac surgery haemorrhage, where the same physics applies.

3. Diagnostic Nuances: Separating Good from Great Clinicians

3.1 The Clinical History That Most Residents Miss

When confronted with a bleeding ICU patient, the master clinician asks three questions that residents routinely skip:

a) What is the bleeding pattern?

Mucocutaneous (petechiae, gingival ooze, epistaxis, purpura) → platelet problem (quantitative or qualitative)
Deep tissue (haematomas, haemarthroses, retroperitoneal collections) → factor deficiency
Diffuse oozing from ALL puncture sites and surgical wounds → DIC until proven otherwise
Isolated site bleeding → local cause (wrong vessel puncture, inadequate pressure) — do not over-medicalise

b) What changed in the last 24–48 hours? A sudden drop in haemoglobin without overt external bleeding is occult haemorrhage — retroperitoneal, intrathoracic, or intra-abdominal — until proven otherwise. Do not accept "the lab must have haemolysed the sample."

c) What drugs was the patient given in the last 72 hours?

Antibiotics → vitamin K depletion (especially cephalosporins, fluoroquinolones)
Proton pump inhibitors → rarely relevant (good for prophylaxis)
Heparin — was the dose adjusted for renal function? Anti-Xa level is NOT routinely checked and accumulation is common in AKI
NSAIDs, aspirin — often omitted from ICU medication reconciliation
Azole antifungals → dramatically potentiate warfarin via CYP2C9 inhibition

3.2 The Examination That Tells the Story

Beyond the bleeding site, the systematic examination should specifically document:

Skin: purpura, petechiae, ecchymoses disproportionate to minor trauma
Mucous membranes: gingival ooze, epistaxis → primary haemostatic failure
Joints and muscle compartments: feel for tense haematomas
Surgical drains: character of output — fresh blood (arterial problem), altered blood (venous ooze), or mixed (DIC pattern)
Catheter sites: oozing from ALL sites simultaneously is the DIC signature sign
Fundoscopy: retinal haemorrhages in thrombotic microangiopathy (TMA) or extreme thrombocytopaenia
Abdomen: guarding/rigidity in a coagulopathic patient with dropping Hb = retroperitoneal/visceral haemorrhage until proven otherwise

3.3 Interpreting the Laboratory: The Tests Behind the Tests

🔬 Diagnostic Pearl: The standard coagulation screen (PT, APTT, platelet count, fibrinogen) tells you about plasma, not about whole blood clot formation. These tests are performed at 37°C in ideal laboratory conditions — far removed from the acidaemic, hypothermic, hypocalcaemic milieu of the bleeding ICU patient. A "normal" PT in a patient who is pH 7.2 and temperature 34°C does NOT mean normal coagulation.

The Lethal Triad (acidosis + hypothermia + coagulopathy) mutually amplify each other. Enzymes of the coagulation cascade lose approximately 10% of their activity for every 1°C fall below 37°C. At 33°C, factor activity is reduced by ~40%.

Interpreting the APTT:

Prolonged APTT alone: heparin effect, factor VIII/IX deficiency (haemophilia), lupus anticoagulant
Prolonged PT alone: early warfarin effect, isolated factor VII deficiency, mild liver disease
Both prolonged: DIC, severe liver disease, massive transfusion, supratherapeutic direct oral anticoagulants (DOACs)
Both normal but patient still bleeding: platelet dysfunction (uraemia, aspirin), hyperfibrinolysis, vascular cause

Fibrinogen: The Underappreciated Vital Sign of Haemostasis Fibrinogen is an acute phase reactant. A "normal" fibrinogen of 2.0 g/L in a systemically inflamed patient who would be expected to have a level of 4–5 g/L represents functional fibrinogen deficiency. This is the most commonly missed laboratory nuance in ICU haemostasis. Target fibrinogen >2.0 g/L for prophylaxis, >2.5 g/L when actively bleeding.

The D-dimer trap: An elevated D-dimer in an ICU patient is almost always present and is diagnostically unhelpful in isolation. D-dimer confirms fibrinolysis has occurred — not DIC, not PE, not DVT specifically. Use it in context, not in isolation.

