Data doctors: saving money and the planet

Specialist anaesthesiologist Dr Richard French turns to milk to help him explain a critical and costly part of the anaesthesia process, when a patient has been put under by intravenous drugs and are being kept sedated by anaesthetic inhalants. The anaesthetist is closely monitoring both the patient’s vital signs and the flow of the anaesthetic vapours.

Describing the gas flow, French, who is clinical director of anaesthesia for Canterbury District Health Board in Christchurch, New Zealand, says: “If your goal is to fill a glass with milk right to the top, you can either do it by very, very slowly filling the glass—which is quite painstaking—or you can just keep pouring the milk until it’s flooding over the top and going across the counter. Either way, the glass doesn’t get any more full.”

Dr Richard French (right), clinical director of anaesthesia for Canterbury District Health Board in Christchurch, New Zealand and Professor Ross Kennedy worked together with GE to pilot a study using data to improve the use of anaesthetic machines.

Expensive anaesthetic inhalants are commonly in the top 10 of a hospital’s drug spend and contain volatile chlorofluorocarbons, potent greenhouse gases. During an operation, needlessly flowing a high volume of the vapours doesn’t benefit or harm patients, but it’s bad for hospital budgets and for the planet. So it makes sense that operating theatres want, as French explains with his analogy, “to minimise flow rates and avoid spilling the milk”.

That anti-wastage drive led Christchurch Hospital, already proud of their standing as a “low-flow shop”, to partner with GE in a pilot study, where data from the hospital’s anaesthetic machines was analysed in near-real time by GE data scientists in the US.

French says it was the years of work by his renowned colleague Professor Ross Kennedy, the pilot’s scientific lead, that made Christchurch Hospital the perfect partner for GE. “Over 15 years, Ross has been collecting data from anaesthetic machines, which is a very laborious process,” says French. Kennedy has published several academic papers on his research.

GE Healthcare’s Stephen Attkins says: “Christchurch Hospital had the right clinical engagement, the right IT infrastructure, they were already collecting data and using GE technology—that enabled us to do some truly groundbreaking work.”

French also praises his hospital’s administrators: “Clearly this sort of project can only thrive with really supportive senior management and an environment that promotes positive change; they smoothed the way so that Ross and I could focus on the clinical side of things.”

The result of the global-first pilot is GE’s Carestation Insights, a suite of applications built on Predix, GE’s operating system for the Industrial Internet, which produces clear, actionable data analysis from information collected from operating theatres’ anaesthetic machines. The algorithms developed and refined during the Christchurch pilot can now be deployed to give other clinicians clear visibility of how these pricey gases are flowing in their own operating theatres.

“It’s the perfectly virtuous project,” says French. “It’s not removing anything from patient care, it’s reducing waste and it has an environmental positive. Everyone can understand how this is a good thing.”

Agent cost anaesthetic dashboard
The beta dashboard that GE first presented to Christchurch Hospital was extensively overhauled to later become part of the Carestation Insights suite of applications.

Over the nine months of the pilot from early 2016, the anonymised data of some 5,000 patients was analysed and the beta dashboard that GE first presented to the hospital was extensively overhauled, based on feedback from Kennedy and French.

“The cycle of development has been extremely fast,” says French. “On both sides, we’re pretty confident playing in each other’s patch.” He and Kennedy made a couple of trips to the US, which included a workshop at GE’s Software Centre of Excellence in San Ramon, California, and had fortnightly conference calls with GE Healthcare’s data scientists in Madison, Wisconsin.  “It was really exciting,” say French, “because we were building off what we knew we could already do, and then you start seeing the huge opportunities in extracting data from your anaesthesia delivery units and remotely processing it.”

French and Kennedy expected the pilot to validate the many years of work and data audits that they’d done to make Christchurch Hospital a low-flow institution, so they were surprised by the findings. “It confirmed that we are a low-flow shop, but there are significant outliers: gas use can vary quite markedly, even with the same anaesthetist on the same list. You wonder why, and you can only answer that with really good data analytics.”

Kennedy came up with a model to investigate the gas flows in the two main phases of anaesthesia when anaesthetic vapours are used: the induction flow at the start of an operation, when flows are set high, and the maintenance phase, when the flow of the anaesthetic agent is reduced. They found that the high-flow phase at the beginning often continued longer than needed. This wasn’t, says French “because the anaesthetist is doing anything wrong, they’re just busy looking after the patient, but you can end up with the first few minutes of the anaesthetic costing as much or more than the entire rest of the anaesthetic.”

The unnecessarily long high-flow phase now seems an obvious thing to look out for, says French, but only with the hindsight the data has afforded them. “That’s what data gives you, it spurs you to go looking for the solution to your problems,” he says. “If you don’t have the data, you’ll never know to go and look.” Confident in the analysis of their hospital’s data, they acted on it.

Back in Christchurch, Kennedy ran an educational session on the findings for anaesthesia staff, followed up by more information via email. They arranged for the hospital’s admin department to send text-message reminders on the day an anaesthetist was working in a Carestation Insights-equipped theatre, reminding them to reduce their flow as soon as possible.

Assistants were also primed to remind the specialists. “These were low-level prompts, because we know that our staff will do the right thing once we’ve given them the information,” says French. “It’s important to use data to change behaviour across an entire department, not to go after individuals … a rising tide will lift all boats.”

And so it did at Christchurch Hospital. During the pilot, using the insights from the data analysis, the already low-flow institution decreased flow rates by an additional 13%, reducing the use of those expensive, environmentally harmful anaesthetic agents, and cutting anaesthesia costs by 12%. In a busy hospital—75 specialist anaesthetists, 30 trainees and around 32,000 surgeries each year—that adds up to significant financial and environmental savings.

“Huge data sets can be absolutely overwhelming,” says French. “You need data scientists and technology to make data human and relevant.” He and Kennedy have pondered that their finding of frequent unwarranted and wastefully long high-flow phases in operations now seems obvious, and yet no studies around it have been previously published and hospitals weren’t actively watching out for it.

“We’ve never had the good data to make us go think about it,” he says. “We weren’t expecting any massive insights into our own clinical practice, and it’s been gratifying and surprising that we got that. It’s a good moral to the story: you may think you’re doing well, but until you’ve got data to prove it, you can’t say that for certain.”

GE Healthcare has released Carestation Insights to the market, and French and Kennedy are presenting the results of the study at the prestigious Euroanaesthesia 2017 congress in Geneva.

“Every dollar that’s given to healthcare can only be spent once,” says French. “If we can avoid wastage of anaesthetic agents, we can spend those dollars on other things that benefit patients.”

Download the case study on the pilot that produced Carestation Insights here.

Load Comments

comments