Publication

• Title: Prehospital Resuscitation with Type O Whole Blood for Trauma and Hemorrhage
• Acronym: TOWAR
• Year: 2026
• Journal published in: New England Journal of Medicine
• Citation: Sperry JL, Guyette FX, Cotton BA, Luther JF, Utarnachitt RB, Kutcher ME, et al; TOWAR Study Group. Prehospital resuscitation with type O whole blood for trauma and hemorrhage. N Engl J Med. Published online May 18, 2026.

Context & Rationale

• Background

  • Major traumatic haemorrhage remains a leading cause of potentially preventable early trauma death.
  • Modern trauma resuscitation has moved away from crystalloid-heavy care towards earlier haemostatic resuscitation with blood products.
  • Prehospital transfusion has biologic and temporal plausibility: the intervention occurs before definitive haemorrhage control, before severe trauma-induced coagulopathy is fully established, and before shock becomes irreversible.
  • Prior prehospital blood-product trials had produced context-dependent results: PAMPer suggested benefit from prehospital plasma during air medical transport, whereas COMBAT did not show survival benefit during rapid urban ground transport.¹²
  • Low-titre group O whole blood is attractive because it provides red cells, plasma, and platelets in a single product, with simpler logistics than separate components and potential haemostatic advantages.
  • Before TOWAR, civilian whole-blood evidence was dominated by observational studies and a small prehospital-to-in-hospital pilot trial rather than large definitive randomised comparisons.³
  • The storage age of whole blood was an important operational uncertainty, because older stored whole blood may have altered platelet and coagulation function, while using older units could reduce waste and improve programme feasibility.
• Research Question/Hypothesis

  • TOWAR asked whether prehospital transfusion of up to 2 units of low-titre group O whole blood reduced 30-day all-cause mortality compared with prehospital transfusion using as-indicated blood components.
  • It also asked whether whole blood stored for 15 to 21 days was associated with similar outcomes to whole blood stored for 1 to 14 days.
  • The investigators hypothesised that whole blood would reduce 30-day mortality compared with component therapy.
• Why This Matters

  • Prehospital whole blood requires blood-bank infrastructure, donor screening, cold-chain management, product rotation, traceability, training, and governance.
  • A mortality benefit would support widespread civilian air medical adoption and inform military far-forward blood-supply planning.
  • A neutral result would not refute prehospital transfusion, but would narrow claims that whole blood is clinically superior to component-based haemostatic resuscitation.
  • The storage-age substudy had immediate practical implications for wastage, shelf-life policy, and the feasibility of smaller whole-blood programmes.

Design & Methods

• Research Question: In injured adult patients at risk of haemorrhagic shock who were selected for prehospital blood-product transfusion during air medical transport, does transfusion of up to 2 units of low-titre group O whole blood reduce 30-day all-cause mortality compared with transfusion using blood components?
• Study Type: Pragmatic, multicentre, phase 3, open-label, cluster-randomised trial conducted from May 2022 to June 2025 across 44 air medical bases transporting patients to 11 trauma centres. Air medical bases were assigned in a 2:1 ratio during fixed 1-month blocks to whole blood or component therapy. Enrolment occurred under exception from informed consent with community consultation, public disclosure, and post-enrolment notification.
• Population:
  • Adults aged 18 to 90 years.
  • Injured patients transported from the scene or a referring emergency department to a participating trauma centre.
  • Patients had initiation of blood or blood-component transfusion and met physiologic criteria for haemorrhagic shock risk.
  • Physiologic inclusion required either systolic blood pressure ≤90 mm Hg with heart rate ≥108 beats/min, or systolic blood pressure ≤70 mm Hg irrespective of heart rate.
  • Key exclusions included isolated standing-height fall, drowning or hanging, isolated burns without traumatic injury, pregnancy, legal detention, traumatic arrest with >5 consecutive minutes of CPR without return of vital signs, penetrating brain injury or exposed brain matter, inability to obtain intravenous or intraosseous access, patient/family objection, or a “NO TOWAR” opt-out bracelet.
  • Receipt of blood products before randomisation was not an exclusion criterion.
• Intervention:
  • Air medical bases assigned to the intervention were stocked with 2 units of low-titre group O whole blood.
  • Low titre was defined as anti-A and anti-B antibody titres <256.
  • Eligible patients received up to 2 prehospital units of whole blood.
  • Administration of the second unit was guided by clinical status and site protocol.
  • Whole blood shelf life in the trial was up to 21 days.
  • No other aspect of prehospital or in-hospital care was protocolised.
• Comparison:
  • Air medical bases assigned to the comparator were stocked with blood components: red cells, plasma, or both.
  • Eligible patients received component therapy according to standard transfusion practice without protocol-defined minimums or maximums.
  • Subsequent prehospital and in-hospital care remained at the discretion of treating clinicians.
• Blinding: The intervention was open-label because blood-product administration required documentation, traceability, and look-back procedures. Personnel assessing trial outcomes were unaware of treatment assignment. The primary outcome was objective, but open-label treatment still allowed possible performance bias in co-interventions and downstream transfusion decisions.
• Statistics: A total of 1020 patients from 40 participating sites was required to detect a 10 percentage-point absolute reduction in 30-day mortality, from 26% to 16%, with 80% power, a two-sided type I error rate of 0.05, and an intraclass correlation coefficient of 0.02. The primary comparison used a two-sided Donner-Klar test for proportions, with mixed-effects logistic regression adjusted for trial month and baseline imbalances. Analyses were modified intention-to-treat: patients who withdrew after emergency enrolment were excluded.
• Follow-Up Period: Primary outcome assessment was 30 days after randomisation. Secondary outcomes included mortality within 3, 6, and 24 hours, in-hospital mortality, blood-product use during the first 24 hours after arrival, haemostasis, organ failure, infection, ARDS, coagulation measures, adverse events, and the whole-blood storage-age substudy.