3.4 Viscoelastic Testing: The Revolution at the Bedside

Thromboelastography (TEG) and rotational thromboelastometry (ROTEM) have transformed haemostatic management in trauma and cardiac surgery by providing a whole-blood, real-time picture of clot formation, strength, and lysis within 15–20 minutes. Key parameters:

Parameter	TEG Equivalent	What It Tells You
Clot Initiation Time (CT/R)	R-time	Prolonged = factor deficiency or anticoagulants → give FFP/PCC
Clot Formation Time (CFT/K)	K-time	Prolonged = fibrinogen/platelet deficiency → give cryoprecipitate
Alpha angle (α)	α angle	Reduced = fibrinogen deficiency → give cryoprecipitate/fibrinogen concentrate
Maximum Amplitude (MA/MCF)	MA	Reduced = platelet dysfunction or deficiency → give platelets
Lysis Index (LY30/ML)	LY30	Elevated = hyperfibrinolysis → give tranexamic acid IMMEDIATELY

⚡ Clinical Hack: In centres without VET, a surrogate for hyperfibrinolysis is the "clot observation test" — mix 2 mL patient blood with 0.1 mL thrombin on a glass slide. If the formed clot lyses within 30–60 minutes, significant fibrinolysis is present. Crude, but life-saving when VET is unavailable.

4. Clinical Pearls 🪙

Pearl 1: The Direction of Travel Matters More Than the Number A platelet count of 80 × 10⁹/L is reassuring if it was 40 × 10⁹/L yesterday. It is alarming if it was 220 × 10⁹/L three days ago. ICU haematology is about vectors, not points. Review the trend, not just the current value.

Pearl 2: The Most Common Cause of Thrombocytopaenia in the ICU Is NOT What You Think Most residents reach for "heparin" or "immune thrombocytopaenia" when they see a falling platelet count. The truth? Sepsis is the commonest cause of thrombocytopaenia in the ICU, accounting for 40–60% of cases in most series. Sepsis causes bone marrow suppression, splenic sequestration, platelet consumption by microvascular thrombi, and immune-mediated destruction — all simultaneously.

Pearl 3: You Cannot Correct Coagulopathy in a Bleeding Patient Who Is Still Shocked Haemostatic therapies are physiologically futile in the presence of ongoing hypoperfusion. Lactic acidosis of 8 mmol/L will overwhelm any dose of FFP you give. Fix the shock first. The sequence is: restore perfusion → restore temperature → correct acidosis → then correct coagulopathy.

Pearl 4: The INR Does Not Predict Bleeding Risk in Liver Disease The INR was developed to monitor warfarin therapy — it measures only the procoagulant arm of the coagulation system. In cirrhosis, both procoagulants AND anticoagulants are reduced, often in balance. A patient with cirrhosis and INR 2.5 may have near-normal overall haemostasis (balanced coagulopathy) and carries a far lower bleeding risk than an INR 2.5 warfarin patient. This is why "correcting the INR" with FFP before procedures in cirrhotic patients is largely an evidence-free ritual.

Pearl 5: Protamine for Heparin — Dose It Precisely or Don't Use It Excess protamine is itself anticoagulant — it inhibits platelet function and can cause paradoxical bleeding. The dose is 1 mg protamine per 100 units of unfractionated heparin administered in the preceding 4 hours. Do not give more. Do not give it empirically without knowing the heparin dose.

Pearl 6: Recombinant Factor VIIa is a Last Resort, Not a Rescue Drug rFVIIa (NovoSeven) has been aggressively marketed and irrationally used in ICU bleeding. Outside of haemophilia with inhibitors and a few surgical contexts, its evidence base for ICU bleeding is weak. It increases thromboembolic events by ~10% and does not reduce mortality. Reserve it for truly refractory life-threatening haemorrhage where all other options have been exhausted.

