Clinical Calculators

Red Cell Transfusion Threshold Guide

Pick a clinical setting and see the haemoglobin trigger each authority actually states for it — AABB 2023, the 2025 AABB myocardial infarction guideline, ACG 2021, NICE NG24 and the Surviving Sepsis Campaign — with the randomised trial each figure rests on. Where two societies genuinely disagree, both numbers are shown attributed rather than averaged, and settings every document excludes return no number at all. There is no single universal trigger and this page never prints one.

1 · Clinical setting
2 · Haemoglobin
Switching units converts the number you have already typed, so the clinical value never changes silently. 1 g/dL is exactly 10 g/L. Leave the field blank to read the thresholds without comparing anything to them.
3 · Anything that changes the scope

Each of these is a situation a guideline explicitly writes itself out of, or one where a guideline adds a separate statement. Only the boxes that matter for the chosen setting are shown.

What each authority states
What this page reads from

No haemoglobin figure on this page originates here. Every trigger, every strength of recommendation and every certainty rating is transcribed from one of five published documents, each read from primary text rather than from a summary of it, and each named on the card that carries its number. The table below lists all five in full, with the identifier needed to open the source and check the wording against what this tool prints.

DocumentCitation and official sourceMethodAs of
AABB 2023
Association for the Advancement of Blood & Biotherapies, with the International Society of Blood Transfusion, the International Collaboration for Transfusion Medicine Guidelines, the Society of Critical Care Medicine, the European Blood Alliance and the Society for the Advancement of Patient Blood Management
Carson JL, Stanworth SJ, Guyatt G, et al. Red Blood Cell Transfusion: 2023 AABB International Guidelines. JAMA. 2023;330(19):1892-1902. Official source
DOI 10.1001/jama.2023.12914 · PMID 37824153
GRADE, 45 randomised controlled trials with 20,599 adult participants and 7 trials with 2,730 paediatric participants2023-11-21
AABB 2025 (AMI)
Association for the Advancement of Blood & Biotherapies international panel
Pagano MB, Stanworth SJ, Dennis J, et al. Red Cell Transfusion in Acute Myocardial Infarction: AABB International Clinical Practice Guidelines. Ann Intern Med. 2025;178(10):1469-1477. Official source
DOI 10.7326/ANNALS-25-00706 · PMID 40825204
GRADE, meta-analysis of randomised trials of restrictive versus liberal strategies in acute myocardial infarction using Cochrane methods2025-08-19
ACG 2021 (UGIB)
American College of Gastroenterology
Laine L, Barkun AN, Saltzman JR, Martel M, Leontiadis GI. ACG Clinical Guideline: Upper Gastrointestinal and Ulcer Bleeding. Am J Gastroenterol. 2021;116(5):899-917. Official source
DOI 10.14309/ajg.0000000000001245 · PMID 33929377
GRADE, systematic reviews of predefined clinical questions2021-05
NICE NG24
National Institute for Health and Care Excellence (England)
Blood transfusion. NICE guideline NG24. National Institute for Health and Care Excellence, published 18 November 2015. Official source
NICE guideline NG24
NICE guideline development; recommendations are worded ‘use’ or ‘consider’ rather than GRADE-rated in the published recommendation text2015-11-18
Surviving Sepsis 2026
Society of Critical Care Medicine and European Society of Intensive Care Medicine
Prescott HC, Antonelli M, Alhazzani W, et al. Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock 2026. Crit Care Med. 2026;54(4):725-812. Official source
DOI 10.1097/CCM.0000000000007075 · PMID 41869847
GRADE2026-03

The trial evidence underneath. Nine randomised controlled trials sit behind those recommendations, and the tool opens only the ones that enrolled the population selected above: TRICC (1999), TRISS (2014), Villanueva 2013 (2013), TRIGGER (2015), TRICS III (2017), FOCUS (2011), MINT (2023), REALITY (2021), TRIPICU (2007). Each trial panel reproduces that trial’s own restrictive and liberal arm definitions, its primary outcome as the authors defined it, the published result with its confidence interval, and the authors’ own concluding sentence. Nothing is restated in friendlier words.

One figure here is a convention rather than a published value. The conversion between the two unit systems is definitional: one gram per decilitre is exactly ten grams per litre. Every threshold is therefore shown in both units with no rounding in either direction, because United Kingdom and European documents state these numbers in g/L and United States documents state them in g/dL, and a reader comparing NICE with AABB is otherwise comparing numbers that look ten times apart.

