Highlights
Restrictive transfusion strategies (Hb ~7–8 g/dL) reduce red cell exposure by ~42% without increasing 30‑day mortality across most adult and paediatric populations. In gastrointestinal bleeding, restrictive transfusion lowered 30‑day mortality. In contrast, patients with acute brain injury had better neurological outcomes with liberal transfusion. Evidence for physiological triggers remains heterogeneous.
Background and clinical context
Red blood cell (RBC) transfusion is a cornerstone therapy for severe anaemia and acute bleeding, but blood is a scarce resource and transfusion carries potential harms, costs, and logistical burdens. Key clinical questions are whether transfusing earlier (liberal strategy) improves outcomes versus limiting transfusions to a lower haemoglobin (Hb) threshold (restrictive strategy), and whether additional physiologic criteria should guide transfusion decisions.
Historically, transfusion practice has been driven by convention, physiologic reasoning, and expert opinion. Randomized trials over the past three decades have compared restrictive and liberal haemoglobin thresholds across surgical, medical, critical care and paediatric settings. The 2025 Cochrane review by Carson and colleagues synthesizes randomized evidence across 69 trials and provides the most contemporary, high‑quality aggregate assessment of threshold strategies, with attention to subgroup effects and non‑mortality outcomes that clinicians value.
Study design and scope of evidence
Carson et al. performed a systematic review and meta-analysis of randomized controlled trials that compared restrictive versus liberal RBC transfusion strategies in adults and children. Searches covered major bibliographic databases and trial registries through 14 October 2024. The analysis included 61 adult trials (27,639 participants) and eight paediatric trials (2764 participants), plus trials evaluating physiologic triggers.
Restrictive thresholds were most commonly Hb 7.0–8.0 g/dL; liberal thresholds centered on Hb 9.0–10.0 g/dL. Primary outcomes included exposure to RBC transfusion, 30‑day mortality, and clinically important endpoints (myocardial infarction, stroke, infection, thromboembolism, and neurologic outcomes in brain‑injured patients). Risk of bias was generally low and GRADE was applied to determine certainty of evidence.
Key findings
Reduced transfusion exposure
Across adult and paediatric trials, restrictive strategies markedly reduced the proportion of patients who received at least one RBC transfusion (risk ratio [RR] 0.58, 95% confidence interval [CI] 0.52–0.65; high‑certainty evidence). This represents a roughly 42% relative reduction in exposure to transfusion, with considerable heterogeneity (I2 = 97%) reflecting differences in clinical contexts and trial protocols rather than a consistent direction of effect.
Mortality and major clinical endpoints
When pooled across clinical contexts, restrictive thresholds did not increase 30‑day mortality (RR 1.01, 95% CI 0.90–1.14; 44 studies, 22,575 participants; high‑certainty evidence). Similarly, there was no convincing difference between restrictive and liberal strategies for myocardial infarction, stroke, infection, thromboembolism, or congestive heart failure (moderate to high‑certainty evidence for most outcomes).
Two clinically important exceptions emerged:
– Gastrointestinal bleeding: In pooled trials of patients with acute GI bleeding, a restrictive transfusion strategy reduced 30‑day mortality (RR 0.63, 95% CI 0.42–0.95; 4 studies, 1,574 participants). The most influential trial in this area (Villanueva et al., 2013) and subsequent data suggest that a restrictive policy may reduce re‑bleeding and adverse cardiovascular events in some bleeding phenotypes.
– Neurocritical illness (traumatic or non‑traumatic brain injury): For critically ill patients with acute brain injury, liberal transfusion strategies produced better long‑term neurologic outcomes. Pooled estimates showed a relative increase in unfavourable neurological outcome with restrictive transfusion (RR 1.14, 95% CI 1.05–1.22; 4 studies, 2,297 participants; moderate‑certainty evidence), indicating that maintaining higher Hb may be neuroprotective in this subgroup.
Pediatrics
The paediatric evidence base was smaller and more heterogeneous (8 trials, 2,764 participants). There was no clear mortality difference between strategies (RR 1.22, 95% CI 0.72–2.08; low‑certainty evidence), and data on other patient‑important outcomes were underpowered. As with adults, restrictive approaches reduced transfusion exposure.
Transfusion reactions and adverse events
Transfusion‑specific reactions were uncommon overall but occurred more frequently with liberal transfusion policies (Peto odds ratio 0.47, 95% CI 0.36–0.62; 18 studies, 11,505 participants). This is consistent with fewer units transfused when adopting restrictive thresholds.
Physiological triggers
Ten trials evaluated physiological criteria (e.g., central venous oxygen saturation, lactate, mixed venous oxygen) alone or combined with Hb thresholds. Heterogeneity in interventions, patient populations, and outcome measures precluded meta‑analysis; the existing evidence is insufficient to endorse a single physiologic algorithm over Hb‑based thresholds, though individualized physiologic assessment remains logical in complex or ischemic states.
Interpretation and clinical implications
The 2025 Cochrane synthesis reinforces a simple, pragmatic message: for most hospitalized adults and children, adopting a restrictive RBC transfusion threshold (commonly Hb 7–8 g/dL) reduces transfusion exposure without increasing short‑term mortality or major complications. This supports restrictive policies as a general standard of care, aligns with prior landmark trials (e.g., TRICC, FOCUS) and guideline recommendations (e.g., AABB), and has important implications for patient safety, resource stewardship, and blood supply management.