5. Oysters 🦪 — Hidden Gems Most Clinicians Miss

Oyster 1: Acquired von Willebrand Syndrome in Critical Illness Continuous-flow LVAD patients, severe aortic stenosis, and ECMO circuits destroy high-molecular-weight vWF multimers through shear stress. The result is an acquired von Willebrand syndrome (AVWS) that perfectly mimics platelet dysfunction. These patients bleed from mucocutaneous sites (gut, nose, skin), have a prolonged PFA-100 but normal vWF antigen, and do not respond to platelets or FFP. Treatment is desmopressin (DDAVP), vWF concentrate, or removal of the offending shear force. This diagnosis is missed in virtually every ICU that doesn't specifically look for it.

Oyster 2: Supratherapeutic Heparin Effect from Renal Failure — The Silent Accumulator Low-molecular-weight heparins (LMWH) are renally cleared. In a patient with AKI, prophylactic enoxaparin 40 mg once daily may accumulate to therapeutic or supratherapeutic anti-Xa levels within 3–5 days. Routine anti-Xa monitoring is not standard of care in most units — yet the haemorrhagic consequences are severe. In any ICU patient with AKI and unexplained bleeding on prophylactic LMWH, check the anti-Xa level (target: 0.2–0.4 IU/mL for prophylaxis). Consider switching to UFH (renal failure-safe, reversed by protamine) or fondaparinux avoidance in eGFR <30 mL/min.

Oyster 3: Hypofibrinolysis — The Other Side of the DIC Coin While hyperfibrinolysis is well-recognised, the opposite state — hypofibrinolysis — is equally deadly and equally ignored. In late-phase sepsis, PAI-1 levels rise dramatically, shutting down fibrinolysis and causing microvascular fibrin deposition. These patients develop multi-organ failure not from bleeding, but from microvascular thrombosis. Paradoxically, their coagulation tests may be "improving" (fibrinogen rising, PT shortening) while their kidneys and liver are being destroyed by microthrombi. TEG/ROTEM with fibrinolytic profiles can detect this; standard coagulation tests cannot.

Oyster 4: The Platelet Count Does Not Tell You What Platelets Are Doing A platelet count of 150 × 10⁹/L in a patient with uraemia, post-cardiopulmonary bypass, or on aspirin/clopidogrel may be haemostatically equivalent to a count of 30 × 10⁹/L in a normal individual. Conversely, patients with essential thrombocythaemia may have counts of 1000 × 10⁹/L with completely dysfunctional platelets. PFA-100 closure time or TEG maximum amplitude (MA) are far better surrogates of platelet function than the count itself.

Oyster 5: Citrate Anticoagulation in CRRT Can Cause Systemic Hypocalcaemia and Bleeding Patients on continuous renal replacement therapy (CRRT) using regional citrate anticoagulation (RCA) are at risk of citrate accumulation — particularly in liver failure, where citrate metabolism is impaired. Citrate chelates ionised calcium. Systemic hypocalcaemia (ionised Ca²⁺ <1.0 mmol/L) causes platelet dysfunction and impaired coagulation factor activity — a perfectly correctable cause of ICU bleeding that is found only if the ionised calcium is actually measured. The clue: a high total calcium with low ionised calcium ratio (>2.5) is pathognomonic of citrate accumulation.

Oyster 6: Stress-Dose Steroids and Adrenal Insufficiency Can Masquerade as Coagulopathy Relative adrenal insufficiency in septic shock causes capillary fragility and poor vascular responsiveness to haemostatic therapies. A patient who bleeds refractory to all haemostatic interventions but whose bleeding resolves dramatically after hydrocortisone 200 mg/day has adrenal insufficiency until proven otherwise. Vasopressin 0.04 units/min has a direct haemostatic effect at the V1 receptor on vascular smooth muscle — use it not just for vasopressor support but as a haemostatic adjunct.

6. Clinical Hacks & Tips ⚡

Hack 1: The "Fibrinogen First" Rule In any actively bleeding ICU patient, give fibrinogen concentrate (4 g IV) or cryoprecipitate (10 units) before anything else — before FFP, before platelets. Fibrinogen is the substrate of clot formation. Without it, nothing else works. This is the single most evidence-based shift in contemporary haemostatic resuscitation.

