9. Blood Transfusion Thresholds in Critical Illness

Quick Recap

Cross-cutting supportive-care protocol — applies across all ICU systems; cross-reference Massive Transfusion Protocol (Hematology System) for the distinct, ratio-based approach to active major hemorrhage, which is a different clinical problem from the stable, non-bleeding anemic ICU patient this protocol addresses. This topic has a well-established general rule (restrictive is safe and preferred) with genuinely important, evidence-based population-specific exceptions that have emerged from major 2023–2024 trials — the exceptions are the most clinically important recent development here.

1. Definition

Restrictive transfusion strategy: transfusing red blood cells only when hemoglobin falls below a threshold of approximately 7 g/dL (with some protocols using 7–8 g/dL), transfusing to bring hemoglobin back above that threshold rather than to a higher target.

Liberal transfusion strategy: transfusing at a higher threshold, historically 9–10 g/dL, reflecting older practice patterns that assumed higher hemoglobin was uniformly protective.

The general critical care paradigm (established by TRICC and reinforced by numerous subsequent trials): for the general, hemodynamically stable, non-actively-bleeding critically ill patient, a restrictive strategy is non-inferior to liberal transfusion and is associated with fewer complications — this has been standard of care for over two decades. The clinically important recent development is the emergence of specific patient populations where this general rule does not apply, most notably acute myocardial infarction and acute brain injury (Section 11).

2. Pathophysiology

Why restrictive transfusion is generally safe: critically ill patients tolerate moderate anemia better than historically assumed, given compensatory mechanisms (increased cardiac output, increased oxygen extraction ratio) that maintain adequate tissue oxygen delivery down to hemoglobin levels well below the traditional 10 g/dL threshold, in the absence of a specific vulnerability that impairs this compensation.

Why transfusion itself carries real risk, not just theoretical risk: transfusion-associated circulatory overload (TACO), transfusion-related acute lung injury (TRALI), immunomodulation, infection risk, and simple volume/iron burden all represent genuine costs of transfusion that a purely "more oxygen-carrying capacity is better" framework ignores — this is the mechanistic rationale for restrictive strategies being preferred by default rather than liberal transfusion being a "safe extra."

Why specific populations may need a different calculus: in acute myocardial infarction, the myocardium supplied by a critically stenosed or recently infarcted vessel may be uniquely vulnerable to further ischemic injury from reduced oxygen-carrying capacity, in a way that other tissue beds are not — anemia may directly exacerbate ongoing ischemia in a way that compensatory mechanisms cannot adequately buffer in this specific, localized vascular territory. In acute brain injury, the injured brain has impaired autoregulation and may be similarly less able to compensate for reduced oxygen delivery than other organs, and unlike most tissue beds, neuronal injury from suboptimal oxygen delivery may be irreversible in a way that transient reduced oxygen delivery elsewhere is not — this is the biological plausibility underlying why these two specific populations have now shown a different result in dedicated trials (Section 11).

3. Immediate Stabilization (ABCDE) — Transfusion Threshold as a Bundle Element

Not an acute stabilization scenario in the actively hemorrhaging patient (cross-reference Massive Transfusion Protocol for that scenario); this protocol addresses the ongoing, day-to-day transfusion threshold decision in the stable ICU patient:

Checklist:

4. Focused History

• Underlying reason for anemia (chronic vs. acute, ongoing vs. resolved blood loss)
• Cardiac history, specifically active or recent myocardial infarction/ischemia
• Neurological history, specifically acute brain injury (traumatic brain injury, subarachnoid hemorrhage, other acute intracranial pathology)
• Cardiopulmonary reserve/comorbidity affecting anemia tolerance
• Ongoing symptoms potentially attributable to anemia (angina, dyspnea, altered mental status) versus asymptomatic laboratory finding

5. Comprehensive System-wise Examination

• Cardiovascular: signs of ischemia, hemodynamic tolerance of current hemoglobin level
• Neurological: mental status, focal deficits (relevant to acute brain injury population-specific threshold)
• General: signs of active bleeding requiring a different management pathway (cross-reference Massive Transfusion Protocol, or relevant bleeding-source protocol) rather than a simple threshold-based transfusion decision

POCUS integration: cardiac function assessment may inform individualized tolerance of anemia in a patient with reduced cardiac reserve, though this is adjunctive to, not a replacement for, the population-based threshold evidence in Section 11.

