Pre-Hospital Whole Blood in Traumatic Haemorrhage

Background

The idea of giving whole blood early in trauma has an almost irresistible appeal. It feels right, it is logistically elegant, and most importantly it appears to align with the direction of travel in modern haemostatic resuscitation. But as always in trauma, what feels right doesn’t always survive contact with good evidence. We saw this from the RePHILL trial (discussed on St Emlyns here), where although blood transfusion feels right for bleeding patients over saline resuscitation, the evidence didn’t show a benefit in that paper.

The Study of Whole Blood in Frontline Trauma (SWiFT) trial, recently published in the New England Journal of Medicine, attempts to answer the question: is prehospital whole blood better than component therapy in traumatic haemorrhage? The full paper is here. As always, we encourage you to read it in full.

The Abstract

Background – Whole-blood transfusion has recently gained favor in the management of severe hemorrhage; however, data from large clinical trials evaluating its clinical effectiveness and safety are lacking.

Methods – We conducted a pragmatic, phase 3, multicenter, unblinded, randomized, superiority trial across 10 air ambulance services in England. Patients with major traumatic hemorrhage who were attended by a participating air ambulance service were randomly assigned to receive either whole-blood transfusion (up to 2 units) or standard care with blood components (up to 2 units each of red cells and plasma) before arrival at the hospital. The primary outcome was a composite of death from any cause or massive transfusion (≥10 units of blood components or products) within 24 hours after randomization.

Results – A total of 942 patients underwent randomization. After the exclusion of participants with nontraumatic hemorrhage or traumatic cardiac arrest, 616 were included in the analysis (314 in the whole-blood group and 302 in the standard-care group). A primary-outcome event occurred in 48.7% of the participants in the whole-blood group and in 47.7% of those in the standard-care group (relative risk, 1.02; 95% confidence interval, 0.80 to 1.31; P=0.84). The incidence of death from any cause at all time points, massive transfusion, and other secondary outcomes appeared to be similar in the two groups. Prothrombin times were above the normal range in 40.7% of the participants in the whole-blood group and in 30.5% of those in the standard-care group. More serious adverse events occurred in the standard-care group than in the whole-blood group (37 and 31, respectively). The incidence of thrombotic events appeared to be similar in the two groups.

Conclusions – Among participants with life-threatening hemorrhage, prehospital transfusion of 2 units of whole blood was not superior to standard care in reducing the risk of death or massive transfusion within 24 hours.
Smith JE, et al. Prehospital Whole Blood in Traumatic Hemorrhage – a Randomized Controlled Trial. N Engl J Med. 2026 Mar 17.

What kind of study is this?

This is a pragmatic, multicentre, phase 3 randomised controlled trial conducted across 10 participating UK air ambulance services.

Patients were randomised (via pre-packed blood boxes) to receive:

Whole blood (up to 2 units)
Or standard care (red cells + plasma)
2 units of whole blood is approximately 940mL. They are larger than units of components, which are usually around 250-300mL each. 2 units of whole blood was therefore chosen as being roughly equivalent (in volume) to the standard care alternative of 2 units of red cells and 2 units of thawed plasma.

The trial was:

Unblinded (inevitable in this context)
Pragmatic (clinician-led decision to transfuse)

Tell me about the patients

Adult and paediatric trauma patients attended by participating UK air ambulance teams with suspected major haemorrhage.

Key features:

Median ISS ~33 (severely injured cohort)
Majority blunt trauma (~70%)
Balanced baseline characteristics
616 patients were included in the primary analysis after exclusions.
Traumatic cardiac arrest patients were excluded from the primary analysis, removing the sickest subgroup.

What were the measured outcomes in this study?

The primary outcome was a composite of death from any cause, or massive transfusion (≥ 10 units of transfusion) within the first 24 hours.

Secondary outcomes included:

Death from any cause at 6 hours, 24 hours, 30 days and 90 days after randomization
Massive transfusion
Days free from organ failure (up to 30 days)
Days in critical care and hospital (up to 90 days)
Number of units of blood component or product and cell salvage received within 24 hours after randomization (including prehospital transfusions)
Use of additional hemostatic agents within 24 hours after randomization
Prothrombin time that was above the normal range
Acid-base disturbance

What are the main results?