Hack 2: The 4T Score — Use It Every Time You See a Dropping Platelet Count on Heparin HIT (heparin-induced thrombocytopaenia) occurs in 0.5–5% of patients on UFH and is catastrophic if missed. Calculate the 4T score mentally at every encounter with unexplained thrombocytopaenia:

Component	2 points	1 point	0 points
Thrombocytopaenia	>50% fall to nadir ≥20	30–50% fall OR nadir 10–19	<30% fall OR nadir <10
Timing of fall	Days 5–10 or ≤1 day if prior heparin exposure	>10 days OR timing unclear	<4 days without prior exposure
Thrombosis or other sequelae	New thrombosis, skin necrosis	Progressive/recurrent thrombosis	None
oTher cause of thrombocytopaenia	None apparent	Possible	Definite other cause

Score ≥6 = high probability HIT → stop ALL heparin including flushes → start argatroban or fondaparinux → send anti-PF4 antibody assay.

⚡ Critical Hack: HIT patients MUST NOT receive warfarin until the platelet count has recovered to >150 × 10⁹/L. Starting warfarin in active HIT causes catastrophic skin necrosis from protein C depletion. This is a high-stakes pitfall that kills patients.

Hack 3: The MTP Activation Trigger — Don't Wait for Lab Results In massive haemorrhage (>1 blood volume in 24 hours, or >50% blood volume in 3 hours), activate the Massive Transfusion Protocol (MTP) based on clinical assessment — NOT laboratory values. By the time the labs come back, the patient is further behind. Use the ABC Score (Assessment of Blood Consumption): ≥2 of: penetrating mechanism, systolic BP ≤90 mmHg, HR ≥120 bpm, positive FAST — trigger MTP immediately.

Hack 4: Four-Factor PCC (Prothrombin Complex Concentrate) Beats FFP in Urgent Reversal For warfarin reversal in life-threatening haemorrhage, 4F-PCC (Beriplex, Octaplex) achieves complete reversal within 15 minutes, versus 6–12 hours for vitamin K alone and 30–60 minutes for FFP. The dose is 25–50 IU/kg depending on INR. It requires no blood group matching, no thawing, and delivers a tiny volume (20–40 mL vs 1–1.5 L for FFP). In 2024, 4F-PCC is the standard of care for urgent anticoagulation reversal — yet FFP remains the default in many centres, purely out of inertia.

Hack 5: Tranexamic Acid — Timing Is Everything The evidence from CRASH-2 (trauma) and WOMAN (postpartum haemorrhage) is unambiguous: tranexamic acid (TXA) saves lives when given within 3 hours of haemorrhage onset. Beyond 3 hours, it may increase the risk of thromboembolism without mortality benefit. TXA dose: 1 g IV over 10 minutes, repeated after 30 minutes if bleeding continues. In ICU patients with suspected hyperfibrinolysis (TEG LY30 >7.5%), give TXA regardless of aetiology.

Hack 6: The Calcium Imperative Ionised hypocalcaemia is the most reversible, most ignored haemostatic defect in massive transfusion. Citrate in blood products chelates ionised calcium. After 4–6 units of packed red cells, most patients develop clinically significant hypocalcaemia. Supplement with 10 mL 10% calcium gluconate IV after every 4 units of blood product, or target ionised Ca²⁺ >1.1 mmol/L. This is not optional — it is a core component of damage control resuscitation.

7. The Differential Diagnosis Framework: A Structured Approach

When you face a bleeding ICU patient, run through this mental checklist in order:

Step 1 — Is This a LOCAL or SYSTEMIC Problem?

Oozing from a single site (drain, wound, catheter) is usually local — surgical haemostasis issue. Bleeding from multiple sites simultaneously (arterial line AND ET tube AND peripheral IV AND urine) is systemic coagulopathy. Do not treat local bleeding with systemic haemostatic agents.

Step 2 — What Is the PLATELET Situation?

Count <10 × 10⁹/L → transfuse regardless of bleeding status (spontaneous intracranial bleeding risk)
Count 10–50 × 10⁹/L + active bleeding → transfuse to >50 × 10⁹/L
Count >50 × 10⁹/L + bleeding → platelet function, not count, is the problem (uraemia, drugs, AVWS)

Step 3 — What Is the COAGULATION Factor Situation?

