Most studies on trauma outcomes start at the hospital door. This one doesn't.
A study just published in the British Journal of Surgery by Perkins, Ter Avest and colleagues from London's Air Ambulance analyzed 497 trauma deaths over two years, within one of the world's most developed trauma systems. What they found should redefine how we think about where to look for the next improvements in trauma survival.
The answer isn't in the operating room. It's in the field.
The Numbers That Matter
Between 2019 and 2020, London's Air Ambulance attended 3,089 patients with major trauma. Of these, 497 died: a mortality rate of 16.1% in a physician-led HEMS system, operational 24/7, integrated into a mature regional network with four Major Trauma Centres and 22 Trauma Units.
The figure every prehospital clinician should keep in mind: 77.1% of those deaths occurred before hospital arrival. The median time from traumatic event to traumatic cardiac arrest was 12 minutes. Not 30. Not an hour. Twelve minutes.
The Curve Has Shifted. The Problem Hasn't.
Baker and Trunkey described the trimodal distribution of trauma deaths in 1980: immediate deaths at the scene, early deaths within hours from hemorrhage or brain injury, and late deaths from multiorgan failure. That model shaped an entire generation of trauma system design. Damage control surgery, massive transfusion protocols, specialized trauma centers: all of this was built to address the second and third peaks.
And it worked. The late peak has essentially disappeared. Early in-hospital deaths occur sooner and less frequently. What remains is a highly left-shifted, unimodal curve, what Holcomb defined as the "consistent curve of death." Deaths concentrate early and concentrate outside the hospital.
The London data confirm this pattern with a granularity that most studies lack. Nearly 15% of deaths were instantaneous. An additional 37% occurred within 15 minutes of trauma. Adding the next time bracket, over half of all deaths have already occurred. The hospital never had a chance to intervene.
Hemorrhage Dies in the Field. Head Trauma Dies in Hospital.
The most relevant finding of this study isn't the overall distribution. It's the cause-specific separation between prehospital and in-hospital deaths.
Hemorrhage caused 36.8% of all overall deaths. But 96.2% of hemorrhage deaths occurred before hospital arrival. Only 7 out of 183 patients who died from exsanguination reached the hospital. The system that was designed to save these patients—the trauma center with its blood bank and surgeons—almost never sees them anymore.
Conversely, 84.2% of in-hospital deaths were due to head trauma. The hospital has become very effective at stopping fatal hemorrhage. It has much less to offer when the primary damage is neurological.
Every single death from airway obstruction, traumatic asphyxia, tension pneumothorax, and cardiac tamponade occurred in the prehospital phase. These are reversible causes. All of them. And all killed before the patient reached a trauma center.
Penetrating Trauma: A Separate Epidemic
The penetrating trauma data deserve separate analysis. Of 144 penetrating trauma deaths, 95.1% occurred in the prehospital phase. These patients were younger (median 26 years versus 40 in blunt trauma), more often male, and died overwhelmingly from hemorrhage (84%) or cardiac tamponade (7.6%).
This is a population where the intervention window is measured in single-digit minutes. The median time from event to cardiac arrest isn't published separately for penetrating trauma in this study, but the overall median of 12 minutes almost certainly overestimates the time available for knife and gunshot victims.
The implications are clear: for this population, survival depends on what happens before the ambulance arrives. Bystander hemorrhage control, public access to bleeding control devices, and first responder training aren't accessory programs. They are the intervention.
The Survival Bias Nobody Talks About
This article puts in black and white something that is often perceived but rarely stated in trauma literature: most outcome studies exclude the dead.
If you only analyze patients who survive to admission, you're looking at the healthiest 23% of the mortality pool. You build your understanding of trauma, design protocols, and allocate resources based on a population that has already been filtered by a massive survival bias.
The practical consequence is that hospital registries systematically overestimate the time available for effective hemorrhage control. When a study reports that patients with hemorrhage die at a certain point after admission, it's describing a population that had sufficient physiological reserve to survive transport. Those who exsanguinated in 8 minutes are simply absent from the dataset.
Perkins and colleagues state this explicitly: the causal profile of trauma death appears fundamentally different when the prehospital phase is included. Ignoring it means optimizing for the wrong problem.
What This Means for System Design
This study doesn't argue against hospital trauma care. It argues that hospital care has done its job. The late mortality peak has disappeared. In-hospital hemorrhage deaths have halved in the last decade in the London system. The remaining margins for improvement, the next measurable reduction in trauma mortality, now depend on what happens before arrival.
