The data that informs air ambulances and improves patient care

Spending a morning on top of the Royal London Hospital, home of London’s Air Ambulance helipad requires a head for heights that may be too much for some. Inspired by my visit, I speak to the data analyst who has been working out why sometimes it’s quicker to send a car rather than a helicopter when every second counts.

Christine Henry has a long career in data analysis and digital projects in and around the healthcare world. She is a problem solver and since joining London’s Air Ambulance (LAA) in March 2020, she has been using data to find solutions to help them to improve patient care.

LAA is a charity, it raises funds to meet the £10m annual running costs for an organisation that employs around 100 staff. It works with doctors and consultants from Bart’s Hospital for its clinical needs and with the London Ambulance Service who provide paramedics on secondment and dispatch facilities from its Emergency Operations Centre. It is currently focused on raising £15 million to replace its helicopters by 2024 through its Up Against Time Appeal.

The problem that Christine was trying to solve was simple: is it quicker to send a car or a helicopter to incidents within a predefined footprint of the helipad at the Royal London Hospital?

Recognising the constraints

To put this in context, it’s worth describing the geography of this part of east London. It’s densely populated urban area, a mixture of old and new housing stock with Stratford accessed by a major road further east and beyond the green extent of Victoria Park.  Like many areas of London, traffic is slow moving and access can be difficult even under blue lights. There aren’t that many places to land a helicopter either.

Faced with these constraints, the LAA dispatching paramedic needs to decide whether to dispatch an LAA car or the helicopter. Christine’s job was to work out what the last four years’ worth of data was telling her about the decisions made about what resource to dispatch and whether any improvements could be made for future calls. LAA already had a footprint map of the local area to show where a car could be dispatched but didn’t have any evidence to show that it was more efficient than sending a helicopter.  

Christine needed data from not only from LAA systems but from those held by Bart’s Hospital and London Ambulance Service meaning that data sharing agreements had to be in place. Data sharing is a challenge for any project particularly involving sensitive personal data, although in this case all the personal data could be stripped out as it wasn’t needed for this project.  

Geospatial view

Using open-source programming tools, Christine used data from the three systems and created a footprint map where the local area is cut up into 500m square sections. On average, LAA attends 5 or 6 jobs a day. This amounts to 1,500 calls a year but only ten per cent of these fall in the local footprint of the Royal London Hospital.

There are limitations on when the helicopter can be used as it only flies in daylight hours, during the winter months, it won’t be deployed after 1600. At night, it is flown to Northolt, a short hop along the A40 west out of London. There are also weather-based restrictions, so heavy fog and rain can also limit its use.

Checking against perceptions

With four years of data, Christine was able to map out where all the calls were located, how long it took to reach each patient and what response was used. She then wanted to make sense of this with clinical colleagues to see if it matched their perceptions by setting up an expert working group.

It’s interesting to hear about the conversations that Christine had with her colleagues about whether what the data was telling them matched their own understanding based on responding to incidents in the local area. The data was simply reflecting in a geospatial way the historical record of previous incidents, but it didn’t always match with what they remembered about these calls.

Using the footprint map

Christine will review the data based on the new footprint map that has been operational for about a year now. She wants to see if the changes have been beneficial. I was interested in whether the change to the footprint had made dispatching decisions harder, but Christine says no. “The new footprint is automatically updated on the computer map. For the dispatchers, it should make zero difference.”

Explaining how the change affects patients, she adds, “Now that I’ve got a full year of data, I can run the numbers and look at how many patients were affected by this. What do the times look like and are they vastly different from historical data. Even a couple of minutes  saved in getting to a patient can make a difference.” Determining success is not simple as Christine points out, it’s hard to know the counter factual. “What we’ll have to look at is does it match our prediction based on the historical experience.”

Forging ahead with data

In 2019, before Christine joined LAA, the organisation commissioned external consultants to develop a digital transformation strategic plan. This covered improving operational processes throughout the organisation, removing paper processes and replacing them with digital improvements. She says the impetus came from a data champion at the strategic clinical level of LAA and they are now looking at data maturity to see how far they have come in the last three years. She’s optimistic about how LAA values its data and now that she’s got this project under her belt, she’s in a great position to do even more with data to improve patient care.

---

This article originally appeared in the February 2023 print edition of Emergency Services Times with the headline ‘Footprints and flight paths.’