Key Results

This trial was not stopped early. It completed planned enrolment: 9554 trauma patients were assessed for eligibility, 1020 underwent cluster randomisation, and 993 were included in the modified intention-to-treat primary analysis after 27 post-randomisation withdrawals.

• The primary result was clearly neutral for benefit and numerically unfavourable to whole blood: 25.9% versus 20.5% 30-day mortality; adjusted OR 1.24; 95% CI 0.87 to 1.76; P=0.24.
• No secondary clinical outcome showed a convincing benefit of whole blood, including early mortality, haemorrhagic death, 24-hour transfusion requirements, haemostasis, organ failure, infection, or ARDS.
• The storage-age substudy supports operational use of 15–21-day whole blood within this trial context, but it does not establish randomised equivalence and does not address longer 35-day storage strategies.

Internal Validity

• Randomisation and Allocation: Randomisation was cluster-based at the air medical base level during 1-month blocks rather than patient-level allocation. This was logistically sensible for prehospital blood-product stocking and wastage control, but it means treatment assignment was known operationally and could theoretically influence enrolment, transfusion initiation, or transport behaviour.
• Dropout and Exclusions: The trial enrolled 1020 patients, but 27 were excluded from the modified intention-to-treat analysis because of withdrawal after emergency enrolment. Primary outcome data were otherwise highly complete: 972/993 had known 30-day survival status before imputation.
• Missing Data: Missing primary outcome data were handled conservatively by imputing missing vital status as survival, and sensitivity analyses gave similar absolute risk differences. Mechanistic secondary outcomes were less robust: thromboelastography and platelet mapping had substantial missingness, limiting interpretation of haemostatic mechanisms.
• Performance and Detection Bias: Blinding of treating teams was not feasible. Outcome assessors were unaware of treatment assignment, and the primary endpoint was objective. However, open-label care could influence co-interventions, component supplementation, in-hospital transfusion decisions, and cause-of-death attribution.
• Protocol Adherence: Assigned-product transfusion occurred in 964/1020 patients (94.5%), which is strong for a prehospital emergency trial. However, crossover and mixed-product exposure were clinically important: 226 whole-blood-assigned patients received components, and 26 component-assigned patients received whole blood.
• Protocol Deviations: The supplement reported 365 deviations in the whole-blood group and 152 in the component group. Major categories included first 24-hour laboratory samples not collected, samples drawn out of window, enrolment of ineligible patients, consent-process issues, and non-adherence to the assigned study arm.
• Baseline Characteristics: Groups were broadly similar, but the whole-blood group had numerically more traumatic brain injury (41.4% vs 35.7%), more prehospital intubation (54.4% vs 48.4%), lower median Glasgow Coma Scale score (11 vs 13), and slightly higher median Injury Severity Score (25 vs 23). The primary model adjusted for traumatic brain injury and prehospital intubation.
• Severity: The cohort was clinically severe enough to test a mortality intervention: median Injury Severity Score was 25, 39.7% had traumatic brain injury, 52.5% underwent prehospital intubation, 62.2% underwent operative intervention within 24 hours, and overall 30-day mortality was 24.3%.
