Prehospital Blood: A Review of the Evidence

Working review, compiled August 2026. Built from the trial reports rather than from summaries.

Four claims, in descending order of how well the evidence supports them:

• Whole blood does not outperform components at up to two units. Two trials, both null, read directly.
• The case that prehospital transfusion beats standard care is weaker than it is usually presented, because one of the two trials that supposedly disagree was never powered to answer the question.
• Three trials have produced three incompatible coagulation findings.
• No randomized trial has enrolled a nontraumatic hemorrhage patient. What proportion of ground transfusion goes to that population is not reliably known.

1. Whole blood against components

• Two randomized trials, TOWAR and SWiFT, reported in 2026. Neither found benefit.
• The trials differ substantially: different primary outcomes, different populations, different comparators. They agree in direction, not in design.
• TOWAR's severe hypotension subgroup is the finding most likely to be overlooked. Among 543 patients with systolic pressure at or below 70 mmHg, 30-day mortality was 34.5% with whole blood against 24.1% with components, odds ratio 1.59, 95% CI 1.02 to 2.47. The interval excludes 1. The authors report no apparent differences across the seven prespecified subgroups, which is defensible on interaction testing and multiplicity, but the nominal signal is there and it is in the sickest patients.
• TOWAR's mortality estimates favor components at every timepoint, with the largest effect earliest: 3 hours OR 1.72, 6 hours 1.55, 24 hours 1.25, 30 days 1.24. None significant. The consistency of direction is worth more attention than any single estimate.
• TOWAR reports two p-values for the primary outcome. The prespecified Donner-Klar test gave a between-group difference of 0.082, P = 0.08. The mixed-effects model gave P = 0.24. Both null; different impressions.
• Crossover was substantial in TOWAR. 45 patients assigned whole blood received none, 31.6% of the whole blood arm received additional components, and 26 component-arm patients received whole blood. The authors list this among factors raising type II error risk.
• Both trials capped whole blood at two units. Neither capped the comparator the same way: TOWAR's component arm had no protocol-defined minimum or limit, and SWiFT's standard care arm received up to two units each of red cells and plasma, prespecified as volume-equivalent to two units of whole blood.
• SWiFT's authors state that the dose of up to two units may have been insufficient to demonstrate an effect.
• TOWAR's storage age substudy was null through 21 days: 27.1% mortality at 15 to 21 days against 26.4% at 1 to 14 days, adjusted OR 0.99, CI 0.74 to 1.32. Operationally this is the most immediately useful result in either trial.

Table 1. TOWAR and SWiFT

ISS, injury severity score. SBP, systolic blood pressure. HR, heart rate. mITT, modified intention to treat. OR, odds ratio. RR, relative risk. CI, confidence interval.

Table 2. Mortality by timepoint

Every point estimate in both trials favors the comparator. None is significant. Confidence intervals are unadjusted for multiplicity and should not be read as hypothesis tests.

2. Transfusion against standard care

• This evidence is usually described as discordant, with PAMPer positive and COMBAT and RePHILL null. That description does not survive reading the trials.
• COMBAT was not powered for mortality. Its sample size was calculated to detect differences in international normalized ratio and thromboelastography G index. On those parameters it could only have detected a mortality difference of 19 percentage points, from 25% to 6%. It was stopped for futility after 144 of a planned 150 enrolments.
• So COMBAT did not contradict PAMPer on mortality. It was never able to speak to it. Listing the two side by side as opposing mortality results misrepresents both.
• PAMPer's result is real but fragile. 30-day mortality 23.2% against 33.0%, absolute difference 9.8 percentage points, 95% CI 18.6 to 1.0. The upper bound is a single percentage point.
• PAMPer's plasma group also received less crystalloid, median 500 vs 900 mL, and fewer prehospital red cell transfusions, 26.1% vs 42.1%. Both imbalances favor plasma. Adjustment for them leaves the effect intact, adjusted OR 0.61, CI 0.40 to 0.91, but the trial was unblinded and cluster-randomized.
• The mechanism is unclear. A roughly 10 percentage point mortality reduction from 600 mL of plasma is a large effect for a small intervention, and the INR difference of 1.2 against 1.3 does not plausibly account for it.
• Severity scores are not comparable across these trials. COMBAT reports New Injury Severity Score, which sums the three highest AIS scores regardless of body region and therefore runs higher than ISS on the same patient. PAMPer and RePHILL report ISS.
• Transport interval remains the most plausible reconciliation. COMBAT ran 16 to 19 minutes scene to hospital in urban Denver. Its authors conclude explicitly that plasma may benefit austere settings with longer transport times and that the financial burden is not justified in an urban environment close to mature trauma centers.

Table 3. PAMPer, COMBAT and RePHILL

NISS, New Injury Severity Score. pp, percentage points.

