An ambulance crew is, among other things, a rolling site of live clinical research. Three separate stories from the past few months show how much that setting now contributes to life sciences beyond simply getting someone to hospital. London’s cardiac arrest survival figures have just reached their highest point in six years, credited largely to what happens in the minutes before paramedics arrive. A major international trial run out of UK air ambulances has just delivered a firmly negative result on whether whole blood transfusion in the field saves more lives than standard care. And ambulance services across the country are quietly testing whether artificial intelligence can take notes so clinicians do not have to. None of this happens in a laboratory. It happens on the way to one.
The Difference Bystanders Make
London Ambulance Service reported in January 2026 that cardiac arrest survival in the capital had reached 10.9 per cent, roughly one in nine people, matching pre-pandemic levels and the best result in six years. More than 2,500 cardiac arrest patients received bystander CPR before paramedics arrived, but the figures also expose a stark postcode lottery. Islington recorded bystander CPR in 74.4 per cent of cases, the highest in London, while Bromley and Barking and Dagenham sat at 54 per cent, a 20 percentage point gap between the best and worst performing boroughs.
Consultant Paramedic Mark Faulkner said early chest compressions and defibrillator use from bystanders “more than doubles a person’s chance of surviving,” and that these are skills that take only minutes to learn. That claim is backed by a sustained public health effort rather than a single campaign: nearly 47,000 Londoners have been trained through the London Lifesavers programme since September 2021, and around 11,000 public defibrillators are now sited across the city. The improvement in survival is credited to a combination of that bystander training, faster ambulance response times and quicker 999 call answering, which is a useful reminder that some of the most significant gains in emergency medicine now happen before a clinician is anywhere near the patient.
A Trial That Said No
Not every piece of ambulance research produces a positive result, and the SWiFT trial is a good example of why that matters. The trial, sponsored by NHS Blood and Transplant and run with Queen Mary University of London, Barts Health NHS Trust, participating air ambulance services and Defence Medical Services, tested whether giving trauma patients whole blood in the field, rather than red blood cells and plasma administered separately as is standard practice, improved outcomes. Over two years from December 2022, 942 patients were recruited, with 616 eligible for the main analysis, 314 receiving whole blood and 302 receiving standard component therapy.
The result, published in the New England Journal of Medicine in March 2026, was a clear no. Death or the need for a massive transfusion within 24 hours occurred in 48.7 per cent of the whole blood group compared with 47.7 per cent of the standard care group, a difference the researchers described as not statistically significant, with similar mortality and safety outcomes across both arms throughout. A negative trial of this size is not a failure. Whole blood is more complex and expensive to store and deploy in an ambulance than separate components, and this result gives services solid evidence that switching to it would not be worth that cost, rather than leaving the question to guesswork or enthusiasm for a plausible-sounding idea.
Listening In on the 999 Call
The third strand is less about treatment and more about time. South East Coast Ambulance Service is piloting an AI tool, built by the company Tortus, that listens to calls handled by clinicians in its Emergency Operations Centre and automatically turns the conversation into a structured medical record, a task that would otherwise mean a clinician manually typing notes during or after the call. Clinicians still review and approve every note before it is used. The pilot, running through February 2026, is being tested alongside similar trials at London Ambulance Service and South Western Ambulance Service, and sits within the NHS’s wider 10-year plan commitment to bring AI into clinical operations under close supervision rather than at arm’s length.
The stated aim is straightforward: give clinicians back the minutes currently spent writing up a call so they can spend them on the next patient instead, while producing a more consistent clinical record than handwritten notes typically allow. It is a small, unglamorous piece of technology compared with a landmark drug trial, but in a service where minutes measurably change survival odds, as the cardiac arrest figures show, time saved on documentation is not a trivial gain.
The Common Thread
A bystander with two minutes of training, a randomised trial testing an expensive new protocol against the one already in use and software quietly reclaiming a clinician’s time are not obviously related. What connects them is that an ambulance is one of the few places in medicine where the gap between an intervention and its outcome is measured in minutes rather than months, which makes it an unusually honest testing ground. Ideas that sound sensible, like giving trauma patients whole blood, still have to prove themselves against a stopwatch and a control group. Skills that sound too simple to matter, like chest compressions from a stranger, turn out to double survival odds. Ambulances were built to move people quickly. What is becoming clearer is how much they also move evidence.
Sources: London Ambulance Service NHS Trust; New England Journal of Medicine; Queen Mary University of London; South East Coast Ambulance Service NHS Foundation Trust.
