I regularly listen to The Resus Room. In the episode "Papers of August 2026" Simon Laing and Rob Fenwick discuss three papers, and the first intrigued me enough to read the original text. It's a study by London's Air Ambulance published in the Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine, and it addresses a question that has circulated in the prehospital debate for thirty years without ever having a clean numerical answer: how much time, in minutes, does each advanced procedure we decide to perform on scene cost.
The paper doesn't answer the question everyone expects. It doesn't say whether it's better to intervene or load and go, because it doesn't look at mortality or functional outcome. It does something more circumscribed and, in my opinion, more useful: it puts a price on the clinical procedure. And once that price exists, the discussion about stay and play and scoop and run changes nature, because it stops being a question of philosophy and becomes a question of arithmetic applied to one's own system.
What they did
Gandolfi and colleagues compared two time windows separated by a precise discontinuity. Between 2010 and 2017 London's Air Ambulance introduced into current practice four procedures it didn't have before: prehospital blood transfusion, REBOA, arterial cannulation and central venous access. The authors therefore analyzed patients treated in the month of July in the years 2005-2010 (group 1, 728 patients) and 2017-2021 (group 2, 629 patients), deliberately leaving empty the interval in which the change occurred.
Eight procedures were classified as advanced: prehospital anesthesia, blood transfusion, thoracostomy, thoracotomy, REBOA, central venous access, arterial cannulation and treatment of traumatic cardiac arrest. The multivariate model then isolated the contribution of each variable to scene time, defined as the time between the team's arrival at the patient and departure from the scene.
The number
Each additional advanced intervention is associated with a 41% increase in scene time (relative change 1.41; 95% CI 1.35-1.48; p < 0.001), after adjustment for period, age, mechanism, team size, mode of arrival, time and day of the week.
The detail that matters is that the effect is multiplicative and not additive. We're not adding five minutes per procedure. We're compounding percentages on a growing base. The translation into absolute time is brutal: 17 minutes median on scene for patients who receive no advanced intervention, 57 minutes median for those who receive five. REBOA, with only three cases and therefore with all due caution, is associated with the highest median of all: 61 minutes.
Prehospital anesthesia, by far the most frequent procedure (309 patients, 22.8% of the cohort), is associated with a median of 37 minutes, identical in both periods. Twenty years of accumulated experience, checklists, standardization and simulation have not shortened the time of a PHEA. This is a datum worth looking at for a long time.
The second independent predictor is the mechanism. All else being equal, blunt trauma has a scene time almost double that of penetrating trauma (1.91; 95% CI 1.78-2.05). In crude terms: 25 minutes median versus 10.
The three quarters who receive nothing
There's a datum that almost never appears in the debate. 74% of patients treated by London's Air Ambulance receive no advanced intervention. In the more recent group the proportion drops to 70.6%, but remains largely in the majority.
This datum must be read carefully, because it lends itself to a wrong interpretation. It would be natural to conclude that these are simply less severe patients, and in part this is true: those who work in dispatch centers know perfectly well that the proportion of truly critical missions is a fraction of the total, and that the added value of a system is played out at least as much on patients who won't end up in the operating room. But here the population is not the general one. It's a cohort filtered twice: by a dedicated dispatcher who listens to calls in real time to intercept major trauma, and by the exclusion of 154 patients declared dead on scene. The median Injury Severity Score documented in previous studies by the same service is 17.
In other words: even in a preselected population of urban major trauma, served by the team with the widest range of capabilities in Europe, three out of four patients are treated and transported without any advanced procedure. The caricature of stay and play, of the physician who persists on scene because they have the tools to do so, doesn't even exist there. Which shifts the question: not "how much do we intervene", but "on whom".
The missing denominator
Here comes the point that from my context is impossible not to see, and that a London study couldn't address.
A 41% increase in scene time is a cost. But a cost only has meaning in relation to what you give up obtaining. And what you give up, in trauma, is the time to arrival at definitive care. Which is not a constant: it's a geographic variable.
In London the median transport time is 15 minutes in the recent period. One more intervention on scene means delaying by a few minutes access to an operating room that is anyway a quarter of an hour away. The operating room remains the real alternative, immediately available, and the logic of minimizing time on scene is consistent with the structure of the system.
In Valle d'Aosta, as in any mountain or rural system, that denominator changes category. Transport to the only regional hospital can exceed an hour. In the side valleys, in winter, with the helicopter unavailable, much more. The same 41% is measured against a wait that is no longer a quarter of an hour but a time in which the patient may simply not arrive.
The reasoning is inverted. Where definitive care is fifteen minutes away, advanced intervention on scene is a delay to be justified. Where it's seventy, the same intervention is often the only therapeutic window available, and the question becomes its exact opposite: what capabilities must we bring on scene, given that the hospital is not a short-term alternative.
