Skip to content Skip to navigation

In Ontario, Using Medical Device Integration and Predictive Analytics to Improve Clinical Outcomes

October 13, 2015
by Rajiv Leventhal
| Reprints

Earlier this year, Alison Fox-Robichaud, M.D., led a study with the aim to determine the pattern of decline prior to an inpatient cardiac arrest. Specifically, Dr. Robichaud and her team implemented the Hamilton Early Warning Score (HEWS), system, created at the Ontario, Canada-based Hamilton Health Sciences, within its electronic vital signs documentation to track and trigger care for deteriorating patients. Multiple early warning scores have been developed and implemented to reduce cardiac arrests on hospital wards, but in contrast to previous observational studies, the research team chose to implement a score modified from the published EWS system using the consensus opinion of a steering committee and evaluate the score in real time, according to the study’s analysis.

The research team came up with a series of seven components, a mixture of vital signs and observations, and when they took those values and put them together, it would give them a “score.”  Based on that score, you could predict if a patient was tending towards an event that could lead to a cardiac or respiratory event, explains Mark Farrow, CIO of Hamilton Health Sciences, a medical group consisting of seven hospitals with more than 1,100 beds. Since the project took off, Hamilton Health Sciences has gone live with a system that automatically feeds data from over 400 vital sign monitors into the electronic medical records (EMRs) of patients.

The HEWS system, created in less than three months leveraging software from Boxford, Mass.-based  Iatric Systems, is powered by an algorithm of various, real-time data points from the devices that predict which patients are at a higher risk for a medical emergency. By putting this data to use, the hospital is able to deploy a designated team to intervene before a “code blue” is called, which has led to a significant reduction in such code blues and unplanned ICU admissions, according to Farrow.

As such, Hamilton Health Sciences is using integration and predictive analytics to provide clinicians with a plethora of data and ways to use that data, enabling them to often act before a patient situation worsens. “It quickly became clear that having this information in real time would allow us to alert our critical care teams so they can intervene with a patient before they got into serious trouble. Obviously it’s lot easier to deal with a patient before they have a cardiac arrest compared to dealing with a cardiac arrest and bringing the patient back to a satisfactory point,” Farrow says. .

He adds that the “real-time” aspect makes a world of difference. “It doesn’t help to capture a lot of electronic health information, but then eight hours later at the end of your shift to come back and say your patient probably had a cardiac arrest four hours ago. Instead we can say that we think your patient could in trouble in the next couple hours and you may need to intervene in their care to move them in the right direction. Our system shows this,” Farrow says.

The vital sign monitors that were recently integrated are located in the hospitals’ ICUs, CCUs and neonatal units, but Farrow says more focus now will be on other ward levels because integration already exists in the ICU and CCU from monitors into Hamilton’s Meditech EMR so that clinicians can be analyzing that data as it comes through. “A key to project was around the ward where you don’t have the patient hooked up to the monitor most likely. How do you capture that information as quickly as possible?” Farrow asks.

According to Farrow, the scoring system starts with a 1, for a healthy person, and goes up to a 5, for one at serious risk. By capturing the data from devices and looking at other components, such as visual signs, a score will be generated. “We can say your patient is trending towards more challenges such as breathing or getting oxygen, and can most likely end up with heart attack, which is a code blue,” Farrow explains. “We are using the evidence and science that says by tracking these six or seven pieces of information you can use predictive analytics to say what a patient is going to do. It requires real-time documenting, putting it into a system that will allow you do to simple analytics. It’s not difficult analytics. And then it’s pushing that score back to the [caregiver] as well as alerting the care team that there’s a potential problem,” he says.

To this end, Farrow says that predictive analytics is the next big wave in terms of being able to improve outcomes. “To me this is really getting to the crux of why we do what we do,” he says. “For so many years we documented, but at the end of the day a large part of that documentation is for medical, legal, and billing reasons. We should be documenting for care reasons,” he says, adding that it becomes useful for care when you can do analytics in real time and start to do the predictive analytics to provide information back.

“This isn’t artificial intelligence; it’s predictive analytics saying the trend is showing a certain thing and then letting a health professional saying ‘thank you let’s take the next steps.’ At first you get a sigh in return, but when you can tie it to outcomes and benefits, it changes attitudes. Nursing is about caring for patients, not about documenting,” he says. Farrow adds, “What we’re trying to do is prevent deaths or unplanned ICU admissions. We have had good success with that. Taking discrete data and bringing it together to do predictive analytics will open a huge new frontier for how we care for patients.”