Data Analytics

How Predictive Analytics Can Play an Influential Role in Operating Rooms

The added foresight helps improve patient outcomes and boost operational efficiencies.

Andrew is the web editor for HealthTech magazine. His experience includes marketing for a major IT services company and digital strategy for WashingtonExec.

A collaborative mission drives the staff at UCHealth: Keep costs down while improving quality of care. Accomplishing this goal can be tricky, but current models of predictive analytics for healthcare are helping achieve that balance.

“Our use of analytics creates a win-win-win scenario for patients, payers and providers,” Steve Hess, CIO of UCHealth, tells HealthTech. “And that’s something that is very rare in healthcare.”

Four years ago, the Aurora, Colo.-based organization began moving down a path toward predictive analytics by focusing on patient surveillance in outpatient care. And that has since compelled UCHealth to leverage an existing universal database to better predict patient outcomes in clinical settings — including operating rooms.

Dr. Clint Devin, an orthopedic spine surgeon at UCHealth Yampa Valley Medical Center in Steamboat Springs, Colo., helped launch that database in 2008 while working at Vanderbilt University School of Medicine. Formerly known as the Quality Outcomes Database (QOD) Spine registries, the effort was established in coordination with nonprofit NeuroPoint Alliance to collect, sort and analyze the safety of care, patient-reported outcomes and satisfaction of care across multiple healthcare settings in spine surgery.

Overall, the database is intended to record critical variables on patient outcomes through opt-in questionnaires before and up to a year after elective surgery. These questionnaires are designed to provide doctors with actual and actionable data that can inform patient management decisions.

“We all know that there’s a lot of patient characteristics that really affect how people do, even if there's an indicated procedure done and it's executed technically well,” Devin tells HealthTech.

Despite growing interest in and supportive evidence for predictive analytics in a clinical setting, the approach will take time to become an industry standard.

A 2019 Deloitte survey of 56 health system CTOs, CIOs and chief analytics executives found that 84 percent believe analytics will be an important part of their strategies in the coming years. But only 36 percent of those same leaders believe analytics are extremely important today.

Devin Helps Establish the QOD, Brings It to UCHealth

UCHealth first adopted the QOD registries into its spinal surgery workflows in 2018. But in order for the database to be utilized for predictive analytics, an immense amount of data — about 10,000 patients’ worth — had to be collected. That process took Devin and his NeuroPoint Alliance partners nearly four years.

The legwork was necessary, however. “You have to get a huge number; for every variable you test, you have to have a certain number of patients so that the analysis being done is statistically sound,” says Devin. “If you have too few patients, things will show up as being statistically significant but in fact they're not because you just don't have enough patients to really make that judgment call.”

The key with these predictive analytics is finding those modifiable characteristics. You could say, ‘Hey, if you stop smoking or if we got your opioids down for two months, here’s how it would change that.’"

Dr. Clint Devin Orthopedic Spine Surgeon, UCHealth

Data collection can be challenging. Although most of the work falls on a site coordinator, surgeons are ultimately most influential in encouraging patients to participate, Devin says.

“I just tell them the importance of it, why we do it, and that you're going to be receiving an email, and it's important to me that you fill it out, and here's why,” says Devin, noting that UCHealth staff offer patients the option to complete the form together via tablet or collect details by phone.

Thanks to those efforts and positive response from participants, the QOD has grown to become the largest spine registry in the world. In September 2019, the American Association of Neurological Surgeons and the American Academy of Orthopaedic Surgeons announced a new partnership, the American Spine Registry, which will be owned and developed by both organizations. As the AANS states in its news release, the ASR is intended to transform the QOD Spine registries “into a more far-reaching program that facilitates the participation of all North American spine surgeons in a shared, quality data-collection platform.”

UCHealth is now beginning to use the data from the QOD registries, coupled with data from a patient’s electronic health record, to predict surgery outcomes in its operating rooms.

UCHealth Moves Predictive Analytics to the OR

For some patients dealing with severe pain, spine surgery can be a logical and effective remedy. Yet despite what imaging methods or examinations might suggest, outcomes can be affected by preoperative conditions.

By introducing predictive analytics into their ORs, UCHealth helps draw a connection between those conditions — such as a patient’s weight, health habits, opioid use or a past reaction to anesthesia — and potential outcomes before a surgery even takes place. It can also reduce the odds of costly readmittance.

“The key with these predictive analytics is finding those modifiable characteristics,” says Devin. “You could say, ‘Hey, if you stop smoking or if we got your opioids down for two months, here’s how it would change that.’”

Predictive analytics technology is also helping the organization to cut costs and operate more efficiently. By using tools that pull from EHR data, the organization is able to free up capacity in its ORs and create a more transparent process for measuring utilization.

For example, analytics have shown that if an operation is done on a Monday or Tuesday, a patient’s length of stay tends to be shorter by almost a day compared with those performed on a Thursday or Friday.

“You could see how hospitals would care about that,” Devin says. “They may front-load their surgeries, and then do other smaller outpatient procedures on Thursday and Friday, knowing that any of the inpatient procedures would then have a shorter length of stay, potentially improving their margin, which matters.”

Do Predictive Analytics Make Sense for All Healthcare Organizations?

As long as data has been around, people have tried to use it to predict outcomes. The stakes, though, are significantly higher when it comes to predicting patient outcomes that are potentially life-altering. It’s one reason why some healthcare organizations hesitate to get on board.

“I would say that healthcare is still very skeptical of machine learning and AI,” says Hess, the UCHealth CIO. “Once you start predicting deterioration and outcomes, you hear the age-old mantra of ‘my patients are different.’”

Hess agrees with the fundamental idea that each person is unique but wants to emphasize that predictability of outcomes is improving through this technology. By marrying clinicians with data scientists, organizations gain deeper insight into data patterns and delivering better solutions.

For organizations wishing to explore the use of predictive analytics, Devin suggests they start by contacting the American Academy of Orthopaedic Surgeons or the American Association of Neurological Surgeons, both of which will have the American Spine Registry. Another method would be to embed predictive analytics directly into an organization’s electronic health record workflow via business intelligence solutions.

“I think hanging a shingle and just providing care isn’t personally enough anymore,” says Devin. “I think you need to be doing some of this stuff to really improve patient care.”

aldomurillo/Getty Images