A new computer-based decision support tool for sepsis at HCA Healthcare harnesses pivotal data in real time.
HCA Healthcare has developed an effective computer-based decision support tool for the early detection of sepsis.
Sepsis and the body's response to the infection is one of the deadliest medical syndromes in the United States, according to the Centers for Disease Control and Prevention. About 1.7 million adult Americans develop sepsis annually and the condition claims approximately 270,000 lives each year. About one-third of patients who die in hospitals succumb to sepsis.
The computer-based decision support tool is called Sepsis Prediction and Optimization of Therapy (SPOT), and it can detect sepsis 18 hours earlier than the best clinicians, says Jonathan Perlin, MD, PhD, president of clinical services and CMO at the Nashville-based health system.
"This is the future. Military fighter planes can't fly without decision support. Healthcare is equally complex. To think that we can manage all the variables without assistive technology is inconsistent with how we think about high-reliability endeavors like aviation and healthcare," he says.
HCA started adopting elements of the Surviving Sepsis Campaign in 2013. From 2013 to 2017, sepsis mortality at HCA's hospitals fell 39%.
The health system launched the SPOT initiative in 2018. From 2017 to 2018, sepsis mortality at HCA's hospitals dropped nearly 23%. "SPOT doubled our effectiveness in surviving sepsis," Perlin says.
The health system estimates that the combined effort of adopting the Surviving Sepsis Campaign and SPOT has saved about 7,800 lives.
How SPOT works
SPOT features an algorithm embedded in HCA's electronic health record that was built with Red Hat open source software. To indicate the onset of sepsis, the SPOT algorithm combines factors such as patient demographics data and medical history with continuous monitoring for signs and symptoms of sepsis as well as key elements of clinical care:
- Body temperature
- Blood pressure
- Heart rate
- Platelet count
- Laboratory tests
- Patient transfers such as moves to an ICU
"The SPOT algorithm surveils 24 hours a day, seven days a week to look for the signs and symptoms of sepsis. When those signs are found, they are teed up and presented to the caregivers," Perlin says.
When the algorithm detects a likely case of sepsis, SPOT initiates an alert similar to a heart attack or stroke code that prompts clinical care teams to take action. Caregivers who receive the alerts include telemetry units, nurse leaders, sepsis code teams, and rapid response teams.
An essential component of the SPOT initiative is the algorithm's diagnostic accuracy, Perlin says.
"We were able to train the algorithm to be more than 100% sensitive—we picked up cases of sepsis that the care providers did not see, and our rate of false positives was half that of care providers. So, the specificity was twice as good as clinicians. It not only improved care but also the efficiency of doctors and nurses," he says.
He continues: "On their own, clinicians can look piecemeal for sepsis signs and symptoms; but the computer can constantly look for those signs and symptoms, and when the computer has a hit, that information is immediately given to the caregivers at the bedside. That signal is not just an alert but also a representation of the sepsis criteria, so there is credibility and explainable data."
How the SPOT algorithm was developed and implemented
Three primary steps led to the development and implementation of SPOT.
1. Robust EHR and data management capabilities: The foundational step that made SPOT possible was the adoption of meaningful use and a data warehouse at HCA a decade ago, Perlin says.
"The past 10 years of building from meaningful use to become a learning health system created the platform for doing things like SPOT. We realized that we would have tremendous power through the data warehouse to learn at scale. Part of the rationale for the data warehouse was to be able to have a resource to be able to train computer algorithms through machine learning and other applications for artificial intelligence to support clinical workflow more effectively."
2. Pilot phase: Before SPOT could be implemented at more than 160 HCA hospitals, the algorithm had to be tested and proven effective, he says. "The computer algorithm was developed through our data warehouse. We piloted the algorithm at a couple of our facilities to test it against clinicians. At a certain point, the algorithm started to outperform the clinicians, and we began to implement SPOT."
3. Clinician engagement: During the launch of SPOT at HCA hospitals, the decision support tool was presented as a way to put critically important information into the hands of clinicians, Perlin says. "To gain acceptance at the bedside, we didn't just say, 'The computer sees sepsis, start treating it.' We said, 'This is what the computer sees; do you agree?' "
The SPOT algorithm's accuracy was vital to the clinician engagement effort, he says. "In addition to being able to show the clinicians what the computer saw, the fact that there were not burdensome false positives was incredibly important. We didn't waste people's time. There have been other sepsis algorithms at other institutions, but some of them have been turned off because there were so many false positives."
An exciting benefit of the SPOT initiative is harnessing HCA's electronic health record and data warehouse to improve clinical care, Perlin says.
"All of us hear about the burdens of electronic health records, but without electronic health records we couldn't have done SPOT. We think this is one of the ways we can drive excellent care at scale. It's the payback for the challenge of using electronic health records. It's been received with a great deal of enthusiasm from clinicians because the caregivers at the bedside see this in real time saving lives," he says.
SPOT represents a significant step forward in the nationwide effort to develop computer-based decision support tools in healthcare, he says. "This is an extraordinarily exciting time to be in healthcare and at HCA because we can cast data at scale. Our data warehouse and computer algorithms like SPOT are bringing us closer than ever to our mission, which is a commitment to the care and improvement of human life."
Christopher Cheney is the senior clinical care editor at HealthLeaders.
An algorithm developed at HCA Healthcare can detect sepsis about 18 hours earlier than the best clinicians.
In one year, the HCA algorithm reduced sepsis mortality by 22.9%.
Since 2013, sepsis initiatives at HCA have saved about 7,800 lives.