3. The coagulation problem

• SWiFT: prothrombin time above the normal range in 40.7% of the whole blood group against 30.5% of standard care, RR 1.31, CI 1.10 to 1.56. Interval excludes 1.
• COMBAT: INR above 1.3 on hospital arrival in 44% of the plasma group against 24% of controls, RR 1.84, CI 1.08 to 3.14, P = 0.02. Interval excludes 1.
• TOWAR: INR above 1.5 within 1 hour of arrival in 20.3% of the whole blood group against 26.4% of components, OR 0.69, CI 0.46 to 1.02. Points the other way and nearly excludes 1.
• Three trials, three coagulation findings, two of them significant and in opposite directions from each other.
• Each set of authors offers a local explanation. SWiFT attributes its finding to plasma age, up to 21 days inside whole blood units against under 5 days in the standard care arm. COMBAT attributes its finding to plasma units themselves having INR above 1.3, so plasma cannot correct a mildly raised INR. TOWAR offers no explanation, since its finding is in the expected direction.
• I have not found any work reconciling the three, though I have not searched systematically. That the interventions intended to correct coagulopathy repeatedly worsened laboratory coagulation, without any corresponding difference in clinical outcome, is either an artifact of what these assays measure in this setting or a signal about product age. Both possibilities matter operationally.
• COMBAT also found that coagulopathy was largely absent at the scene: INR above 1.3 in 2 of 36 plasma and 2 of 29 control patients. The intervention was correcting a derangement most enrolled patients did not have.

Table 4. Coagulation findings across three trials

PT, prothrombin time. INR, international normalized ratio.

4. Nontraumatic hemorrhage

• No randomized trial has enrolled a patient transfused for nontraumatic hemorrhage. All four trials above excluded them by design.
• SWiFT enrolled 33 non-trauma patients anyway and excluded them from analysis, because the trial boxes contained the only blood products most participating services carried. The trial could not keep nontraumatic patients out of its own blood supply.
• No governing document addresses nontraumatic indications. See section 5.
• The available outcome literature is retrospective and physiologic. Transfusion improves hemoglobin and INR. Where an untransfused comparison group exists, mortality, ICU days, length of stay, and rate of intervention did not differ.
• What share of ground transfusion is nontraumatic is not reliably known, and the estimates in circulation do not support the precision with which they are quoted.

On the utilization estimates. The most-cited figure is 46.5% (95% CI 34.3 to 59.2), from Mapp 2020. Read directly, it will not carry the weight placed on it:

• It is 58 consecutive transfusions from two Texas services between September 2017 and December 2018, not a cohort study.
• The combined figure masks near-total disagreement between the two services: HCESD 48 at 68.0% (48.4 to 82.8) nontraumatic, San Antonio Fire Department at 30.3% (17.4 to 47.3).
• Both protocols restricted transfusion to males aged 12 and over and females over 50, to limit RhD alloimmunization risk. Screening out younger women structurally inflates the nontraumatic share.
• HCESD 48 is a suburban service of 150,000 whose territory the authors describe as containing a large number of skilled nursing, memory care, and assisted living facilities. Its high GI bleeding proportion follows from its catchment.
• The authors state the findings are specific to South Texas and not generalizable, and that observational studies of this kind should not be used to justify widespread adoption of whole blood programs.

Other frequently quoted figures (Treichel 2025 at 43%, Hendley 2025 at 12.1%) come from a conference summary rather than the papers, and the two may not share a denominator: one reports nontraumatic share, the other medical conditions. Until the primary sources are read, neither belongs in a table.

5. Current guidance

Three documents govern most civilian practice. All are scoped to trauma. All predate TOWAR and SWiFT.

• 2023 joint position statement, American College of Surgeons Committee on Trauma, American College of Emergency Physicians, and NAEMSP. Patients with signs of hemorrhagic shock should receive prehospital blood products whenever available. Whole blood is preferred over packed red cells; if only components are available, transfuse 1:1. Supporting claim is a 37% reduction in 30-day mortality drawn from observational literature. Every listed indication is trauma-specific. No nontraumatic indication appears in the document.
• Treatment specifics from the same statement: 1 g calcium gluconate per 1 to 2 units transfused, on the basis that prehospital transfusion in civilian trauma is associated with hypocalcemia, which itself predicts decreased survival and massive transfusion requirement; and transfusion through a blood warmer to a delivery temperature of 38°C, not exceeding 42°C.
• 2025 NAEMSP Prehospital Trauma Compendium on transfusion. Three PICO questions, all trauma. Recommends blood components over crystalloid. Suggests LTOWB as first-choice product. The distinction between recommend and suggest is deliberate and appears in the same sentence of the conclusion.
• The compendium's literature search ran January 2000 to February 2023. Its evidence base closed more than two years before publication and misses most of 2023 to 2025 as well as both 2026 trials.
• The compendium performed no formal risk of bias assessment and no grading of evidence strength, citing resource limitations. Screening and full-text extraction were single-reviewer with a 5% verification sample.
• The compendium is explicit that its LTOWB recommendation is interim. It states there are no data supporting components over LTOWB prehospital, that ongoing randomized trials will determine efficacy and safety, and that until those trials report the recommendation rests on biologic rationale, safety, and logistics. TOWAR and SWiFT are those trials. The condition the authors named has now been met.
• 2025 Prehospital Blood Transfusion Coalition clinical practice guideline for civilian EMS. Not read; listed for completeness.

