When Caregivers Huddle for Health, Data Makes a DifferenceMay 2016
Topics: Healthcare IT, Software (General)
Seven thousand patients. Three primary care physicians. If that sounds like an unbalanced equation to you, you’re right—but it's the reality at many healthcare practices nationwide.
In such an overcrowded system, how can healthcare providers be proactive to keep their patients healthy? One emerging practice called the "care team huddle" is helping ensure patients get the care they need. And now, a MITRE research team is working to improve that process with the Intervention Engine (IE) platform. IE not only focuses on proactive care for the highest-risk patients, but will eventually use the data to improve care based on patient outcomes.
Huddles Take a Team-based Approach to Healthcare
In a care team huddle, physicians, social workers, nurse practitioners, and other specialists meet frequently to review a curated roster of patients and assess which, if any, interventions each one needs. Huddles evolved from the lean management principles popularized by Toyota. The car company took the focus off individual machines in a factory to the flow of the product through the total process.
The idea is similar in a clinical setting. For example, an office administrator might discover that a patient often misses scheduled visits due to a lack of transportation or lacks the financial means to pay for a prescription. These issues might never come up with the doctor. But they have a definite impact on the patient's overall health.
According to Chris Moesel, a software engineer who specializes in health technology and leads the IE team, huddles ensure that every caregiver who sees a patient has a more holistic understanding of the patient’s risk factors and circumstances. This consistent view allows them to plan interventions—from home visits to switching to a generic prescription—accordingly.
But it's not perfect. Prepping for a huddle is usually a manual process, often involving time-intensive and error-prone tools such as Excel spreadsheets. In addition, historical data records spread across multiple systems, and not in the spreadsheet, means the huddle team can still miss important opportunities.
Intervention Engine Replaces Recollections with Data
Intervention Engine replaces the clunky spreadsheet with a software platform that makes care team huddles easier and much more data-driven. Developed with an open source Apache 2.0 license, it integrates custom population filters, risk assessment algorithms, and modern visualization techniques into a web-based interface that helps the care team select and discuss patients for each huddle.
The care team can view each patient's high-level medical record, risk scores, and other pertinent information. "These tools help drive the discussion," Moesel says. "The result is that the care team can plan interventions based on data, rather than recall."
For over a year, Moesel and his team have been working with MedStar Health’s research division, MedStar Health Research Institute, and MedStar Good Samaritan Hospital's Center for Successful Aging in Baltimore, Maryland. The Center helps patients with the specific healthcare needs of a geriatric population, from chronic conditions to dementia to fall prevention. They're early adopters of the huddle system. Together, they and MITRE are working to improve it.
Intervention Engine's focus on risk stratification is an example of how MITRE and the Center for Successful Aging hope to improve the traditional huddle concept. Currently, the Center already uses a risk stratification model for care coordination. By integrating that risk stratification model into Intervention Engine, the software can intelligently schedule huddle discussions based on patient risk scores and circumstances. Rather than allowing scheduled office visits to determine when a patient is discussed, these data-driven huddles can promote discussion when the patient needs it most, regardless of schedule.
Currently, the Intervention Engine runs in a test environment, using some Center data. The MedStar and MITRE teams meet regularly to discuss the project status, demo the platform, and get feedback that helps Moesel's team refine the design and add new features.
While IE is not yet used at the point of care, that milestone will come during a pilot program tentatively scheduled for this summer. During the pilot, the MITRE and MedStar teams will collect data to compare outcomes for patients with IE-enabled huddles to a control group receiving general care.
They expect the study to show that using huddles may increase staffing costs for preventative care, but that increase will be outweighed by a decline in costly interventions such as ER visits and hospital admissions.
Teaching Intervention Engine to Learn—and Advise
The team has another goal for the Intervention Engine: to become a "learning health system." This means the software can use the data to look at patient outcomes and make recommendations for other patients.
"The purpose of huddle discussions is to be more proactive," says MITRE's Andy Gregorowicz, who initiated the project with MedStar last year. "Right now, IE supports the discussion, but it doesn't provide a way to record the results, track them, and see how they've worked.
"We're looking at allowing the software to look back over time to see if particular interventions were effective. If we can actually see that certain interventions work better, we can help guide the care team to use them more often."
The team is equally excited to provide Intervention Engine as open-source software. "Using the Apache 2.0 license lets other providers take this software and run it on their own network—which aligns with MITRE's mission to work in the public interest," Gregorowicz says.
"We see this as one of the best avenues to get this technology into the hands of healthcare providers, so they can use it to deliver value to their patients, faster."
—by Tricia C. Bailey