INR >1.5 + APTT >1.5× normal + active bleeding → give fibrinogen concentrate FIRST, then FFP or 4F-PCC
Isolated APTT prolongation → heparin effect? Factor deficiency? Lupus anticoagulant?
Isolated PT prolongation → early liver disease, warfarin, vitamin K deficiency

Step 4 — Is There FIBRINOLYSIS?

Clinical: wounds ooze blood that won't clot; clot forms and dissolves
TEG: LY30 >7.5% or CL30 >15%
Lab: rapidly falling fibrinogen despite fibrinogen replacement
Treatment: TXA 1 g IV immediately

Step 5 — Is There a REVERSIBLE DRUG or TOXIN Cause?

Warfarin → Vitamin K 10 mg IV + 4F-PCC
UFH → Protamine (1 mg per 100 units)
LMWH → Protamine (partial reversal, 60–80%)
DOACs → Idarucizumab (dabigatran), Andexanet alfa (factor Xa inhibitors)
Thrombolytics → TXA + cryoprecipitate + FFP

8. Management Intricacies: Drug Choices, Doses, Timing, Pitfalls

8.1 Blood Products — The Evidence Has Shifted

Fresh Frozen Plasma (FFP):

Contains ALL coagulation factors at approximately 1 unit/mL concentration
Dose: 15–20 mL/kg for significant coagulopathy (typically 4–6 units)
Pitfall: Contains citrate and must be ABO compatible. Takes 20–30 minutes to thaw. Volume overload in cardiac failure. Poor choice for urgent reversal.
Modern indication: Primarily within 1:1:1 massive transfusion protocol or when specific concentrates are unavailable.

Cryoprecipitate:

Rich in fibrinogen (10× concentration of FFP), factor VIII, vWF, factor XIII
Dose: 10 units (raises fibrinogen by ~1 g/L in a 70 kg adult)
When to use: Fibrinogen <2.0 g/L with active bleeding; fibrinogen <1.5 g/L prophylactically in high-risk patients.
Better alternative: Fibrinogen concentrate (Haemocomplettan, RiaSTAP) — 4 g raises fibrinogen by ~1 g/L with smaller volume, no thawing, pathogen-reduced.

Platelet Concentrates:

1 adult therapeutic dose (ATD) raises count by ~30 × 10⁹/L in a non-refractory patient
Pitfalls: ABO/Rh compatibility matters. Refractoriness (failure to increment) suggests HLA antibodies, consumption, or splenic sequestration. CMV-negative products for immunocompromised patients.
Do not transfuse solely for a low platelet count without bleeding in sepsis (TOPIC trial: no benefit, possible harm).

8.2 Haemostatic Agents — Indications Clarified

Tranexamic Acid (TXA):

Mechanism: competitive inhibitor of plasminogen → antifibrinolytic
Dose: 1 g IV over 10 min, may repeat once after 30 min
Evidence: CRASH-2, CRASH-3, WOMAN trials — proven mortality benefit in trauma, PPH; reasonable evidence for GI bleeding
Caution: Avoid if history of seizures (high doses lower seizure threshold). Contraindicated in haematuria from upper tract source (ureteric obstruction risk).

Desmopressin (DDAVP):

Mechanism: releases vWF and factor VIII from endothelial Weibel-Palade bodies
Dose: 0.3 mcg/kg IV over 30 min
When to use: Platelet dysfunction (uraemia, aspirin effect, post-bypass), type 1 vWD, mild haemophilia A
Pitfall: Tachyphylaxis occurs after 2–3 doses (stores depleted). Causes dilutional hyponatraemia — restrict free water for 24 h after each dose. Avoid in cardiovascular disease.

4-Factor Prothrombin Complex Concentrate (4F-PCC):

Contains factors II, VII, IX, X + proteins C and S
Dose: 25–50 IU/kg (INR-guided), maximum 5000 IU
Advantage: Rapid, small volume, no blood group matching, no thawing
Caution: Risk of thrombosis (particularly in HIT, HITT, or hypercoagulable states). Do not use for prophylactic INR "correction."