In short: no difference.

Primary outcome:
- Whole blood: 48.7%
- Standard care: 47.7%
- Relative risk: 1.02 (p = 0.84)

Mortality at all time points, rates of massive transfusion and rates of organ failure were all similar between the whole blood and standard care groups. There were no meaningful differences in thrombotic events or major safety concerns. However, prothrombin time was elevated in 40.7% of the whole blood group versus 30.5% of the standard care group.

What about the limitations?

There are a couple of considerations to have in mind, but ultimately I think this is a highly applicable and generalisable study to UK pre-hospital care practice. The points I think are of interest are:

Was enough whole blood given to really see a difference?
Were the patients sick enough to really benefit?

On the first point: 2 units of whole blood was chosen as an equivalent volume to approximately 2 units of red cells and 2 units plasma. From RePHILL we saw that, in a very similar cohort of patients, 2 units of red cells and 2 units of plasma were non-superior to crystalloid. Having received whole blood, the intervention arm patients went on to receive a median of 4 units of red cells and 4 units of FFP – the same as the control arm. Whilst this is good practice for assessing a purely pre-hospital intervention in that the subsequent cohort matching was very good, it is reasonable then to question…are 2 units of anything ever going to be the difference? The conclusion from this is probably that whole blood is not a an effective pre-hospital only intervention, but SWiFT hasn’t tested the efficacy of whole blood being used throughout an entire trauma system or trauma patient’s journey.

On the second point: The patients had a median ISS of 33. Of note, the median shock index of 1.15 in both groups was high, and predictive of major haemorrhage. The lactates on arrival to hospital were 5, and the base excess -6.5. So there is anatomical, physiological and metabolic evidence of haemorrhage and shock in this cohort. 44.9% of the whole blood group and 42.4% of the standard care group went on to receive massive transfusion (as defined by ≥ 10 units in the first 24 hours) – this is also a lot. So I think the trial has identified the correct cohort to test here . However, the median numbers of units (including pre-hospital transfusion) given tell a different story – 8 and 7 in each group respectively. This either indicates very rapid surgical haemorrhage control (we know this is less likely based on previous UK data), or patients who weren’t having rapid ongoing exsanguinating haemorrhage (who are probably the group that stand to benefit the most from pre-hospital transfusion). The physiology points this way too, as truly exsanguinating patients requiring rapid surgical haemorrhage control often have a shock index approaching or exceeding 2.

Some may say that the composite outcome is a weakness of SWiFT. In trauma trials, mortality alone is often a blunt and relatively infrequent endpoint, especially in systems with good survival rates. Detecting a mortality difference therefore requires very large sample sizes. By adding massive transfusion, the investigators aimed to capture patients with severe haemorrhage physiology and increase the event rate (~48%), thereby improving statistical power. However death and massive transfusion are fundamentally different outcomes. Death = patient-centred, definitive. Massive transfusion = process-driven, subjective. They do not carry equal weight: Preventing transfusion ≠ preventing death and an intervention could reduce transfusion but not mortality (or vice versa). The validity of the SWiFT results is strengthened by there being no difference in the composite outcome, and also no difference in the mortality and massive transfusion components of the composite.

Is there anything else to look at?

I think there are some interesting elements around traumatic cardiac arrest (TCA), rapid-sequence induction and intubation (RSI), and coagulopathy worth investigating. TCA and RSI are both subgroups, and hence hypothesis generating only. Coagulopathy, though, was a secondary outcome, so let’s start there

Coagulopathy: There was an abnormal prothrombin time in 40.7% of the whole blood group vs 30.5% of the standard care group. This has an odds ratio of 1.54 with confidence intervals which do not cross 1, so there is some significance here, but the authors point out that there was no impact of this on the primary outcome. The optimist could argue that whole blood allowed more severely injured (and therefore coagulopathic) patients to arrive at hospital (survivorship bias); I think this is probably a bit of a stretch given how well the cohorts and primary outcomes are matched. I think the more likely cause of this is how good FFP is as a product. The plasma is frozen very early after donation, and thus clotting factors are preserved. Units of ‘thawed plasma’ which most of the patients in the standard care arm received, have a maximum of 5 days at fridge temperatures for clotting factors to degrade, whereas cold-stored whole blood (unlike ‘fresh whole blood’, which is a different product) can be administered after up to 21 days of refrigeration.

