Health System Uses Predictive Analytics to Reduce Readmissions
Advocate Aurora Health, a large health system with dual headquarters in Downers Grove, IL, and Milwaukee, is reporting success with a program that uses predictive analytics to identify outpatients with an increased risk of unnecessary hospitalization. Those patients are then provided special intervention to prevent admissions.
The health system uses a predictive modeling platform that integrates 30 to 40 sources of data, explains Tina Esposito, vice president of information and technology innovation at Advocate Aurora Health.
The program was developed in 2012 as part of the health system’s move to value-based care, Esposito says. Advocate Aurora Health has an accountable care organization with more than 1 million participants, so there is a strong incentive to prevent unnecessary hospitalizations.
“As we thought about how we could be successful for our patients in the new model of care, we realized there was a bit of a gap in understanding how they moved through our health system. In a fee-for-service world, you’re very focused on today and the visit at hand,” Esposito says.
“So our data had been very siloed in that way, with hospital data in one silo and home healthcare data in another, and we wanted to look at this in a much more holistic way,” she continues. “A primary first step was just getting our data organized in a way that would allow us to understand how care was being delivered in our system to the patient overall, rather than just episodes of care.”
That required bringing on more experts in data analytics. Once the health system had a better grasp on its data, it began looking for ways to apply it to patient care. Population healthcare managers approached hospital leadership with the idea that they could be more successful if they could better leverage the data for patients at risk of certain utilizations.
Advocate Aurora Health leaders realized that they needed to use data that allow an intervention in time to make a difference in preventing hospitalization, not long after the opportunity was gone.
“We understand now that once you identify a high-cost patient, that patient doesn’t necessarily stay high-cost, but a big realization was that the care managers were very dependent on claims data, and that is very latent data,” Esposito says. “If you have latent data, by the time you see that something has occurred to the patient and try to dispatch a care manager or any other intervention, that patient likely has already regressed to some baseline level of spending. So you’ve now leveraged a resource that in all likelihood isn’t needed any longer but could have been effective months prior.”
The health system first used the model on heart failure patients at high risk for unnecessary utilization.
The program is designed to be prescriptive in its approach, Esposito explains, focusing on a disease-specific action plan that can prevent unnecessary hospitalization. A key goal is reducing subjectivity in the care management process.
The pilot program determined that patients who are actively engaged with another care team in the health system, such as those addressing transplants or active cancer, are not a good fit for this approach.
The model includes educating patients about their conditions, symptoms to watch for and how to respond, and frequent contact from care managers by phone and in person.
“Part of the intervention is to get the patient ready to no longer need these regular phone calls. We think it is important to have these patients graduate to a level of self-management because you will never have enough care managers to continue this attention indefinitely,” Esposito says.
The average length of time in the program was 70 days.
With 350 patients involved in the pilot, Advocate Aurora Health achieved a 23% reduction in hospitalization, ER use, and observational stays. Half of them achieved all the milestones in the model’s prescriptive workflow.
Esposito says the following were some of the key lessons from the pilot study:
• A predictive model alone does nothing to keep patients out of the hospital. Directed intervention with a disease-specific action plan is required to get results.
• Knowing a patient is at risk of an event doesn’t necessarily mean one can do anything to prevent it.
• Connecting with patients early and often is essential.
• Focus on chronic disease self-management first.
• Provide care managers with clear objectives and milestones to ensure consistency across the team. Hiring care managers focused on key attributes such as a commitment to improving patients’ health, refined phone etiquette, and a personality that was engaging and authentic.
The health system plans to expand the approach to other conditions, such as COPD. Esposito says the program is an example of how data analytics can affect the bottom line, but only if used strategically.
“There is no ROI in analytics unless someone does something with the information you’re providing them,” Esposito says. “The partnership with operations and clinicians has to be very, very tight to ultimately realize any value. Whatever analytic endeavor you’re after, you have to make sure it’s aligned to a very tangible business goal, rather than being just an academic exercise.”
Advocate Aurora Health is reporting success with a program that uses predictive analytics to identify outpatients with an increased risk of unnecessary hospitalization. Those patients are then provided special intervention to prevent admissions.
Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.