Recombinant Factor VIIa (rFVIIa):

Dose: 90 mcg/kg IV (haemophilia); 20–30 mcg/kg in "off-label" ICU use
Narrow indications: Haemophilia A/B with inhibitors, acquired haemophilia A, Glanzmann thrombasthenia, refractory post-partum haemorrhage
Evidence gaps: Does NOT improve mortality in ICU bleeding outside these indications; increases arterial thromboembolism by ~10% (RR 1.45, 95% CI 1.02–2.05 in meta-analyses)

8.3 Reversal Agents for Anticoagulants

Drug	Reversal Agent	Dose	Onset
Warfarin (urgent)	4F-PCC + Vitamin K 10 mg IV	25–50 IU/kg	15 min
Warfarin (non-urgent)	Vitamin K 5–10 mg IV		6–12 h
UFH	Protamine sulfate	1 mg/100 units heparin	5 min
LMWH (<8 h ago)	Protamine sulfate	1 mg/100 anti-Xa units	Partial (60%)
Dabigatran	Idarucizumab (Praxbind)	5 g IV (2 × 2.5 g)	Minutes
Rivaroxaban/Apixaban	Andexanet alfa	Weight/dose-adjusted	2 min
Fondaparinux	rFVIIa (off-label)	20–30 mcg/kg	Limited evidence

9. State-of-the-Art Updates: Evidence Changing Practice

9.1 Viscoelastic-Guided Haemostatic Therapy: Now Level 1 Evidence in Trauma

The ITACTIC trial (2020) and multiple systematic reviews have demonstrated that VET-guided transfusion reduces blood product usage, reduces allogeneic transfusion exposure, and in some studies reduces mortality versus conventional coagulation test-guided therapy. ROTEM/TEG algorithms have now been incorporated into the European Trauma Guidelines (2023, 7th edition) as the primary haemostatic monitoring tool. Extension of this approach to medical ICU coagulopathy is an active area of investigation.

9.2 Sepsis-Associated Coagulopathy Scoring and the ISTH DIC Score

The ISTH overt DIC score (platelets, PT, fibrinogen, D-dimer) has become the clinical standard for DIC diagnosis. A score ≥5 = overt DIC. The Japanese Association for Acute Medicine (JAAM) DIC score adds SIRS criteria and is more sensitive in sepsis. Recent studies (SCARLET trial, 2019) investigated recombinant thrombomodulin for sepsis-associated DIC — though SCARLET did not achieve its primary endpoint (28-day mortality), post-hoc analyses suggest benefit in the subgroup with coagulopathy but NOT thrombocytopaenia. The story is not closed.

9.3 Fibrinogen Concentrate vs. Cryoprecipitate: The FIBRES Trial

The landmark FIBRES trial (NEJM, 2019) — 735 cardiac surgery patients — demonstrated non-inferiority of fibrinogen concentrate to cryoprecipitate for haemostatic efficacy, with fewer units administered and equivalent safety. This trial has accelerated adoption of fibrinogen concentrate as the fibrinogen replacement product of choice in bleeding cardiac surgery patients and is reshaping practice in other ICU contexts.

9.4 The Transfusion Threshold Revisited: TRISS and Beyond

The landmark TRISS trial established that a restrictive transfusion threshold (Hb 70 g/L) is non-inferior to a liberal threshold (90 g/L) in septic shock. The TRICC, TRISS, and TRICS-III trials collectively support a Hb threshold of 70–80 g/L in most ICU patients without active cardiac ischaemia. Transfusing to a Hb >90 g/L is now considered inappropriate in most ICU contexts — it does not improve outcomes and increases transfusion-related complications (TACO, TRALI, immunomodulation).

Exception: Patients with acute coronary syndrome, symptomatic cardiac failure, or neurocritical injury may tolerate a threshold of 80–100 g/L. Individualise; do not protocolise blindly.