TCA: Table S15 in the supplementary data shows outcomes for the patients in TCA on arrival of the air ambulance services. These were (I think rightly) excluded from the primary analysis, and have been reported separately. Whole blood appears to come off slightly worse in this cohort: death 78.8% SC, 85.1% WB, massive transfusion 18.6% SC, 24.8% WB, composite outcome 89.9% SC, 94.1% WB. The odds ratio between these 2 groups was actually quite high, 1.75, in favour of standard care, but the confidence interval is very wide (0.26 – 11.67) so really no inference can be made of this (very high event rate in both groups, low numbers) . I do wonder though if there is another cohort of patients, the ‘super massive transfusion’ group, who are much more sick than the vast majority of the study cohort, but are much more resuscitable than the TCA cohort, where benefit may be found.

RSI: The patients who were anaesthetised met the primary outcome 49.5%/54.4% (SC/WB) of the time. Patients who weren’t anaesthetised met the primary outcome 46%/44% of the time (SC/WB). They didn’t compare these groups directly so I’ve done a little stats work myself. The event rate (death/massive transfusion) in the combined (WB & SC) anaesthesia group was 52.1%, and in the non anaesthetised group it was 45%. The odds ratio is 1.33 in favour of not being anaesthetised. I suggest this is due to the patients who had TBI / multi-system trauma rather than an isolated bleeding pathology, but this is impossible to glean from the data. In London though, we are increasingly choosing to intentionally not anaesthetise our bleeding patients (without other injuries/priorities/indications) as we believe (and observational data shows) the time and mortality associated with doing so is higher. A meta analysis was used to write a consensus paper in 2025 supporting this concept of addressing massive haemorrhage and circulation before definitive airway management; perhaps SWiFT is also pointing us towards the concept that “you can’t ventilate your way out of haemorrhagic shock” (Hutch, 2026).

Regarding TBI, the SC & WB groups were well matched for traumatic brain injury Abbreviated Injury Scale (AIS) >/= 3, so this is unlikely to be skewing the data. Lots of the patients did have a head AIS >/= 3 though, and this will be contributing to some of the deaths in both groups.

Should we change practice based on this study?

In short, no. Whether you’re using whole blood or components, SWiFT shows us that what we give between those two options isn’t the important factor. Perhaps what else we are doing around our transfusion, namely avoidance of anaesthesia, aggressive en-route management of bleeding, and minimising time to definitive haemorrhage control, are the targets we need to be aiming for and was referenced here by Qasim et al (coining the term “scoop and control”). At London HEMS we are teaching the damage control pre-hospital care strategy for this highly selected group of patients as well. Also I have some (perhaps a bit radical) ideas to rethink the entire sequencing of how we care for trauma patients where their main or only problem is haemorrhage.

Thoughts from Our Man in America – Zaf Qasim

This trial has gained a lot of interest this side of the pond as well – in the last decade, many EMS agencies have implemented a prehospital blood program, with a mix of using whole blood or component therapy. The differences in care delivery across systems are important to highlight. Although I have the privilege of being part of a physician-delivered prehospital care team, this is significantly less common than in the UK and Europe. Arrival of such a team may delay care when the more prudent course of action would be to start moving to definitive treatment and providing advanced intervention en-route.

For that reason, it was important to empower paramedics across the US to have blood administration included in their protocols. Many agency and state protocols (though not all) have now changed to allow this. I think this is an important variable that is not clearly outlined in the SWiFT trial – what was the time from 999 call to air ambulance team arrival and by extension to initiation of blood products? I would imagine there is inevitably a delay of some time between the first-on-scene ground ambulance team and the HEMS team due to logistics and geography. Does this time lost translate to any change in outcomes? This study does not answer that completely.

