Connected Health: Big data has yet to produce big ROI

BOSTON—“I don’t think big data has generated a big return on investment for sick or healthy people,” said Chris Kryder, MD, MBA, chairman of Valence Health, speaking during a panel discussion about big data’s impact on healthcare at Partners HealthCare’s 10^th Annual Connected Health Symposium.

Michael Weintraub, chairman and CEO of Humedica, however, said “we’re seeing seven-figure reductions in costs by keeping people out of the emergency department, pushing information to ensure follow up and compliance with medications.” He said some organizations are publishing 75 percent reductions in readmissions for patients with a variety of heart ailments.  

Stephen McHale, CEO of Explorys, also said he’s seen “tremendous results by engaging teams. The hardest part is getting people to learn how to deliver medicine differently. That’s critical to driving down the cost of care.” The majority of time is spent changing behavior, he said, which “wasn’t our plan. The change management issues here are significant. We’ve changed the way we approach the effort and resources to apply into this environment.”

Physician behavior won’t change until the incentives really change, said Kryder. “We’re at the early stages in moving from pay-for-performance to risk. When that moves, the ROI will be dramatic. Right now we have information overload. It’s very hard to generate valuable nuggets.”

When asked what lessons healthcare can learn from other industries, Jean Drouin, MD, MBA, director of McKinsey & Company, said there are 600 specific customer journeys in banking. “To fundamentally improve business processes, we need to get that granular in healthcare.”

“There’s a saying that culture eats strategy for lunch. Your strategy doesn’t matter if the incentives and economics aren’t there,” Weintraub said. “The reality is that big data currently is compartmentalized. It means different things to different people. It’s going to evolve very quickly.” Provider organizations can’t just look at inpatient data, he said, because patients go to clinics and other points of care. And, providers won’t use the data unless they are “pushed to the point of care, influencing easy actionability. Knowledge has to meet action or it won’t be used.”

Weintraub also said the industry is only in the second inning and predicts three-month phases of advances over the next three to five years. “Once information is pushed to those who consume the data, we’ll start seeing the investment eclipsed by the savings. It’s going to take several years and a bunch of failure.”

The role of the EMR will continue, according to Weintraub. “There has always been room for business intelligence, science and analytics to the left and right of the technology systems that are the plumbing.” With increasing consolidation, he said providers will end up with practices that have a range of EHR systems. “There’s a role to homogenize all that and create a common data store.”

When asked about the future of reimbursement, Kryder said his company’s “one single operating goal is to help providers become payers. I think it’s over for the payers—I just don’t know when. As we remove the complexity of the third-party system, it’s going to be better for patients. It’s going to happen.”

But Drouin said plans won’t sit idly by. “The functions that they provide will be required whether the ultimate owner is a provider or a payer. There’s an enormous amount of diversity in the payer space right now. It does appear that the trend is toward disruptors that are becoming more integrated. That trend will take awhile to wend its way.” Meanwhile, he said payers are going to focus on using data because of the pressure they face to lower their costs and simplify their business.

Non-healthcare data could play a growing role, especially in big data, said Weintraub, because factors such as education level impacts medication adherence and how willing a person is to accept communication from various sources. “With hypersegmentation of the public, you can use that behavioral information. Behavior consumption patterning information is available today using cohort matching and statistical analysis. We can predict, with high degrees of confidence, that certain cohorts will behave in certain ways.” With 40 percent of Americans having at least one chronic disease, that means a lot of people are consuming drugs. There are ways of using consumption pattern data and behavioral data and “mashing that with specific big data at the clinical and financial level to impact what you do to push that information into the workflow and mind’s eye of the clinician.”