Reference data build date. The underlying data file was built 2026-08-01 and carries a re-verification date of 2027-02-01. Guidelines are revised; a reader using this page after that date should confirm each figure against the linked source.

How to read the result

The tool reports position, not permission. Entering a haemoglobin does not produce a recommendation to transfuse or to withhold. It produces an arithmetic statement — this value is 0.4 g/dL below the figure that source states — next to the sentence the source actually wrote. The distance is the useful part: a value 2 g/dL under every threshold on screen is a different conversation from one sitting 0.1 g/dL under a single conditional recommendation.

A trigger and a target are not the same number. Most of these documents state only the first. NICE NG24 is the only source implemented here that also gives a post-transfusion haemoglobin target — 70 to 90 g/L in general, and 80 to 100 g/L in acute coronary syndrome — which is why target ranges appear on NICE cards and nowhere else. The absence of a target on the other cards is not an oversight; those panels did not state one.

Strong and conditional carry different weight. A strong recommendation means the panel judged that most well-informed patients would want the recommended course. A conditional or weak recommendation means the panel judged that choices would reasonably differ. Only three statements on this page are strong: AABB Recommendation 1 for the stable hospitalised adult, AABB Recommendation 3 for the critically ill child, and the Surviving Sepsis Campaign directional recommendation. The acute myocardial infarction, haematology and oncology, paediatric congenital heart disease and upper gastrointestinal bleeding recommendations are all conditional, and the certainty rating printed beneath each card says how much trial evidence stands behind it.

Exactly at the number is genuinely ambiguous. Some documents say “less than 7 g/dL” and others say “a threshold of 7 g/dL”. At exactly 7.0 those are not the same statement, and the tool says so rather than resolving it. The range in which any of these figures is meaningful is roughly 6 to 10 g/dL; outside it the question is no longer which threshold applies.

The number moves with the population, not with the reading. Changing the setting changes the trial evidence, and the whole reason this page exists is that the two most recent documents point in opposite directions: 7 g/dL for a stable inpatient, 10 g/dL after a myocardial infarction. Averaging those would misrepresent both, so no single universal trigger is printed anywhere on this page.

A worked example

Consider an inpatient in their seventies admitted with melaena, circulation stable after initial fluids, no chest pain and no known coronary disease, whose haemoglobin comes back at 6.8 g/dL. Selecting Acute upper gastrointestinal bleeding and entering 6.8 returns the ACG 2021 statement — a conditional recommendation, low-quality evidence, threshold 7 g/dL — and marks the entered value as 0.2 g/dL below the figure that source states. Alongside it sits AABB Recommendation 1 at the same 7 g/dL, strong and moderate certainty, with a note that the AABB guideline never names upper gastrointestinal bleeding as a subgroup at all.

The Villanueva trial panel opens underneath and is where the 7 g/dL came from: 921 patients, restrictive arm transfused below 7 g/dL, liberal arm below 9 g/dL, six-week survival 95 versus 91 per cent (hazard ratio for death 0.55, 95% CI 0.33 to 0.92, P = 0.02), further bleeding 10 versus 16 per cent, and 51 per cent of the restrictive arm transfused not at all.

Now consider the same reader ticking Hypotensive at presentation. The cards do not change, but the ACG’s own narrative appears beneath them with no separate figure attached: haemoglobin in a hypotensive patient will be lower after fluid resuscitation even without further bleeding, so the panel agreed it is reasonable to transfuse before 7 g/dL is reached. Ticking Major or exsanguinating haemorrhage instead removes every number from the screen, because the trial that produced the 7 g/dL figure excluded exactly those patients and NICE writes major haemorrhage out of its restrictive rule in the recommendation text itself. A calculator that kept printing 7 g/dL there would be quoting a study of a different patient.

Where this tool breaks

A calculator is only trustworthy if it names the cases it should not be used for. These are the ones for this page, and in most of them the tool actively declines to display a figure rather than displaying a misleading one.

