Skip to content Skip to navigation

Developing a Predictive Data Analytics Culture

June 16, 2015
by Gabriel Perna
| Reprints
As the leading healthcare organizations move towards utilizing predictive analytics to improve care and reduce costs for patients, they’re starting to understand what it takes to develop a culture that nurtures this technology. 
Representatives from three provider organizations, Inova Health, a health system in Northern Virginia; Adventist Health, a Gaithersburg, Md.-based health system; and Unity Health Care, a Washington D.C-based federally-qualified health center, shared their predictive analytics experiences at the Institute for Health Technology Transformation’s (iHT2) Washington D.C. Health IT Summit. iHT2 is a sister organization of HCI, under the corporate umbrella of the Vendome Group. 
With most Summit attendees saying they are 2-5 years away from developing predictive analytics, panelists talked about the lessons they’ve learned thus far. Cherie Pardue, Deputy CIO at Adventist, stressed the importance of getting stakeholder buy-in. 
“One of the things we’ve worked on is educating people,” Pardue said. “The COO [has done] an amazing job of highlighting the data and how important everyone’s role is in contributing to the data and interpreting the data. That was the first step into turning our culture into an informational-driven one. Quite often, we get the reports and people don’t understand the role they play in the data that’s presented on the reports with the accuracy.” 
This buy-in also involves breaking down what Pardue says are “information fiefdoms,” between various clinical and operational departments. This means building relationships and getting the different areas of the organization to understand what they’re trying to do. “Is it a perfect world and do we have perfect governance yet? No. But it’s a start,” she said. 
Kim Krakowski, R.N., Director of Informatics and Innovation at Inova, talked about getting buy-in from physicians when using informatics across the enterprise. She said her organization has engaged super users, both from primary care and specialties, and that’s gone a long way. At Unity Health Care, Karin Werner, VP of Strategic Development, said the lean nature of her organization allows her to engage end-user stakeholders easier and understand what they need. 
“A lot of our work finding out what the end users really wanted. We took a quality improvement stance. Instead of creating the reports the executives wanted the end users to use, we asked them what bothers them. What is the one thing that would make their day easier,” Werner said. “That’s helped a lot.” 
Another recurring theme from panelists was focusing on specific predictive analytics-driven efforts, such as tackling sepsis, as both Pardue and Krakowski’s organizations have done. Adventist, Pardue said, has created an algorithm that alerts providers to patients at risk for sepsis. It’s also developed a risk stratification score for readmissions within its accountable care organization. 
The latter project brought on another lesson. “You’re never finish truly collecting or massaging the data. As you work through it, there are always things to tweak and refine,” Pardue said. 
The work they’ve done has been successful thus far, having gone from a rate of 20 percent readmissions to approximately 8 percent, Pardue said. Within that targeted population, the organization is narrowing it down even further by looking at diabetes patients.
The provider panelists closed out the session by revealing what drove their organizations into a predictive analytics culture in the first place. Both Pardue and Werner mentioned business and financial reasons. Pardue said Adventist’s ACO cannot generate revenue based off of value without predictive data. Werner had pointed comments on what gets an organization to believe in predictive analytics. 
“Time and money are what matters to people. If you can find a way to automate a manual process that will save someone time. Those were pieces that could drive change, people had an incentive to look at the data on a regular basis if they could save an hour of time,” Werner said. “Also, we implemented a performance-based incentive plan for physicians. Believe me, they are very interested and engaged in the data now.”