• Heterogeneity: Heterogeneity was substantial and pragmatic. The population included blunt and penetrating trauma, scene and interfacility transports, traumatic brain injury and non-traumatic brain injury phenotypes, variable transport contexts, and variable component strategies. This improves real-world relevance but may dilute a benefit in a narrower haemorrhagic phenotype.
• Timing: The intervention was delivered in the intended prehospital phase, mostly after scene response: 70.8% were transported from the scene of injury and 29.2% from another hospital. This timing was appropriate for a haemostatic-resuscitation intervention, but some patients had already received blood products before randomisation, which reduced the clean contrast between groups.
• Dose: The whole-blood dose was capped at up to 2 units, with the second unit guided by clinical condition and site protocol. This may have been insufficient to overcome severe haemorrhagic shock, especially when subsequent in-hospital transfusion volumes and haemorrhage-control procedures dominated later resuscitation.
• Separation of the Variable of Interest:
  • In the whole-blood-assigned group, 343 patients (48.0%) received 1 unit of whole blood and 326 (45.6%) received 2 units; 46 (6.4%) received no whole blood.
  • In the component-assigned group, 136 patients (44.6%) received 1 unit of red cells and 129 (42.3%) received 2 units; 117 (38.4%) received 1 unit of plasma and 93 (30.5%) received 2 units; 175 (57.4%) received both red cells and plasma.
  • Estimated prehospital transfusion volume differed by product type: median whole-blood volume was 500 mL (IQR 500 to 1000) in the whole-blood group versus 0 mL in the component group; median plasma volume was 0 mL versus 275 mL; median red-cell volume was 0 mL versus 330 mL.
• Key Delivery Aspects: The comparator was pragmatic and clinically realistic, but it was not a single standardised regimen. Component-assigned bases could provide plasma, red cells, or both. This makes the trial a real-world product-strategy comparison rather than a precise physiologic comparison of whole blood versus balanced red-cell-plus-plasma resuscitation.
• Crossover: Crossover was important. Whole-blood-assigned patients could receive components and component-assigned patients could receive whole blood. This likely biased towards the null, but the per-protocol analysis also failed to suggest benefit and was numerically more unfavourable to whole blood.
• Adjunctive Therapy Use: Prehospital co-interventions were similar enough not to explain the primary result: intubation occurred in 54.4% versus 48.4%, CPR in 10.4% versus 7.3%, tourniquet use in 13.1% versus 16.1%, pelvic binder use in 21.7% versus 22.0%, tranexamic acid in 41.8% versus 37.5%, and crystalloids or colloids in 62.6% versus 63.4%.
• Outcome Assessment: All-cause 30-day mortality is objective, patient-important, and resistant to ascertainment bias. Some secondary outcomes, especially cause-specific mortality and time to haemostasis, are more vulnerable to site-level judgement and clinical practice variation.
• Statistical Rigor: The sample size and cluster design were prespecified, and the primary analysis accounted for clustering and baseline imbalances. The trial was powered for a large 10 percentage-point absolute mortality difference, not for smaller effects. Secondary and subgroup analyses were not adjusted for multiplicity and should be interpreted as exploratory.

Conclusion on Internal Validity: Internal validity is moderate to strong for the primary mortality question. Randomisation, objective outcome assessment, high primary follow-up, and prespecified analysis support credibility; cluster allocation, open-label care, crossover, pre-randomisation transfusion, comparator variability, modest whole-blood dose, and baseline imbalance limit confidence in excluding smaller or phenotype-specific effects.