Prehospital and transport times can impact outcomes. A median of 38 minutes were spent on scene once the air ambulance arrived. Given it then took a median of 28 minutes to transport to the hospital, it is important to understand how that time on-scene is spent. At what point was the transfusion started and importantly completed? Are there marginal gains to be achieved to shorten scene time (such as first-arriving ground units establishing intravenous access) and to continue advanced critical care interventions en-route, thus shortening overall time to definitive intervention? Longer on-scene times should not necessarily be seen as a negative, but that time must be used appropriately. The New Orleans group showed that although they spent an average of 4.5 minutes longer on-scene with their blood-carrying crews, the time to initiation and completion of the first unit of blood was >15 minutes faster than if the patient had just been transported to hospital without this intervention. Initiating and completing the transfusion early in a patient in true hemorrhagic shock addresses that shock state earlier, before it progresses to an irreversible stage, and this may lead to an improvement in outcomes. Given the disparities of prehospital care delivery worldwide, these systems questions are important systems and should be taken into account when we evaluate trials like SWiFT.

When this paper came out, I found myself on the phone to Dr. John Holcomb to discuss the relevance to US prehospital care. Dr. Holcomb, author of the PROPPR trial amongst many others, is also on the board of the Prehospital Blood Transfusion Coalition. This is a multidisciplinary collaboration that promotes best practice in prehospital blood transfusion programs in the US. He highlighted several things – including that battlefield experience has clearly shown blood product resuscitation to be superior to crystalloid use, and that military observational data supports the survival benefit of prehospital blood components. “Hence, this trial is probably a good thing – in particular for those who are having challenges in obtaining whole blood for their programs. Any blood product is better than crystalloid!”

This paper likely eases some of the logistical issues of trying to obtain whole blood from blood suppliers within a fledgling program. “Don’t wait for low-titre O whole blood to become available in your area, coordinate with your hospitals and blood supplies.” Holcomb adds. “Make informed decisions about what blood products to use based upon availability, logistics, costs, and operational needs.”

Holcomb further highlighted the importance that some components, in particular plasma, may actually be most beneficial in prehospital trauma, especially TBI patients. After all, the PAMPer trial was the only positive blood component trial that showed benefit in prehospital trauma patients. Liquid plasma is a rapidly transfusable form that does not need to be thawed and is currently the least expensive component in the US, although it suffers from a short shelf life. The US Food and Drug Administration has yet to approve the use of lyophilized plasma in civilian practice – something that our UK and European colleagues have had access to for years. Should this become available in the US, EMS agencies seeing lower volumes of major trauma patients will at least have a product that can be stored on the shelf for a significant period of time and be beneficial to our patients given the current evidence.

For me, in my system, where we have prehospital whole blood available, I think we will continue to use that. The logistical benefit of carrying a smaller volume of product that can be administered faster with at least an equal result to the equivalent of components (and likely one that exposes the patient to fewer donors than if components are used) still makes sense for us. SWiFT certainly supports that for systems that can achieve it. Likewise, in my mind, it also takes away some of the potential moral injury clinicians who only have access to component therapy may face thinking that this may be an inferior treatment choice for their patients.

Summary

Whole blood in trauma makes intuitive sense — but SWiFT reminds us that intuition and evidence don’t always align. In this pragmatic UK prehospital RCT, giving 2 units of whole blood was not superior to standard component therapy for death or massive transfusion, with outcomes essentially identical across all major endpoints.

There are some interesting signals. Coagulopathy was higher in the whole blood group — likely reflecting the relative preservation of clotting factors in FFP and the problems with using older units of whole blood.

The key message is that: Which transfusion product we give pre-hospital matters less than what we do around it.

Minimising scene time, avoiding unnecessary anaesthesia, and maintaining the trajectory to definitive haemorrhage control (while providing advanced en-route care) are likely to have a far greater impact than choosing whole blood over components.

Best wishes, Hutch and Zaf