6. Syndrome Identification — Reframed as Population/Threshold Classification

• General critically ill patient, no specific vulnerability: restrictive threshold (~7 g/dL) per TRICC and the broad supporting trial base
• Acute myocardial infarction with anemia: evidence now supports consideration of a more liberal approach, though this remains genuinely debated rather than a clean reversal (Section 11)
• Acute brain injury with anemia: recent trial evidence (2024) supports a liberal threshold (90–100 g/L) over restrictive (70 g/L) for improved neurological outcomes — a clear, population-specific exception to the general paradigm
• Active hemorrhage: cross-reference Massive Transfusion Protocol — this is a fundamentally different clinical scenario (ratio-based resuscitation, not threshold-based maintenance transfusion)
• Upper GI bleeding: historically studied separately from general critical illness (excluded from TRICC); restrictive strategies have shown similar mortality benefit in this population as well

7. Differential Diagnosis — Reframed as "Why Is This Patient Anemic, and Does the Cause Change the Threshold Decision?"

Must-not-miss:

• Ongoing occult bleeding not yet clinically apparent (distinguishes a threshold-based decision from a resuscitation-based one)
• Hemolysis (may require a different diagnostic and therapeutic approach beyond simple transfusion)

Common:

• Phlebotomy-related blood loss (a genuine, underappreciated contributor to ICU anemia over a prolonged stay)
• Dilutional anemia from fluid resuscitation
• Chronic disease-associated anemia present before this admission

8. Severity/Risk Assessment

General ICU population: hemoglobin threshold itself (7 g/dL restrictive vs. 9–10 g/dL liberal) is the primary severity/decision metric, per the extensive supporting trial base (Section 11/22).

Acute MI-specific consideration: anemia acuity (acute drop vs. chronic baseline anemia) may modulate the risk-benefit calculus — a secondary analysis of MINT trial data specifically examined this dimension, suggesting the acuity of the anemia itself, not just the absolute hemoglobin level, may be clinically relevant, though this remains an area of ongoing analysis rather than a settled, separately validated threshold.

9. Investigations

Immediate bedside: hemoglobin/hematocrit trend (trend more informative than a single value, particularly when assessing for ongoing occult loss vs. stable chronic anemia)

Routine labs: standard anemia workup as clinically indicated to identify a treatable underlying cause (iron studies, hemolysis workup) distinct from the acute transfusion-threshold decision itself

Repeat frequency: hemoglobin monitoring frequency should be matched to clinical stability and bleeding risk, not applied as a fixed daily ritual regardless of clinical context

10. Point-of-Care Ultrasound

Not a primary component of this protocol; may be used adjunctively to assess cardiac function/tolerance of anemia in specific complex cases, or to investigate for an occult bleeding source if acute anemia is unexplained.

11. Evidence-Based Management

The General Rule — Restrictive Transfusion for the Broad ICU Population

• TRICC trial (1999, n=838): restrictive strategy (transfuse for Hgb <7 g/dL) was non-inferior to liberal strategy (transfuse for Hgb <10 g/dL) for 30-day mortality, with fewer complications and lower transfusion-related adverse events in the restrictive arm; a prospective secondary analysis found the liberal strategy was actually associated with an increased in-hospital death rate (p=0.05) — this trial established restrictive transfusion as standard of care and remains foundational more than two decades later
• TRISS trial (2014): no difference in 90-day mortality between Hgb thresholds of 7 vs. 9 g/dL specifically in septic shock patients — extended the restrictive-strategy evidence base into this specific, high-acuity population
• Large pooled meta-analysis (2025, 13 RCTs, 13,705 critically ill adults, PRISMA-guided): no statistically significant difference in 30-day mortality between restrictive and liberal strategies (OR 1.02) — confirms non-inferiority of restrictive transfusion at a large, pooled evidentiary scale
• 2023 AABB International Guidelines: continue to recommend restrictive transfusion strategies as the default for most hospitalized patients, including those with stable cardiovascular disease — the current authoritative guideline position for the general population

Population-Specific Exception #1: Acute Myocardial Infarction — A Genuinely Unsettled, Actively Debated Area