The authors point to three areas. The first is primary prevention, particularly of violence, which fuels penetrating trauma deaths. The second is hemorrhage control by bystanders and first responders: programs like Stop the Bleed and CitizenAID. The third is advanced prehospital capability, including early blood product administration and treatment of obstructive shock.
One sentence from the paper deserves a pause: "This does not imply prolonged on-scene care, but rather the timely delivery of targeted, life-saving interventions alongside rapid transport to definitive hospital care." This is the balance every prehospital system must find. Not "stay and play" versus "scoop and run." Targeted intervention within the 12-minute window, and then move.
The Thread Connecting the Evidence
This paper doesn't exist in isolation. It fits into a growing body of work pointing in the same direction.
Nutbeam and Stassen recently proposed extending the Chain of Survival model, traditionally used for cardiac arrest, to road trauma, mapping five phases from automatic crash notification through rehabilitation. Their argument is structural: improving survival requires optimizing every link, not just the hospital one.
The military literature tells the same story from a different angle. Conflicts in Afghanistan, Iraq, and Ukraine have demonstrated that when you bring effective hemorrhage control closer to the point of injury, you save lives. Tourniquet conversion protocols, whole blood resuscitation, and the "golden hour" policy all arose from recognizing that the critical window isn't an abstract concept. It's a measurable and compressible interval.
And evidence on prehospital transfusion continues to accumulate. A narrative review from 2026 by Burns and colleagues in SJTREM documents the evolution from military experience to civilian implementation, with the shift toward low-titer O whole blood simplifying the logistics that historically limited field blood product availability.
All these lines of evidence converge on the same conclusion. The 12-minute window is real. It's measurable. And the interventions that can make a difference within it already exist. The challenge is deploying them at scale, systematically, at the point of injury.
What I Take Home from This Study as a Field Physician
I work in a system where transport times are long, there's only one hospital, and the terrain doesn't forgive delays. The Aosta Valley isn't London. We don't have four Major Trauma Centres and 22 Trauma Units within reach. But the physiology is the same, and the clock runs at the same speed.
Twelve minutes. That's the median time from event to cardiac arrest in this cohort. In a mountain environment, 12 minutes is barely enough to reach the patient. Which means that in systems like mine, the interventions that matter most are those that can be delivered by whoever is there first: bystander hemorrhage control, tourniquet application, effective dispatch, and the fastest possible deployment of advanced capabilities.
Perkins' study confirms what many of us observe clinically: we've built excellent hospitals. Now we need to build equally excellent systems for the minutes before the hospital.
References
- Baker CC, Oppenheimer L, Stephens B, Lewis FR, Trunkey DD. Epidemiology of trauma deaths. Am J Surg. 1980;140(1):144-150. doi:10.1016/0002-9610(80)90431-6
- Holcomb JB. Transport time and preoperating room hemostatic interventions are important: improving outcomes after severe truncal injury. Crit Care Med. 2018;46(3):447-453. doi:10.1097/CCM.0000000000002915
- Cole E, Weaver A, Gall L, et al. A decade of damage control resuscitation: new transfusion practice, new survivors, new directions. Ann Surg. 2021;273(6):1215-1220. doi:10.1097/SLA.0000000000003657
- Nutbeam T, Stassen W. The road injury chain of survival: a framework for improving trauma outcomes. Injury. 2025. doi:10.1016/j.injury.2025.112285
- Perkins ZB, Greenhalgh R, ter Avest E, et al. Prehospital resuscitative thoracotomy for traumatic cardiac arrest. JAMA Surg. 2025;160(4):432-440. doi:10.1001/jamasurg.2024.7245
- Ter Avest E, Kocierz L, Alvarez C, et al. Improving decision-making for prehospital resuscitative thoracotomy in traumatic cardiac arrest: a data-driven approach. Crit Care. 2025;29(1):485. doi:10.1186/s13054-025-05705-z
- Burns B, Nolan B, Ferguson I, Partyka C, Healy G. Prehospital transfusion strategies in haemorrhagic shock. Scand J Trauma Resusc Emerg Med. 2026. doi:10.1186/s13049-026-01547-y
- Nijhawan A, ter Avest E, Twohig CJ, et al. Epidemiology of trauma deaths across a mature regional trauma system: patterns of pre-hospital and in-hospital fatalities. BJS. 2026. doi:10.1093/bjs/znag030