External Validity

• Population Representativeness: The cohort resembles severely injured patients selected for prehospital transfusion in mature air medical trauma systems. The median Injury Severity Score was 25, most injuries were blunt, and most patients were transported from the scene.
• Comparison with Broader Trauma Populations: The supplementary comparison with the 2023 Trauma Quality Programs Participant Use File suggested that TOWAR patients were younger and more often male than the overall trauma population, but closer to the subset meeting TOWAR physiologic criteria.
• Applicability to Air Medical Systems: Generalisability is strongest for systems that already have prehospital blood-product capability, blood-bank support, temperature-controlled storage, product traceability, and trained air medical clinicians.
• Applicability to Ground EMS: Applicability is weaker for urban ground EMS with short transport times, limited blood-product capacity, or no plasma availability, because transport time and logistical context may alter the value of prehospital haemostatic resuscitation.
• Applicability to Systems Without Prehospital Components: TOWAR does not answer whether whole blood is better than no prehospital blood, crystalloid-only resuscitation, or red-cell-only prehospital transfusion. It answers whether whole blood outperformed a pragmatic component strategy among patients already receiving prehospital blood products.
• Excluded Populations: The results should not be extrapolated directly to children, pregnant patients, prisoners, isolated burns, isolated drowning or hanging, penetrating brain injury, prolonged traumatic arrest, isolated standing-height falls, or patients without vascular access.
• Military and Austere Settings: The findings are relevant to military planning because whole blood is logistically attractive in far-forward care, but civilian air medical systems with trauma-centre access are not equivalent to combat casualty care, prolonged field care, or mass-casualty blood-supply constraints.
• Storage-Age Applicability: The storage-age substudy applies to whole blood stored for up to 21 days. It should not be used to infer equivalence of 22- to 35-day whole-blood storage policies.

Conclusion on External Validity: External validity is good for mature North American air medical trauma systems providing prehospital blood products to severely injured adults. It is limited for paediatric trauma, pregnancy, prolonged field care, low-resource EMS, systems without component therapy, ground-only urban systems, and longer-storage whole-blood programmes.

Strengths & Limitations

• Strengths:
  • Large phase 3 randomised trial in a difficult prehospital emergency research setting.
  • Pragmatic design reflecting real-world air medical blood-product delivery.
  • Clinically important, objective primary outcome.
  • High primary outcome ascertainment.
  • Prespecified cluster-adjusted primary analysis.
  • Careful emergency research oversight, including FDA and Department of Defense oversight, external data and safety monitoring, community consultation, and public disclosure.
  • Embedded observational storage-age substudy addressing an important operational question.
  • Detailed reporting of actual products received, crossovers, protocol deviations, laboratory data, and safety events.
• Limitations:
  • Open-label intervention with potential for performance bias.
  • Cluster randomisation by base and month rather than patient-level randomisation.
  • Known treatment assignment could influence enrolment or transfusion decisions in the field.
  • Modified intention-to-treat analysis excluded 27 post-randomisation withdrawals.
  • Product crossover and mixed-product exposure reduced treatment separation.
  • Pre-randomisation blood-product transfusion was allowed, further diluting the intervention contrast.
  • Component therapy was heterogeneous: red cells, plasma, or both, without protocol-defined balance.
  • Whole-blood dose was limited to up to 2 units and may have been insufficient for the highest-risk haemorrhagic phenotypes.
  • The trial was powered for a 10 percentage-point mortality difference and may not exclude smaller clinically meaningful effects.
  • Secondary and subgroup analyses were exploratory and not multiplicity-adjusted.
  • Mechanistic outcomes were limited by missing thromboelastography and platelet-function data.
  • The storage-age substudy was observational rather than randomised.

Interpretation & Why It Matters

• Primary message

  TOWAR does not support a claim that up to 2 units of prehospital low-titre group O whole blood improves 30-day survival compared with pragmatic component-based prehospital transfusion.
• What it does not refute

  The trial does not refute prehospital blood-product transfusion. All enrolled patients were candidates for prehospital blood or component transfusion; the trial compared product strategies, not blood versus no blood.
• Clinical implication

  Whole blood may remain attractive for logistics, simplicity, donor exposure, and operational resilience, but TOWAR argues against presenting it as mortality-superior to component therapy in comparable air medical systems.
• Storage implication

  The storage-age substudy is practically important: whole blood stored 15 to 21 days did not appear clinically worse than 1- to 14-day whole blood, supporting product rotation and wastage reduction within a 21-day shelf-life programme.
• Research implication

  The next generation of trials should focus less on a universal “whole blood versus components” question and more on phenotype, timing, dose, comparator quality, transport interval, coagulopathy, traumatic brain injury, and system logistics.

