HIMSS14: Predictive analytics with an eye to the future

ORLANDO—Providers can achieve big savings if they focus their lens on preventable visits. Baylor Quality Alliance, a Texas-based clinically integrated accountable care organization (ACO) of the Baylor Health Care System, was highlighted in a session at the Health Information and Management Systems Society's annual conference that showed how utilizing predictive analytics successfully targeted high-risk, high-utilizing patients.

To be successful at analytics, providers must focus on a small subset of patients that represent the bulk of their costs, said A.G. Breitenstein, JD, MPH, vice president of providers at Humedica, a Massachusetts-based clinical intelligence company in partnership with Baylor.  

But they must look beyond past utilization to the “most overlooked” approach to analytics: exploring ways to prevent cases from happening in the first place. This entails asking questions like what is possible in the next one to three years during a meaningful relationship with a patient to improve outcomes, and how much to invest in those opportunities, Breitenstein said.

Timely delivery of insight into which patients require intervention also is essential. “You have to ask what kind of runway you need to effectuate those interventions.”

At Baylor Quality Alliance, the ACO sought to figure out ways to gather data and create an environment where it’s usable, said Michael Sills, MD, CMIO and vice president of informatics technology. The Medicare Shared Savings Program looked to several areas to improve its care: utilizing a narrow network of high performing providers, data aggregation and integration of clinical and claims data, and financial modeling.

“Our system has 700 employed docs and 42 hospitals. It gives us the ability to approach care in an integrated way that hasn’t happened before,” Sills said, also noting that they represent 55 EMRs.

Through analytics, the system identified the 880 chronically ill patients that compromised 60 percent of the costs.

“We were really surprised by the number of people who had a high HCC scores,” he said. Of the 44 patients with the highest risk, five hadn’t been seen in three months. “That was startling.”

The analytics also revealed that more than 20 percent of patients had high blood pressure. “Clearly there are lots of opportunity for improvement. These are not people previously identified with hypertension,” he said.

Once identified, the patients at highest risk received outreach from care coordinators who scheduled an appointment for them. The care coordinators, who attended these appointments, showed the algorithms to the primary care provider to better inform them of their patient’s risks.

“We need to take information into the structure that allows us to come up with information that is usable and actionable without threatening anyone. When we do that, we change outcomes and improve access by finding patients who need to be seen the most and creating an environment in which to see them,” said Sills.