Big data’s big puzzle: Now what?
HHS’ partnership this month with OptumLabs is likely to add momentum to the push to use big data to lower healthcare costs and improve the quality of care delivery.
But the announcement also highlights the divide between health systems that are ahead of the curve in working with sophisticated data analytics and those that are still learning how to collect and use that information.
For decades, many industries, such as financial services, have used data analytics to determine their strategies for everything from marketing to maintenance scheduling. But the healthcare sector as a whole has lagged in using analytics to learn about its customers’ behavior and how to influence it.
“Clearly there are some places that are really innovative, but most systems are just starting this journey,” said Murtuza Mukadam, global head of healthcare strategy and solutions at consulting firm Virtusa.
Through the new partnership, HHS agencies will gain access to OptumLabs’ big-data resources, including de-identified medical claims and clinical data. OptumLabs is a research and innovation center established by UnitedHealth Group’s Optum healthservices division and the Mayo Clinic.
The first research project, to be led by the Agency for Healthcare Research and Quality, will compare the results of the government’s Medical Expenditure Panel Survey to OptumLabs’ real-world claims data.
Large systems like Mayo and Geisinger Health System, which in 2013 launched its own technology and analytics group, xG Health Solutions, are starting to apply their research findings to changing their standards of care. Geisinger has spent $200 million on information technology over the past two decades. By using big data to identify its sickest patients, it has reduced hospital admissions by 27%, the system reported at a 2014 health IT conference.
Other systems are just starting to dabble in data analytics. Until recently, much of healthcare’s IT efforts have focused on adopting electronic health records, or their digital infrastructure, as the government has pushed that transition through meaningful-use payments, Mukadam said.
It’s only recently that providers have turned their attention to data analytics. As more healthcare providers take on financial risk for care outcomes, they are finally seeing the financial incentives for making the analytics investment.
“People are waking up,” said Munzoor Shaikh, a director in the healthcare practice of consulting firm West Monroe Partners. “People are just starting to understand the problem a bit better.”
Yet one of the biggest barriers to effectively using data analytics is confusion over how much and what type of data to collect, according to a survey from Stoltenberg Consulting, which collected responses at this year’s Healthcare Information and Management Systems Society meeting.
About half of respondents (51%) cited “not knowing how much or what data to collect” as the biggest hindrance to using data analytics, while 33% said their organizations did not know what to do with the data they collected. Even the term “big data” elicits confusion, because it can cover anything from administrative data, to clinical data, to patient-generated data.
Health systems are gathering terabytes upon terabytes of data each day through multiple sources, Mukadam said, from claims information, to physicians’ EHR notes, to data from consumers’ mobile and wearable devices.
“If there’s no good, low-cost way to capture that data, it’s going to be pointless,” he said. “I don’t think hospitals are even there yet. A lot of them don’t even have the right skill set.”
A number of vendors have stepped in to help providers manage the load. Mukadam pointed to Hadoop, an opensource platform, which is making it cheaper and simpler to build data lakes.
Quest Diagnostics is targeting the physician market through a partnership with Inovalon, an analytics software company. Quest reaches about half of the country’s physicians through its laboratory services.
The partnership’s on-demand Data Diagnostics service goes live at the end of the year, said Dan Rizzo, Inovalon’s chief innovation officer.
“You’re still in this world of big-data application,” he said. “It’s amazing how much data are still in a disconnected EHR. Where you will see it picking up is when you have entities bringing together different sources.”
While bigger health systems and insurers tend to have more data, they’re not necessarily using it more effectively, Shaikh said. For organizations both large and small, the best place to start is to zero in on a single question they’re trying to answer, such as identifying the cost of care for a given condition in the Eastern region of the U.S., for example.
“We call it small data,” he said. “You have to prove the value in a small and directed way. Don’t invest in everything all at once; invest in something where you can really make a difference.”
“It’s amazing how much data are still in a disconnected EHR. Where you will see it picking up is when you have entities bringing together different sources.”
DAN RIZZO, Inovalon’s chief innovation officer