6. What this leaves

• Whole blood has no demonstrated mortality advantage over components at up to two units, with a nominal signal of harm in the most severely hypotensive patients in one trial.
• Whether it helps beyond two units is untested. TROOP is enrolling roughly 1,100 patients with a 6-hour primary outcome, expected to report around mid-2027.
• Whether prehospital transfusion beats standard care at all rests substantially on one positive trial with a confidence interval reaching 1 percentage point, and a companion trial that was never powered to check it.
• Transport interval is the most plausible modifier and the one most relevant to a rural or long-transport service. It has never been tested prospectively.
• Coagulation findings are incoherent across trials and unexplained.
• An entire population receiving this product has no trial evidence, no guidance, and no reliable utilization estimate.
• None of this argues against running a program. It argues against justifying one on a mortality benefit the trials did not find, and against assuming that trauma guidance transfers to the gastrointestinal bleed at three in the morning.

Sources and other data

Sources

• TOWAR. Sperry JL, Guyette FX, Cotton BA, et al. Type O whole blood prehospital resuscitation for trauma and hemorrhage. N Engl J Med. 2026;394(23):2317-2328. doi:10.1056/NEJMoa2602167
• SWiFT. Smith JE, Cardigan R, Sanderson E, et al. Prehospital whole blood in traumatic hemorrhage: a randomized controlled trial. N Engl J Med. 2026. doi:10.1056/NEJMoa2516043
• PAMPer. Sperry JL, Guyette FX, Brown JB, et al. Prehospital plasma during air medical transport in trauma patients at risk for hemorrhagic shock. N Engl J Med. 2018;379(4):315-326.
• COMBAT. Moore HB, Moore EE, Chapman MP, 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.
• RePHILL. Crombie N, Doughty HA, Bishop JRB, et al. Resuscitation with blood products in patients with trauma-related haemorrhagic shock receiving prehospital care. Lancet Haematol. 2022.
• Mapp JG, Bank EA, Osborn LA, Stringfellow ML, Reininger DW, Winckler CJ. Epidemiological and accounting analysis of ground ambulance whole blood transfusion. Prehosp Disaster Med. 2020;35(1):98-103.
• Berry C, Gallagher JM, Goodloe JM, Dorlac WC, Dodd J, Fischer PE. Prehospital hemorrhage control and treatment by clinicians: a joint position statement. Prehosp Emerg Care. 2023;27(5):544-551.
• Brown JB, Yazer MH, Kelly J, Spinella PC, DeMaio V, Fisher AD, Cap AP, Winckler CJ, Beltran G, Martin-Gill C, Guyette FX. Prehospital trauma compendium: transfusion of blood products in trauma. A position statement and resource document of NAEMSP. Prehosp Emerg Care. 2025. doi:10.1080/10903127.2025.2476195

Other data

• Rogers SO. Prehospital whole blood for traumatic hemorrhage: consistent evidence from two randomized trials. N Engl J Med. 2026;394(23):2372-2373. The consistent-evidence framing is widely repeated and sits awkwardly against the trials' differing populations and primary outcomes.
• Levy MJ, Schaefer RM, O'Byrne H, Krohmer JR, Bank EA, Holcomb JB. Prehospital blood transfusion coalition clinical practice guideline for civilian emergency medical services. Trauma Surg Acute Care Open. 2025;10(3):e001931.
• Thiels CA, Aho JM, Fahy AS, et al. Prehospital blood transfusions in nontrauma patients. World J Surg. 2016;40(10):2297-2304. The published abstract contradicts itself on the trauma and non-trauma split; the subgroup arithmetic indicates 549 non-trauma of 857 total.
• Parker ME, Khasawneh MA, Thiels CA, et al. Prehospital transfusion for gastrointestinal bleeding. Air Med J. 2017;36(6):315-319. Shares authors and overlapping enrolment dates with Thiels 2016; the two cohorts may not be independent.
• Treichel 2025 and Hendley 2025, both quoted for nontraumatic utilization shares. Primary citations not established; both trace to a single conference summary, and the two may not share a denominator.
• NEMSIS blood report 2026 and the ImageTrend EMS Insights Report 2026. The figures in circulation do not reconcile: 62% of 22,000 ground transfusions is about 28% of 48,000, not the 23% reported.
• Sussex HEMS nontraumatic series 2025, and a specialized prehospital response vehicle cohort reporting 12 nontraumatic recipients of 57.
• Carico C, et al. Nationwide trends in prehospital blood product use after injury 2020-2023. Reports that fewer than 1% of hemodynamically unstable trauma patients meeting transfusion criteria received prehospital blood. This reframes the utilization picture considerably.
• AHRQ systematic review on prehospital EMS blood transfusion and fluid interventions, in progress.