• MINT trial (Carson et al., NEJM 2023, n=3,504, 144 sites, 6 countries): patients with acute MI and Hgb <10 g/dL randomized to restrictive (transfuse at 7–8 g/dL) vs. liberal (transfuse at <10 g/dL) — composite outcome of 30-day death or MI occurred in 16.9% (restrictive) vs. 14.5% (liberal), RR 1.15 (95% CI 0.99–1.34, p=0.07) — did not reach conventional statistical significance, but trended toward benefit with the liberal strategy, a notable reversal in direction from the general ICU paradigm
• Patient-level meta-analysis (Carson et al., NEJM Evidence 2024, pooling MINT with 3 prior smaller MI-specific trials — REALITY, CRIT, and an earlier Carson pilot — total n=4,311): examined restrictive (7–8 g/dL) vs. liberal (10 g/dL) specifically in MI populations — this pooled analysis is the most informative current evidence synthesis for this specific population
• 2023 AABB guideline update: now recommends transfusing patients with acute coronary syndrome to Hgb ≥10 g/dL (a liberal approach), a specific, explicit exception carved out from the guideline's otherwise restrictive default — reflecting that "previously available data were misinterpreted" regarding this population per recent authoritative synthesis
• Practical synthesis: unlike the acute brain injury evidence below, the acute MI evidence base remains genuinely contested — MINT's primary result was not itself statistically significant, and there is active ongoing debate in the literature (including a 2026 "con" perspective piece explicitly arguing restrictive triggers remain preferable even in this population) — this is not a settled reversal but an area where reasonable clinicians, and even guideline bodies, currently diverge; a liberal approach (Hgb ≥10 g/dL) in acute MI with anemia is a reasonable, guideline-supported consideration, but should not be presented as unambiguously established practice

Population-Specific Exception #2: Acute Brain Injury — A Clearer, More Recent Reversal

• HEMOTION trial (2024, traumatic brain injury population): restrictive (transfuse at Hgb ≤70 g/L) vs. liberal (transfuse to maintain Hgb ≥100 g/L) — liberal transfusion associated with better neurological outcomes
• TRAIN trial (2024, broader acute brain injury population including non-traumatic etiologies): restrictive (<70 g/L) vs. liberal (<90 g/L) — similarly found liberal transfusion associated with better outcomes in this population
• Both trials maintained their respective thresholds for the full duration of ICU stay (mean ~15 days in HEMOTION, ~22 days in TRAIN), reflecting a sustained strategy rather than a brief initial intervention
• Practical synthesis: this represents a genuinely clear, recent (2024), population-specific reversal of the general restrictive paradigm — for patients with acute brain injury (traumatic or non-traumatic) and anemia, a liberal transfusion strategy targeting Hgb 90–100 g/L is now supported by two dedicated, contemporary RCTs, in explicit contrast to the general ICU and even the more contested acute MI evidence

12. Organ Support

Interacts with overall fluid balance strategy (transfusion volume contributes to fluid balance, relevant to Nutrition Support and Vasopressor protocols' broader hemodynamic management context) and with the underlying condition driving ICU admission in the population-specific exceptions (acute MI, acute brain injury — cross-reference relevant Cardiovascular and Neurology System protocols).

13. Disease-Specific Therapy — Threshold Reference Table

14. Consultation Matrix

15. Monitoring Framework

• Clinical: hemoglobin trend, clinical tolerance of current level (symptoms, hemodynamic stability)
• Escalation triggers: acute drop suggesting new/ongoing bleeding — reassess for a bleeding source rather than simply transfusing to a threshold
• De-escalation criteria: stable hemoglobin above the relevant population-specific threshold → no transfusion indicated regardless of absolute number if well above threshold

16. ICU Bundle Checklist (Daily, When Anemia Is Present)

17. Complications

Early:

• Transfusion-associated circulatory overload (TACO)
• Transfusion-related acute lung injury (TRALI)
• Acute hemolytic/febrile transfusion reactions
• Under-transfusion consequences in a population with a genuine vulnerability (e.g., ongoing ischemia in acute MI, secondary neurological injury in acute brain injury) if the general restrictive default is applied without considering population-specific exceptions

Late:

• Iron overload with repeated transfusion over a prolonged ICU/hospital course
• Alloimmunization

Prevention: population-appropriate threshold selection, single-unit transfusion with reassessment, minimizing iatrogenic (phlebotomy-related) blood loss

Rescue: standard transfusion reaction management protocols; reassessment for an unaddressed bleeding source if hemoglobin fails to respond as expected to transfusion

18. Escalation & De-escalation

Escalation: clinical deterioration attributable to anemia in a population with a specific vulnerability (acute MI, acute brain injury) → consider more liberal transfusion per the population-specific evidence above, rather than defaulting to the general restrictive threshold.

