In implementing BI, hospital officials face KM challenges

For a variety of reasons, healthcare has traditionally lagged behind other sectors in adoption of business intelligence (BI) tools. Other vertical markets, such as banking and manufacturing, have been taking advantage of BI for 20 years.

And the progress health organizations have made has largely been on the financial side of the house. The complexity of medicine and the lack of common standards, practices and clinical vocabularies across similar service lines make getting clean clinical data very difficult.

“Many healthcare organizations are having trouble finding the time and resources to implement electronic health records (EHRs), much less do more sophisticated analyses on the data that comes out of them,” says Mary Sirois, principal in the clinical transformation practice of healthcare consulting firm Divurgent. “You would be missing a lot of clinical decision support potential, if the providers aren’t entering orders electronically.”

KLAS Research, which focuses on health information technology, estimates that fewer than 100 U.S. hospital systems have enterprise data warehouses and sophisticated reporting tools that draw data from both financial and clinical systems. Yet a 2010 KLAS survey of health IT executives highlights how quickly things may be changing. The federal financial incentives being offered to organizations that are “meaningful users” of EHRs require an increasingly sophisticated level of data gathering and reporting. Sixty-nine percent of respondents told KLAS that BI solutions would play an important or critical role in their meaningful use reporting. One provider told the firm: “BI will play a critical role. Never before have we been so carefully scrutinized, and the BI system helps us provide the information and reports needed to be successful.”

KMWorld recently spoke with three hospital executives charged with leading clinical data warehouse/BI initiatives about the knowledge management challenges they are facing and some of the benefits they are already seeing.

Improved analytics in Georgia

Four years ago, Athens Regional Medical Center in Georgia began the journey of adopting a new EHR system, and it now has almost 90 percent of physician orders, such as medication orders, automated. Most of its clinical data is in that system.

As the industry is being driven more and more by quality reporting, pay for performance and analytics, Athens saw a need to extract key data from diverse data sources, says Jim Pirkle, associate director of clinical analytics. “Although our transactional systems are all great at what they do, they aren’t as good at sharing data with the other systems,” he explains. “We needed to view things across multiple dimensions.”

Athens Regional uses multiple BI tools including Tableau Software and Diver Solution from Dimensional Insight. Those two products are capable of extracting data from diverse data systems.

“Depending on the nature of the question we want to answer, we will query one of those two,” Pirkle says. “For instance, we would use Tableau for things like physician practice evaluations and Dimensional Insights for hospital metrics.”

Previously, clinical quality staff would generate Excel files around hospital goals and then post them all over campus. “Now we can generate that electronically, and our users have access to the information at their fingertips,” Pirkle says. “The biggest change has been that if the chief medical information officer asks how many procedures Dr. Smith performed last year, I can tell him the answer in 30 seconds. That used to take several days.”

Pirkle used to work in manufacturing and said that many years ago you could use similar tools to determine exactly how many of your products had been sold at Walmart the day before. “That hasn’t been the case in healthcare, but we are starting to see the power of having access to that level of information,” he adds.

But now that the hospital has the tools to mine the data, the challenge is helping the internal customers understand the subtle nuances behind their questions. For instance, the answer about how many procedures Dr. Smith performed depends on several variables including whether he was primary or secondary surgeon on a case. “We are trying to do an educational campaign with users to help them better define what they want from the data,” Pirkle says. “Otherwise we may provide something that’s accurate but doesn’t deliver the information they want.”

One challenge Pirkle and colleagues face is that any changes to the source systems can impact how data flows into the warehouse. “What seems to be a simple change can have tentacles throughout the reporting structure,” he says. Blending data from different systems can be tricky because something as simple as reporting periods might be on a fiscal year in one system and a calendar year in another. “These BI tools we use offer the flexibility to allow us to define parameters to deal with incongruencies between systems,” he says.

Pirkle is enthusiastic about the hospital’s growing analytical capabilities. “We are able to analyze data that we didn’t even have access to last year. It is helping us define evidence-based medicine by looking at patterns of data,” he explains. If one physician did 80 percent of procedures one way but others in his group did only 40 percent that way, this allows them to look at those disparities and talk about it, he says.

Decision support evolution at HealthEast

HealthEast, a three-hospital system based in St. Paul, Minn., has had a financial warehouse for 10 years and uses McKesson’s Horizon Performance Manager and Horizon Business Insight tools to analyze the data. More recently, the clinical informatics team has been working on adding clinical data from within hospital systems.

Many of those source systems have not been approached before as sources to feed a data warehouse. “We are just opening discussions about whether we can create data streams from many of them,” Skip Valusek, director of clinical informatics, explains. “In finance, there are maybe two systems as sources. On the clinical side, I can’t even begin to count all the islands of data. I call it the archipelago.”

Working under the chief medical information officer, Valusek’s team includes clinical report writers, people working on extracting data from source applications and a clinical outcomes analyst working with users on what should be in the scorecards and dashboards created in Horizon Business Insight.

One example of a new feature is glucose level management. “If we do a better job of watching glucose levels more tightly, we can have better patient outcomes and shorter hospital stays,” Valusek says. “So our analytical tools can give us better operational awareness of how we are doing on a daily basis, and then we aggregate that data into weekly and monthly reports.”

Another upcoming challenge will be to merge data from clinics and physician offices with data from the hospitals. “We use McKesson’s EHR in the hospital, but Allscripts (allscripts.com) in clinics,” Valusek says. “With health reform, we will have to look at it from a system enterprise perspective.”

Asked to offer advice to people just starting down this path, Valusek emphasizes the importance of a committed executive sponsor. “My boss, the CMIO, stresses the central role this plays to our overall strategy. It needs to be a bigger part of our strategic plan, our whole program. It can’t be an ad hoc project,” he says.

When he arrived at HealthEast three and a half years ago, he stressed the need to build a strong team of clinical data analysts. “You might have people who are good at cobbling together data from different sources, but you need people who are good at analyzing it,” he says. “I can create an engine to pull data from all these sources, but who’s going to analyze it?”