Paris study models wider defibrillator access using drones after cardiac arrest

A Greater Paris study presented on 26 September found that placing automated external defibrillators (AEDs) at drone bases could expand access after an out-of-hospital cardiac arrest. The researchers estimated that a network of 100 bases and 26 additional fixed AEDs would put more than 97% of historical cardiac arrest locations within 500 metres of a fixed device or about four kilometres of a base. The result describes potential coverage, not deliveries or lives saved.

The study, presented at the European Emergency Medicine Congress in Paris, examined 28,349 cardiac arrests recorded from 2011 to 2024 in administrative areas of Greater Paris. Central Paris was excluded. Researchers compared the locations with recorded fixed AEDs and modeled where additional devices and drone bases might go. The Guardian reported that the work had not yet been peer reviewed.

What the Paris model found

Of the recorded arrests, only 30% were within a 500-metre ground-network distance of one of 1,893 recorded fixed AEDs. The team estimated that adding 910 fixed devices would raise coverage on that measure to 84.5%. It estimated that 1,712 more fixed devices would be needed to put every recorded case within that distance. Those figures measure proximity to a recorded device, rather than whether a bystander could reach it and use it in time.

A modeled network with 200 drone bases and four additional fixed AEDs would put more than 99% of historical cases within the study's coverage area. The European Society for Emergency Medicine reported 99.4% overall coverage in a model using 200 bases and 871 existing fixed AED sites. These scenarios combine different ways of reaching a device; being within range of a drone base does not itself establish how quickly a person at the scene would receive or use an AED.

Study co-author Matthieu Heidet told the Guardian that only 8% of out-of-hospital cardiac arrest patients in France receive a public AED before emergency services arrive. He cited devices housed in private spaces that are not always accessible as one barrier. That gap helps explain why a map of AED locations cannot, by itself, show how often a device reaches a patient.

What has happened in Sweden

A separate prospective observational study in Sweden provides a check on what drone delivery can accomplish outside a model. Five AED-equipped drones served areas with about 200,000 residents between April 2021 and May 2022. Of 211 suspected cardiac arrest alerts, a drone was dispatched in 72, or 34%. An AED was delivered in 58 of those 72 deployments. The study identified cancellation after dispatchers found that an incident was not an out-of-hospital cardiac arrest as the main reason for non-delivery.

For 55 Swedish cases with both drone and ambulance arrival times available, the drone delivered an AED before the ambulance arrived in 37 cases. Its median lead in those cases was three minutes and 14 seconds. Earlier arrival is a meaningful operational measure, but it is separate from whether someone at the scene attached the device, delivered a shock or survived.

Among those Swedish cases, 18 were confirmed cardiac arrests. A drone-delivered AED was attached in six; two patients with a shockable first rhythm were defibrillated before an ambulance arrived. One of those patients survived to 30 days. Those small numbers document use of the service and an individual outcome. They cannot establish a population-wide survival benefit from drone delivery.

The Swedish study reported no adverse events, but drones could not be dispatched in some circumstances, including unfavorable weather, darkness, no-delivery zones and lack of air traffic control approval. Those limits matter when comparing the service with a model that assumes bases can cover a mapped area. A drone's theoretical range is only one part of access during an actual emergency.

Earlier Swedish research tested 18 simulated flights to historical cardiac arrest locations. In that 2017 study, median drone arrival was five minutes and 21 seconds, compared with 22 minutes for emergency medical services in the historical comparisons. The authors cautioned that the flights were few, short and conducted in good weather, while older emergency-service times might not reflect current traffic. Those results offered an early feasibility measure, rather than evidence of better patient outcomes.

The questions before deployment

The Paris researchers modeled bases at locations including existing AED sites, fire stations and mobile intensive care units. They did not test a working network there. Heidet told the Guardian that drone operating costs must be part of any assessment and that the network offering the widest modeled coverage may differ from one that is economically feasible. Weather, flight permissions, delivery at the scene and whether a bystander can use the device are further practical questions that the Paris coverage figures do not settle.

The findings therefore support a narrower conclusion: drones may extend the area served by AEDs and, under modeled conditions, may shorten delivery times. The observed Swedish service shows that some deliveries did beat ambulances, while also showing how often dispatch and use of a delivered device are separate steps. Whether a Paris network would deliver those gains in practice, or improve survival, remains unknown.