However, the evidence also highlights important caveats. Certain patient groups appear to benefit from higher haemoglobin targets. Neurocritically ill patients (brain injury, intracerebral haemorrhage, severe stroke) had better long‑term neurological outcomes with liberal transfusion strategies, suggesting clinicians should individualize transfusion thresholds and may favor a higher Hb target in patients at high risk of cerebral hypoxia. Similarly, a restrictive strategy reduced mortality in GI bleeding trials—underscoring that the physiologic context (active bleeding, hemodynamic instability, source control) matters.
For patients with acute coronary syndromes, cardiogenic shock, or ongoing myocardial ischemia the evidence remains mixed; decisions should combine haemoglobin values with clinical signs, ECG changes, biomarkers, and consultation with cardiology and critical care teams.
Physiologic parameters (mixed venous O2, lactate, SvO2, near‑infrared spectroscopy for cerebral oxygenation) are attractive as individualized triggers but currently lack robust, consistent randomized evidence to replace Hb thresholds. They can, however, inform decisions in complex cases or when tissue ischemia is suspected despite modest anaemia.
Practical recommendations
– Adopt restrictive transfusion thresholds (Hb 7–8 g/dL) as the default for most clinically stable hospitalized adults and children to reduce unnecessary transfusions and their risks.
– In neurocritical care (acute traumatic brain injury, large ischemic stroke, intracerebral haemorrhage), consider higher Hb targets or a more liberal transfusion approach given data supporting improved neurologic outcomes with higher haemoglobin.
– In acute GI bleeding, current randomized data favour restrictive strategies; however, clinician judgement about ongoing hemorrhage, hemodynamics, and comorbidities remains essential.
– For suspected or confirmed acute myocardial ischemia, cardiogenic shock, severe symptomatic anemia, or inability to assess symptoms (e.g., unconscious patients), individualize transfusion thresholds using clinical context and physiologic monitoring.
– Use multimodal blood‑conservation measures: minimize unnecessary phlebotomy, correct coagulopathy, employ haemostatic techniques, and optimize erythropoietic support when appropriate.
Limitations and research gaps
Important limitations persist. Trial heterogeneity (patient populations, exact threshold definitions, transfusion triggers, co‑interventions) complicates subgroup interpretation. Long‑term functional outcomes beyond 12 months are sparsely reported. Paediatric data remain limited in scope and certainty. The role of physiologic triggers requires better‑designed RCTs with standardized algorithms. Finally, many trials were pragmatic and unblinded, although mortality and hard clinical endpoints are less susceptible to measurement bias.
Future research priorities include targeted RCTs in acute brain injury with refined neurofunctional endpoints, trials in specific cardiac subgroups (e.g., acute MI with ongoing ischemia), robust paediatric studies, and evaluations that combine haemoglobin thresholds with physiologic markers and patient‑centred outcomes (functional status, quality of life, cognitive performance).
Expert commentary and alignment with guidelines
The Cochrane 2025 synthesis aligns with longstanding guidance from the American Association of Blood Banks (AABB) and major practice statements that endorse restrictive transfusion thresholds for most stable hospitalized adults. Landmark randomized trials, including TRICC (Hebert et al., 1999), FOCUS (Carson et al., 2011), and Villanueva et al. (2013) inform contemporary practice. The Cochrane update importantly clarifies subgroup effects—most notably the neurocritical care signal—which should prompt guideline panels to consider context‑specific recommendations rather than a single universal Hb target.
Conclusion
Restrictive red cell transfusion thresholds (typically Hb 7–8 g/dL) represent a safe, effective default strategy across a wide range of clinical settings, substantially reducing transfusion exposure without increasing 30‑day mortality. Exceptions exist: neurocritically ill patients appear to benefit from higher haemoglobin targets for better long‑term neurological outcomes, and GI bleeding trials favour restrictive transfusion in that context. Clinical decisions should integrate Hb values with the patient’s clinical state, ongoing bleeding, tissue‑oxygenation data, and individual comorbidities. Research to refine physiologic triggers and to delineate subgroup‑specific thresholds remains a high priority.
Funding and trial registration
The Cochrane review (Carson et al., 2025) synthesizes randomized trials with variable sources of funding; trial‑level funding and registration details are reported in the original review and individual trial manuscripts. Clinicians should consult the primary publications for trial‑specific funding statements and ClinicalTrials.gov or trial registry identifiers where applicable.
References
Hebert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. New England Journal of Medicine. 1999;340(6):409–417. doi:10.1056/NEJM199902113400601.
Carson JL, Stanworth SJ, Dennis JA, et al. Transfusion thresholds and other strategies for guiding red blood cell transfusion. Cochrane Database of Systematic Reviews. 2025 Oct 20;10(10):CD002042. doi:10.1002/14651858.CD002042.pub6.
Carson JL, Terrin ML, Noveck H, et al. Liberal or restrictive transfusion in high‑risk patients after hip surgery. New England Journal of Medicine. 2011;365(26):2453–2462. doi:10.1056/NEJMoa1012452.
Villanueva C, Colomo A, Bosch A, et al. Transfusion strategies for acute upper gastrointestinal bleeding. New England Journal of Medicine. 2013;368(1):11–21. doi:10.1056/NEJMoa1206145.
American Association of Blood Banks (AABB). Clinical Practice Guidelines: Red Blood Cell Transfusion Thresholds and Strategies. 2016.
Thumbnail image prompt
“A compassionate scene in a modern hospital: a neurointensivist and a transfusion medicine physician discussing a brain CT on a lightbox while a blood bag on a transfusion stand hangs nearby; subtle infographic overlay showing haemoglobin thresholds (7–8 g/dL versus 9–10 g/dL); clinical, documentary style, cool hospital lighting, realistic detail.”