9.5 Gut Microbiome and Vitamin K Synthesis: An Emerging Story

Recent metagenomics research has demonstrated that prolonged ICU admission causes profound dysbiosis of the gut microbiome, reducing microbial menaquinone (vitamin K2) synthesis. Combined with reduced oral intake, malabsorption, and broad-spectrum antibiotics (which eliminate vitamin K-producing bacteria), this creates a state of subclinical vitamin K deficiency that is extraordinarily common and almost universally ignored. A 2023 study in Critical Care Medicine found that 68% of ICU patients on day 7 had vitamin K levels below the normal range, and that empirical IV vitamin K supplementation significantly reduced FFP requirements.

9.6 DOAC-Associated ICU Bleeding: A Growing Crisis

Direct oral anticoagulants now account for >60% of anticoagulant prescriptions in many countries. The ICU clinician in 2025 must be fluent in DOAC reversal. Key updates:

Idarucizumab (dabigatran reversal) is renally cleared — repeat dosing may be needed in severe AKI
Andexanet alfa (Xa inhibitor reversal) — the 2023 ANNEXA-I trial (NEJM, 2023) confirmed superiority over placebo for intracranial haemorrhage, though concerns about thrombotic rebound persist
In the absence of specific reversal agents: 4F-PCC 50 IU/kg provides partial reversal of Xa inhibitors and is the emergency bridge until andexanet alfa is available

10. When to Escalate / When to Watch

The fundamental question at 0300 h is never "should I do something?" but "what is the WORST THING that will happen if I do nothing for the next 2 hours?"

Escalate Immediately When:

Haemorrhagic shock (SBP <90, HR >120, lactate rising)
Bleeding into enclosed space: intracranial, pericardial, retroperitoneal (no room for expansion)
Airway threatened by bleeding (haemoptysis, oropharyngeal haemorrhage)
Platelet count <10 × 10⁹/L (risk of spontaneous intracranial bleeding)
Fibrinogen <1.0 g/L despite replacement (refractory consumption — underlying DIC uncontrolled)
Signs of organ dysfunction attributable to haemorrhage (rising creatinine, hepatic encephalopathy, ischaemic ECG changes)
Failure to respond to first-line haemostatic therapy within 30–60 minutes — reassess diagnosis

Safe to Monitor When:

Stable haemodynamics, haemoglobin stable after initial transfusion
Bleeding confined to a single non-critical site (peripheral IV ooze, minor epistaxis) without coagulopathy
Platelet count 50–100 × 10⁹/L with stable trend, no active bleeding
INR 1.5–2.0 without active bleeding (prophylactic correction is evidence-free in most contexts)
Post-procedural ooze responding to local pressure within 15–20 minutes

The "Watch and Worry" Zone — Mandatory Reassessment in 2–4 Hours:

Platelet count 20–50 × 10⁹/L with stable bleeding pattern
Fibrinogen 1.0–1.5 g/L on replacement — is the DIC being controlled?
Haemoglobin dropping 10–20 g/L per 12 hours without obvious source
Any patient in whom the primary diagnosis driving the coagulopathy (sepsis, DIC, liver failure) has not yet been controlled

11. A Memorable Summary: The BLEED Framework

🩸 B — Bleeding Pattern (local vs. systemic; mucocutaneous vs. deep tissue) 🔬 L — Laboratory Trend (direction of PT, platelets, fibrinogen — not just the number) ⚡ E — Eliminate Reversible Causes (drugs, temperature, acidosis, calcium, HIT, DOAC) 🩹 E — Escalate Fibrinogen FIRST (cryoprecipitate/fibrinogen concentrate before FFP) 💊 D — Definitive Treatment of the Underlying Cause (you cannot haemostatically resuscitate a patient whose sepsis/liver failure/DIC is uncontrolled)