It's the same datum, with two opposite clinical meanings. And it's the reason why mechanically transferring the operational recommendations of large urban services to territorial systems is an error we continue to make. Not because those recommendations are wrong, but because they were calibrated on a denominator that we don't have.
It's worth noting that the study itself signals this limitation, explicitly defining reduced external validity toward non-urban services.
The minutes were added elsewhere
If the paper stopped here it would already be useful. But it contains a second result that receives much less attention in the discussion, and that seems to me the most interesting from a system point of view.
Scene time, in twenty years, has not changed: 22 minutes median in the first period, 20 in the second, non-significant difference. The calendar period was not an independent predictor. It seems like reassuring news.
Total prehospital time, however, has increased: from 59 to 63 minutes, with statistical significance. And the increase doesn't come from the scene. It comes from two different phases:
- activation-arrival on scene: from 21 to 24 minutes
- transport: from 10 to 15 minutes
The second datum has a precise structural explanation. In 2010 London established its Major Trauma System, formalizing the bypass of local hospitals toward designated trauma centers. A correct system choice, with a cost in minutes that was consciously accepted. The first, the increase in activation time, the authors attribute hypothetically to the change in case mix: penetrating trauma requires telephone interrogation before dispatch, while some blunt mechanisms automatically activate the team.
The overall result is that we spent twenty years timing, discussing and optimizing the only phase that clinicians directly control, the one on scene, achieving to keep it stable in the face of an increasing procedural load. Excellent result. Meanwhile the minutes accumulated in the two phases that depend on system design and dispatch logic, where no one was watching with the same intensity.
From the dispatch center point of view this is the most actionable message of the paper. The lever on prehospital time is not only, and perhaps not mainly, in the hands of those on scene. It's in the quality of call interception, in the activation threshold, in the choice of vehicle and in the destination. These are decisions made in a few seconds by those who will never see the patient, and that shift more minutes than one less procedure shifts.
Methodological note: the mean that hides two trajectories
One last element remains, which is the real reason why it's worth reading the paper in full and not just the abstract.
Why has scene time remained stable, if each additional intervention increases it by 41% and interventions have become more frequent? Because in the meantime the case mix has changed. The proportion of penetrating trauma went from 24.2% to 34.2%, and penetrating trauma has structurally short scene times. Two opposite trajectories, which in the aggregate cancel each other out.
The mean datum, therefore, doesn't describe any real patient. It describes the overlap of two populations that are moving in opposite directions. If a service looked only at its own median scene time as a quality indicator, it would conclude that nothing has changed, while in reality almost everything has changed: in the cohort with blunt trauma, the proportion of patients who receive more than one advanced intervention has more than doubled, from 4.9% to 11.6%, while in penetrating trauma it remained unchanged.
This is why aggregate indicators, taken alone, are weak instruments of clinical governance. Not because they're false, but because they're compatible with too many different stories.
The limitations that the authors honestly declare must be added. The study is retrospective. It samples only one month per year. It doesn't report outcomes. Procedures are often performed in parallel and not in sequence, which makes it impossible to attribute to each a proper time cost. And above all there exists confounding by indication that the authors explicitly recognize: patients who require more procedures are also the most severe and complex, and their severity lengthens scene time independently of the procedures themselves. The 41% is therefore not the pure cost of the technical procedure, but the cost of the procedure together with the clinical context that made it necessary.
Finally there's an exclusion that pushes the estimate downward: the 154 patients declared dead on scene were removed from the analysis. They are, in all likelihood, among those who received more interventions and absorbed more time. Their exclusion is methodologically defensible, given that they were not transported toward definitive care, but the reader must know that the real cost in minutes of advanced interventions is likely higher than that estimated.
What I take on shift
The paper doesn't tell me whether to work on scene or leave quickly toward the reference Hub. It tells me three different and more solid things.
- That each procedure has a measurable price, and that price is not linear: the second costs more than the first, the third more than the second.
- That price must always be divided by the time that separates me from definitive care, and that this ratio is the true clinical variable, not the procedure itself. It's also the reason why the same evidence generates legitimately different protocols in London and in an alpine valley.
- That while we discuss minutes on scene, the system adds them elsewhere without anyone commenting on it.
The question that remains for me is systemic, and I gladly turn it to those who work in contexts with long transport times: have we ever measured our denominator? Not the median transport time on paper, but the real one, by valley, by season, by helicopter availability. Because without that number every discussion about what to do on scene remains a theoretical discussion borrowed from someone else.
Reference Gandolfi M, Bird F, Henry CL, Bestwick J, Lockey DJ, Perkins ZB. The impact of advanced pre-hospital interventions on scene time. Scand J Trauma Resusc Emerg Med. 2026;34:108. doi: 10.1186/s13049-026-01613-5
Source of inspiration The Resus Room, Papers of August 2026