Controversies & Subsequent Evidence

• The trial sits within a mixed prehospital transfusion evidence base: PAMPer showed lower 30-day mortality with prehospital plasma during air medical transport, whereas COMBAT found no survival benefit during rapid urban ground transport. A post hoc analysis of PAMPer and COMBAT suggested that transport time may be a key modifier, with plasma-associated survival benefit most apparent when transport exceeded 20 minutes.¹²⁴
• Whole-blood enthusiasm was biologically plausible but not yet definitive: The PPOWER pilot showed that prehospital low-titre group O whole blood was feasible and appeared safe, but it was not a definitive mortality trial. RePHILL, which tested prehospital red cells plus lyophilised plasma versus saline for traumatic haemorrhagic shock, also failed to establish superiority, reinforcing that prehospital blood-product interventions are sensitive to setting, dose, and comparator.³⁵
• Guideline momentum preceded definitive randomised evidence: The European trauma bleeding guideline acknowledged uncertainty around prehospital blood-product strategies. EAST subsequently conditionally recommended whole blood for adult civilian trauma patients receiving transfusion, and the NAEMSP position statement supported low-titre group O whole blood as the preferred prehospital product where a high-quality programme can support it. TOWAR substantially tempers the strength of any superiority claim behind that practice direction.⁶⁷⁸
• TOWAR and SWiFT are complementary negative trials: SWiFT tested up to 2 units of prehospital whole blood against a strong balanced comparator of red cells plus plasma and was also neutral for its primary outcome: 153/314 (48.7%) versus 144/302 (47.7%); RR 1.02; 95% CI 0.80 to 1.31; P=0.84. TOWAR compared whole blood with a more variable, as-indicated component strategy and still found no survival benefit.⁹
• Observational meta-analytic evidence remains favourable but heterogeneous: An updated 2026 JAMA Surgery meta-analysis found an association between whole blood and reduced mortality in civilian trauma, but with wide prediction intervals and substantial heterogeneity. Its search ended before the publication of TOWAR and SWiFT, so these two large neutral randomised trials should materially reshape subsequent evidence syntheses.¹⁰
• The field is moving towards personalised resuscitation: The invited commentary accompanying the updated meta-analysis argued for moving from broad product preference to more personalised acute resuscitation. TOWAR supports that direction: the clinically relevant question is likely not “whole blood for everyone”, but “which bleeding phenotype, in which system, at what dose, and against which comparator?”¹¹
• The severe hypotension subgroup is clinically provocative but not practice-changing: In patients with systolic blood pressure ≤70 mm Hg, mortality was 34.5% with whole blood versus 24.1% with components; adjusted OR 1.59; 95% CI 1.02 to 2.47. This should be interpreted as hypothesis-generating because subgroup analyses were underpowered and not multiplicity-adjusted; nevertheless, it argues against assuming that the sickest shock phenotype necessarily benefits most from a 2-unit whole-blood strategy.
• Comparator heterogeneity complicates interpretation: TOWAR did not compare whole blood with a fixed 1:1 red-cell-and-plasma strategy. Component therapy varied between red cells, plasma, or both, and 57.4% of component-assigned patients received both red cells and plasma. This makes the result pragmatic but less mechanistically clean.
• Dose remains unsettled: A 2-unit prehospital whole-blood ceiling may be too low for patients with major haemorrhage, especially when in-hospital haemorrhage control and massive transfusion dominate later outcomes. The TROOP trial protocol addresses a related but different question: low-titre group O whole blood versus component therapy in patients predicted to require large-volume transfusion in level I trauma centres.¹²
• Storage-age findings are useful but bounded: The observational substudy reassures about 15–21-day whole blood within this trial’s supply chain, but it cannot exclude unmeasured confounding and cannot be extrapolated to 35-day whole-blood storage systems.