De-escalation: stable hemoglobin trend above threshold → no transfusion indicated; reassess threshold appropriateness if the patient's clinical population/vulnerability status changes over the course of the ICU stay (e.g., resolution of the acute neurological injury phase).

19. ICU Discharge Criteria (Transfusion-Relevant Context)

Cross-reference ICU Discharge Criteria & Step-Down protocol. Transfusion-specific consideration: current transfusion threshold/strategy and rationale communicated at transfer, particularly if a population-specific liberal threshold (acute brain injury, acute MI) was in use, since receiving ward teams may default to the general restrictive threshold without this context.

20. Documentation & Medicolegal Checklist

• Transfusion threshold applied and rationale documented, particularly for any population-specific deviation from the general restrictive default
• Hemoglobin trend (not just isolated values) documented to support the clinical reasoning
• Underlying cause of anemia, if identified, documented
• Any transfusion reaction documented with management
• Handoff of threshold strategy at any care transition documented

21. Key Guidelines

• 2023 AABB International Guidelines (Carson et al., JAMA 2023): current authoritative source — restrictive strategy recommended as default, with an explicit liberal (≥10 g/dL) exception for acute coronary syndrome
• Current guidance for acute brain injury populations is expected to evolve to incorporate HEMOTION and TRAIN (2024), representing genuinely new evidence not yet reflected in older guideline documents

22. Landmark Trials

23. Controversies

• Acute MI transfusion threshold remains genuinely, actively contested — unlike the acute brain injury exception, MINT's primary result did not reach statistical significance, and recent literature includes explicit "con" perspectives arguing restrictive triggers remain preferable even in acute MI (e.g., a 2026 Journal of Cardiothoracic and Vascular Anesthesia piece explicitly titled arguing this position) — clinicians should present this as a genuine area of clinical equipoise and evolving practice, not a settled reversal, and should be transparent that the 2023 AABB guideline's liberal recommendation for ACS is itself somewhat ahead of the underlying primary trial's own statistical significance.
• Anemia acuity as a potentially important, under-characterized modifier: secondary MINT trial analyses have begun exploring whether the acuity of anemia onset (rather than the absolute hemoglobin value alone) modulates the risk-benefit calculus in acute MI — an emerging, not yet fully resolved, refinement to simple threshold-based thinking.
• Generalizability of the acute brain injury findings: HEMOTION and TRAIN are recent (2024) and represent the best current evidence for this population, but as with any newly established practice-changing finding, broader real-world confirmation and guideline incorporation are still in progress.
• Single-unit vs. multi-unit default transfusion practice: while single-unit transfusion with reassessment is increasingly recommended as good practice to avoid unnecessary over-transfusion, this remains inconsistently implemented across institutions relative to the threshold-level evidence itself, which has received more research attention than this specific administration-practice question.

24. References

1. Hebert 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.
2. Holst LB, Haase N, Wetterslev J, et al; TRISS Trial Group. Lower versus higher hemoglobin threshold for transfusion in septic shock. N Engl J Med. 2014;371(15):1381-1391.
3. Jimenez Franco DA, Perez Velasquez CA, Rodriguez Lima DR. Mortality in critically ill patients with liberal versus restrictive transfusion thresholds: a systematic review and meta-analysis of RCTs with trial sequential analysis. J Clin Med. 2025;14(6):2049.
4. Carson JL, Brooks MM, Hebert PC, et al; MINT Investigators. Restrictive or liberal transfusion strategy in myocardial infarction and anemia. N Engl J Med. 2023;389(26):2446-2456.
5. Carson JL, et al. Restrictive versus liberal transfusion in myocardial infarction — a patient-level meta-analysis. NEJM Evid. 2024.
6. Carson JL, Guyatt G, Heddle NM, et al. Red blood cell transfusion: 2023 AABB international guidelines. JAMA. 2023;330(19):1892-1902.
7. HEMOTION Investigators. Hemoglobin transfusion threshold in traumatic brain injury optimization. 2024.
8. TRAIN Investigators. Transfusion strategy in patients with acute brain injury. 2024.
9. Ho AK. Liberal versus restrictive transfusion trigger after acute brain injury: more than just higher blood oxygen content. Neurotrauma Rep. 2025.
10. Raasveld SJ, et al; InPUT Study Group. Red blood cell transfusion in the intensive care unit. JAMA. 2023;330(19):1852-1861.
11. The Washington Manual of Critical Care, 4th ed. 2025 — transfusion medicine chapter.
12. ICU Protocols: A Step-wise Approach, 2nd ed. — relevant transfusion content.