Lowering the barriers to information exchange

CHICAGO—Information exchange is increasing partly due to lowering of barriers such as governance and trust, said J. Marc Overhage, MD, PhD, speaking at the Healthcare Leadership Forum on Sept. 30. Currently CMIO of Siemens Healthcare, Overhage previously was founding CEO of the Indiana HIE.

In the 1980s, he said, the emphasis of information exchange was on point-to-point transmissions and how “we get there has continued to evolve.” It’s been challenging at the community level because of barriers including fear and anxiety about protecting both patient’s privacy as well as providers’ economic well-being.

“Direct was conceived as a simpler, easier way to get healthcare information to move and, frankly, it’s gone almost nowhere.” Today, we have almost a federated identity approach to allow users to know who the information is sent to but numerous challenges persist including how to identify patients and providers, how to engender trust and how to create business models that support information exchange.

That fact is, he said, “if data are fragmented across many providers, putting them back together is quite a daunting challenge.”

Indiana took a simple approach with its HIE, Overhage said, first dealing with matching patient identity within organizations and between organizations. Algorithmic methods for patient matching are very reliable with high specificity and high sensitivity. Once that’s been done, providers can know where their patients are.

There’s been moderate progress on this, he said, in terms of standardization of terminology and content. But, there are “a whole host of nursing interventions that are poorly normalized.” To manage that, they did a mapping process between standardized codes and local codes. “We did that once for each organization so that it didn’t have to be repeated by everybody. As we did that over and over, we created a series of silos.” Those silos are separate and segregated which was critically important, he said, “because it allowed trust and control of data to evolve.” Data were not mingled until they were used.

Indiana is getting close to true population-level coverage, he said. “That’s interesting for a lot of reasons. If you look at quality, processes and transitions of care, the whole population becomes critical to understand as a policymaker, a payer who wants to understand what the future looks like and a researcher who wants to understand disease and its progression.”

Once Indiana got a good set of population data, they weren’t sure what to do with it. They created DOCS4DOCS, Overhage said, which “created value by getting information to physicians.” Back in 2006, when this effort began, it cost the typical lab 84 cents to get lab results into a physician’s hands. DOCS4DOCS got the cost down to just 25 cents. The vehicle “made it critical to deliver results in a timely, understandable, interpretable way which drove accumulated data that helped us understand the population."

The Indiana HIE helped hospitals learn, through matching physician identifiers with lab results, that 40 percent of those results were going to someone not on their medical staff. “That came as no surprise to us. We saw these kinds of things over and over.” He noted the “long tail” of the internet world. “The curse of healthcare data is that no matter how you look at it, the data are spread all over the place.”

As hospital labs started to send their data, they didn’t have physician identifiers in the lab system and decided to add that information in from staff management software. However, they learned that 40 percent of lab results were going to someone not on their medical staff. Rather, that chunk was going to another provider down the street to whom the patients went. These data have a long tail, Overhage said. “This is the curse of healthcare data. No matter how you look at it, data are spread all over the place.”

They also learned that many patients have more than 100 emergency department visits a year. By taking advantage of these data flows, they did a six-month trial of nine central Indiana hospitals to determine potential ED utilization. They were then able to reduce non-urgent ED visits 53 percent while increasing primary care office visits by 68 percent--“to what we hope was a more appropriate level of primary care.” That came to $2 to $4 million in savings just during the pilot period.

“The sphere of influence tends to be around the care delivery system,” said Overhage. “When you really start looking at patients and where their data are, they are much more dispersed than you might think. When you have all the data together you can learn great things.”

But, there is still work to do to get there, Overhage noted. “Data are a critical part of understanding the human process and patient dynamics. It’s a fun and challenging area. If your inside house is in order you’re in much better shape to share data.”

Participants have to give data to get data, he said. Mutuality makes exchange work. “It really is a shared resource.” That’s why providers should have a clear understanding of the terminology used so they can exchange information in a meaningful way. They also should not just monitor data but monitor the quality of the data collected. “Most of the time we think we’re collecting good data and we’re not. Nobody notices because we’re very good at adapting and covering up those things.”

Patient-generated health data is going to be “critically important over the long term,” Overhage added. “We have a lot to learn about how we think of that data—who’s going to look at that and how and what thresholds and trends should trigger action.”