Summary

• TOWAR randomised 1020 severely injured adults transported by 44 air medical bases to receive up to 2 units of prehospital low-titre group O whole blood or as-indicated component therapy.
• The primary outcome was negative: 30-day mortality was 25.9% with whole blood versus 20.5% with components; adjusted OR 1.24; 95% CI 0.87 to 1.76; P=0.24.
• Secondary clinical outcomes, including early mortality, haemorrhagic death, transfusion requirements, haemostasis, organ failure, infection, ARDS, and coagulation outcomes, did not show a convincing whole-blood benefit.
• The storage-age substudy found no apparent difference in 30-day mortality between whole blood stored for 15–21 days and 1–14 days: 27.1% versus 26.4%; adjusted OR 0.99; 95% CI 0.74 to 1.32.
• The trial supports the feasibility and apparent safety of prehospital whole-blood programmes but not their survival superiority over component-based prehospital transfusion.

Overall Takeaway

TOWAR is a landmark prehospital transfusion trial that did not demonstrate a survival advantage for whole blood over component therapy. It shows that in mature air medical trauma systems, up to 2 units of low-titre group O whole blood did not improve 30-day survival compared with pragmatic blood-component transfusion, shifting the argument for whole blood from mortality superiority towards feasibility, logistics, stewardship, and selective use.

Bibliography

• 1.Sperry JL, Guyette FX, Brown JB, Yazer MH, Triulzi DJ, Early-Young BJ, et al; PAMPer Study Group. Prehospital plasma during air medical transport in trauma patients at risk for hemorrhagic shock. N Engl J Med. 2018;379(4):315-326.
• 2.Moore HB, Moore EE, Chapman MP, McVaney K, Bryskiewicz G, Blechar R, et al. Plasma-first resuscitation to treat haemorrhagic shock during emergency ground transportation in an urban area: a randomised trial. Lancet. 2018;392(10144):283-291.
• 3.Guyette FX, Zenati M, Triulzi DJ, Yazer MH, Skroczky H, Early BJ, et al. Prehospital low titer group O whole blood is feasible and safe: results of a prospective randomized pilot trial. J Trauma Acute Care Surg. 2022;92(5):839-847.
• 4.Pusateri AE, Moore EE, Moore HB, Le TD, Guyette FX, Chapman MP, et al. Association of prehospital plasma transfusion with survival in trauma patients with hemorrhagic shock when transport times are longer than 20 minutes: a post hoc analysis of the PAMPer and COMBAT clinical trials. JAMA Surg. 2020;155(2):e195085.
• 5.Crombie N, Doughty HA, Bishop JRB, Desai A, Dixon EF, Hancox JM, et al; RePHILL Trial Collaborators. Resuscitation with blood products in patients with trauma-related haemorrhagic shock receiving prehospital care (RePHILL): a multicentre, open-label, randomised, controlled, phase 3 trial. Lancet Haematol. 2022;9(4):e250-e261.
• 6.Rossaint R, Afshari A, Bouillon B, Cerny V, Cimpoesu D, Curry N, et al. The European guideline on management of major bleeding and coagulopathy following trauma: sixth edition. Crit Care. 2023;27:80.
• 7.Meizoso JP, Cotton BA, Lawless RA, et al. Whole blood resuscitation for injured patients requiring transfusion: a systematic review, meta-analysis, and practice management guideline from the Eastern Association for the Surgery of Trauma. J Trauma Acute Care Surg. 2024;97(3):460-470.
• 8.Brown JB, Yazer MH, Kelly J, Spinella PC, DeMaio V, Fisher AD, et al. Prehospital Trauma Compendium: Transfusion of Blood Products in Trauma - A Position Statement and Resource Document of NAEMSP. Prehosp Emerg Care. Published online April 1, 2025.
• 9.Smith JE, Cardigan R, Sanderson E, et al. Prehospital whole blood in traumatic hemorrhage: a randomized controlled trial. N Engl J Med. Published online March 17, 2026.
• 10.Ibrahim W, Meza Monge K, Menzel J, Morales Ojeda L, Dalton M, Fuenmayor Lozada DA, et al. Whole-blood vs component therapy in adult trauma: an updated systematic review and meta-analysis. JAMA Surg. 2026;161(5):497-506.
• 11.Wallen TE, Holcomb JB. From balanced blood products to personalized medicine in acute resuscitation of trauma patients. JAMA Surg. 2026;161(5):507.
• 12.Jansen JO, Pedroza C, Novelo LL, Hao T, DeWildt GR, Coton C, et al. Trauma resuscitation with low-titer group O whole blood or products: study protocol for a randomized clinical trial (the TROOP trial). Trials. 2025;26:266.