Summary Table: ICU Bleeding — Pattern, Mechanism, and Management

Clinical Pattern	Most Likely Mechanism	First-Line Action	Pitfall to Avoid
Ooze from all puncture sites	DIC	Fibrinogen concentrate + treat cause	Giving FFP without fibrinogen first
Isolated thrombocytopaenia on heparin	HIT	4T score → stop heparin → argatroban	Starting warfarin in active HIT
Prolonged APTT, normal PT	UFH accumulation in AKI	Check anti-Xa; consider protamine	Assuming therapeutic heparin is safe in AKI
Bleeding + cirrhosis, INR 2.5	Balanced coagulopathy (not true coagulopathy)	Treat precipitant; TEG-guided therapy	Reflexive FFP for INR without active bleeding
GI bleeding + antiplatelet drugs	Platelet dysfunction	DDAVP 0.3 mcg/kg + PPI	Transfusing platelets before DDAVP trial
Trauma + haemorrhagic shock	Acute traumatic coagulopathy + hyperfibrinolysis	TXA within 3 h + 1:1:1 MTP	Delaying TXA for labs to return
Post-cardiac surgery ooze	Platelet dysfunction + heparin effect	Protamine; TEG-guided; DDAVP	Over-protaminating (excess protamine = anticoagulant)
LVAD patient with GI bleeding	Acquired von Willebrand syndrome	DDAVP; consider device speed reduction	Transfusing platelets (ineffective in AVWS)
CRRT citrate + bleeding	Citrate toxicity → hypocalcaemia	Ionised calcium measurement; calcium supplementation	Missing hypocalcaemia in a "coagulopathic" patient

12. References

Levi M, Scully M. How I treat disseminated intravascular coagulation. Blood. 2018;131(8):845–854.
Hunt BJ. Bleeding and coagulopathies in critical care. N Engl J Med. 2014;370(9):847–859.
Collaborators C-T. Effects of tranexamic acid on death, vascular occlusive events, and blood transfusion in trauma patients with significant haemorrhage (CRASH-2). Lancet. 2010;376(9734):23–32.
Spahn DR, Bouillon B, Cerny V, et al. The European guideline on management of major bleeding and coagulopathy following trauma: sixth edition. Crit Care. 2023;27(1):80.
Curry N, Rourke C, Davenport R, et al. Early cryoprecipitate for major haemorrhage in trauma: a randomised controlled feasibility trial. Br J Anaesth. 2015;115(1):76–83.
Callum J, Farkouh ME, Scales DC, et al. Effect of fibrinogen concentrate vs cryoprecipitate on blood component transfusion after cardiac surgery: the FIBRES randomized clinical trial. JAMA. 2019;322(20):1966–1976.
Holcomb JB, Tilley BC, Baraniuk S, et al. Transfusion of plasma, platelets, and red blood cells in a 1:1:1 vs a 1:1:2 ratio and mortality in patients with severe trauma (PROPPR). JAMA. 2015;313(5):471–482.
Mazzeffi M, Greenfield RA, Tanaka KA. Viscoelastic haemostatic testing in major trauma and cardiac surgery: practical aspects of TEG and ROTEM. Anaesthesia. 2022;77(Suppl 1):42–52.
Hébert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care (TRICC). N Engl J Med. 1999;340(6):409–417.
Vincent JL, Sakr Y, Sprung C, et al. Are blood transfusions associated with greater mortality rates? Anesthesiology. 2008;108(1):31–39.
Connors JM, Levy JH. COVID-19 and its implications for thrombosis and anticoagulation. Blood. 2020;135(23):2033–2040.
Schulman S, Angeras U, Bergqvist D, Eriksson B, Lassen MR, Fisher W. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in surgical patients. J Thromb Haemost. 2010;8(1):202–204.
Faraoni D, Tanaka KA, Donahue BS, et al. Perioperative considerations for the patient with haemophilia A or B undergoing surgery. Semin Thromb Hemost. 2019;45(8):810–821.
Warkentin TE. Laboratory diagnosis of heparin-induced thrombocytopenia. Int J Lab Hematol. 2019;41(Suppl 1):15–25.
Milling TJ Jr, Kaatz S. Preclinical and clinical data for factor Xa and "universal" reversal agents. Am J Emerg Med. 2016;34(11S):39–45.

Disclosure: The author declares no conflicts of interest relevant to this article. This review represents the synthesis of current evidence and clinical experience and should not replace institutional protocols or individual clinical judgement. Evidence-based medicine is a foundation, not a ceiling.

Word count: ~5200

Thursday, April 9, 2026