Haemodynamic instability and major haemorrhage
The tool refuses to print a number rather than guessing one. Every document implemented here writes itself out of those situations, and a haemoglobin drawn during brisk bleeding or before equilibration after fluid resuscitation is a lagging measurement in any case.
Trauma, obstetric haemorrhage and massive transfusion protocols
None of these five documents sets a haemoglobin trigger for them. Practice in those settings is driven by ratio-based protocols rather than by a haemoglobin number, and no such protocol is implemented here.
Chronic transfusion dependence, sickle cell disease and thalassaemia
NICE deliberately excludes patients needing regular transfusions for chronic anaemia from its restrictive thresholds and recommends individual thresholds instead. No disease-specific programme guidance was read for this build, so none is shown.
Preterm neonates and outpatient transfusion
The AABB paediatric recommendations cover critically ill children and children with congenital heart disease only. The guideline states in terms that trial data were insufficient to inform recommendations on outpatient transfusion management, so there is no outpatient number to display.
Documents that could not be read in full
Baveno VII on variceal bleeding and the AASLD portal hypertension guidance are paywalled, and the 2025 ACC/AHA/ACEP/NAEMSP/SCAI acute coronary syndrome guideline returned access errors on both the Circulation and the JACC versions. Those documents are named on this page but no figure is attributed to any of them, because a threshold quoted second-hand would not be checkable.
A guideline that states no number at all
The Surviving Sepsis Campaign 2026 recommendation names a direction of travel, not a haemoglobin concentration. The data file stores a null for it rather than borrowing another society’s figure, so the sepsis setting falls back on the general hospitalised-adult recommendation and says so.
Dose, volume and rate
No transfusion volume, rate or product dose appears anywhere, because these documents state none — with the single exception of the NICE single-unit statements, which are reproduced in full and attributed.

And one that applies everywhere. A haemoglobin concentration is not the decision. The AABB panel says as much in its own concluding text: optimal transfusion practice relies not only on threshold concentrations but on the patient’s symptoms, signs, comorbid conditions, rate of bleeding, values and preferences. This page shows where a number sits against published guidance. It does not know the patient.

Questions this page is asked
Why does this page refuse to give one number?
Because the published documents do not agree on one, and the disagreement is substantive rather than cosmetic. AABB 2023 recommends considering transfusion below 7 g/dL for a stable hospitalised adult. The same organisation’s 2025 acute myocardial infarction guideline suggests a liberal strategy with transfusion below 10 g/dL, and states that a restrictive strategy of 7 to 8 g/dL may result in increased mortality in that population. Those are different populations with different trial evidence, and an average of the two would describe no patient at all.
Why do AABB and NICE differ on acute coronary syndrome?
The gap is dateable. NICE NG24 was published on 18 November 2015 and recommends considering 80 g/L with a post-transfusion target of 80 to 100 g/L in acute coronary syndrome. The two randomised trials that moved the AABB panel reported afterwards, in 2021 and 2023; the larger randomised 3,504 patients and returned a composite of myocardial infarction or death at 30 days of 16.9 per cent restrictive against 14.5 per cent liberal, a risk ratio of 1.15 with a confidence interval of 0.99 to 1.34 that did not reach significance. The AABB panel read the direction of every point estimate as a possible harm signal and moved its number up. NICE has not revisited the question. Both figures are current published guidance from a named body, which is why both are shown.
Is 8 g/dL or 10 g/dL right for a patient with heart disease?
That depends on a clinical judgement the tool cannot make: whether an acute infarction is present. AABB 2023 offers 8 g/dL as the trial-consistent option for pre-existing cardiovascular disease, inside a recommendation whose headline figure is 7 g/dL. The 2025 guideline from the same organisation suggests 10 g/dL for acute myocardial infarction. These are not contradictory; they are different populations. The tool shows both under the cardiac settings precisely so the 2 g/dL consequence of that judgement is visible rather than hidden.
Does a restrictive threshold mean fewer transfusions are always better?
No. Restrictive strategies performed at least as well as liberal ones across a large adult evidence base, and in severe upper gastrointestinal bleeding the restrictive arm showed better six-week survival. But TRICC excepted acute myocardial infarction and unstable angina in its own conclusion, the 2023 myocardial infarction trial could not exclude harm from restriction, and every one of these trials excluded the unstable and the exsanguinating patient. A restrictive threshold is a finding about a studied population, not a universal principle.
Can a patient use this page to check their own haemoglobin result?
It will show what published guidance says about a concentration, which may be useful context, but it cannot say whether a transfusion is indicated for any individual. The page states no dose and no volume, it does not know the rate of bleeding, the symptoms or the comorbidities, and several of the numbers it prints are conditional recommendations resting on low-certainty evidence. Questions about a personal result, or about a transfusion that has been offered or declined, belong with the doctor responsible for that care.
Was this reference useful?