Analytics, accountable care, research set to improve care outcomes

BOSTON—With studies indicating that only 55 percent of patients receive the recommended care, “there are big opportunities for improvement," said David Bates, MD, MSc, chief innovation officer at Brigham & Women’s Medical Center in Boston, speaking at the Big Data & Healthcare Analytics Forum.

Accountable care is being used for both quality and cost, “but many of us haven’t figured out what our populations are," added Bates. "We don’t have tools to sort that out. We’re still at the starting gate.” Accountable care pushes toward a paradigm shift to care coordination and a focus on the population. “It’s a hard transition to make. Health IT is key.”

Walmart is one company that has invested a lot in analytics, Bates said. The company tried to figure out how best to plan for coming hurricanes with what goods to supply. Analytics showed that the two biggest sellers are beer and strawberry Poptarts. They loaded up at the next sign of bad weather and did very well.  

Although “healthcare has been pretty late to this particular party,” efforts are underway, according to Bates. For example, the Partners Big Data Commons links disparate islands of patient data across Partners and the academic medical centers onto a common platform. It includes both clinical and research data.

Partners also is working on setting up a phenotype discovery center to include DNA, blood and images. Patients who consent to inclusion of their specimens will allow for advanced processing, Bates said.

Big Data in clinical care aims to better control the spending for high-cost patients. About 5 percent of patients account for half of all spending. The first step in managing a population, he said, is identifying this group and managing them pretty aggressively. Bates said Partners has been able to bend the cost curve quite substantially. High-cost patients get a case manager and socioeconomic status, mental and behavior health and other information is collected because it is “enormously predictive of how someone will do. About half the time people with a mental health issue won’t do well. Others have social needs or lack of transportation. If we don’t find that out and deal with it, we won’t get to a point where we can be much more cost effective.”

Readmissions is another issue on providers’ radar but “they’re tricky because some are not preventable.” But, Bates said everyone should be using algorithms to measure the likelihood. The key differentiators, he said, are tailoring interventions to the patient. How do you get the patient into the doctor’s office within a week of discharge? It’s not enough to get the appointment scheduled. If the patient doesn’t show up, you need a Plan B. “Building systems to accomplish that is pretty hard.”

Although clinical data are now nearly ubiquitously available most organizations haven’t yet figured out how best to leverage them. “Novel approachs are most likely to provide further improvement,” Bates said, but healthcare needs more research based on experiences in other industries.

His own organization is “not where we need to be,” Bates said. They cut back on their analytics investment since reimbursement was so poor. While they receive a lot of patient-reported outcomes data, they are trying to build up their capabilities around that.

The good news is that “everybody wants to do better. Many of our systems make it hard to do that.” But, “data analytics will be foundational in every care redesign effort. If we do it well, we will be able to do much more.”