All Hype?

Data layers

"Gartner was using the term business intelligence 17 years ago. It's hot now because it takes time for these things to mature," says Kurt Schlegel, a consultant with Stamford, Conn.-based Gartner Inc. Operational reporting for a particular business process traditionally went directly against the data source of the application to do reporting but, according to Schlegel, that has become too limiting. "Insightful reports need to tap into the secondary uses of data."

Einbinder agrees. "If you're looking for a system, be it clinical or otherwise, you should be thinking about using the data for secondary purposes. How are you going to get the data out? What data needs to be captured in a reliable, high-quality way? If you don't think about that up front, reporting is often an afterthought, and you're up the creek without a paddle," he says.

The profit motive drove many of the enterprise resource planning (ERP) models that have been in use for years. The difference between ERP and BI is that the ERP models traditionally are more core business oriented, including materials management, payroll, general ledger, billing and accounts payable. Adding clinical information in a way that allows for interoperability in reporting is how most view BI today.

But isn't it possible just to run reports off an application? "I feel very strongly you don't report off the applications," Einbinder says.

Schlegel agrees. "You have more due diligence that the data is correct when using a data warehouse than when going directly from an application. To build reports that are more insightful they need to be based upon multiple domains and that's where the data warehouse is born," he says.

Looking back

The data warehouse is a retrospective way to combine data across multiple sources — it's a way to aggregate and combine information in a manner that makes sense. Partners reuses information captured in the clinical information systems for secondary use: reporting quality improvement, research, clinical understanding and population management. "Our clinical data has strategic value — the treatment, payment and operations," says Einbinder. "A lot of data is captured due to all those processes. One way to think of it is focusing on populations instead of the individual."

Most say the retrospective view of a data warehouse is sufficient to make informed business decisions. "For looking at quality or trying to understand practice patterns, having real time data is not that important — and the closer to real time you get, the more expensive it gets," he says. "There's usually a sweet spot where you try and optimize for your business purposes. Are people doing real time data warehousing?" Einbinder says. "The short answer in healthcare is, no."

But with so many sources of data, from clinical and business information systems, is a single data warehouse necessary or can a good BI system run off multiple domain-specific data warehouses? And can IT systems like Lawson and Oracle provide enough interoperability to achieve BI? For the time being, anyway, they may have to. "A lot of your mid-sized hospitals don't even have data warehouses," says Schlegel. "And technically it is possible to integrate data." That's especially true with the siloed systems most hospitals have today.

A data warehouse facilitates BI at Brigham and Women's Hospital, part of Partners, for what executives call a "balanced scorecard initiative." With over 70 feeds, it's viewed as one stop shopping for management and users. "What we've done successfully here is a partnership with information systems and the clinical decision support group," says Sue Schade, CIO of Brigham. Schade says there are four quadrants to the scorecard — financial, external to customers, internal to employees and lastly, operations, quality and efficiency. "You've really got to look at your performance in a balanced way and use the key metrics within all those quadrants." For the balanced scorecard, Brigham defined key metrics on an institutional level and drilled down to the departments. "You're always comparing apples to apples," Schade says. "And if you have questions, every indicator has an owner." Brigham is using a SAS (Cary, N.C.) data warehouse sitting under their balanced scorecard with 70 data feeds.

Warehouse troubles

The biggest problem with data warehouses is ... the data. A hospital can only warehouse the data that it has — and most don't have clinical data in a standardized coded way that can be aggregated. "The source data is not quite there," says Einbinder. "For that reason the focus has been more on administrative data or billing. And frankly, that's where the money's been." The Partners approach is multiple data warehouses — though not necessarily by design. "We have data warehouses for the hospitals and several at the enterprise level for claims data, EMR data, and billing data, to name a few."

But if a hospital lacks the capital or technology for multiple data warehouses — or even for one data warehouse — what are some options? Schlegel says hospitals are already finding ways. "In lieu of a big data warehouse, you're going to grab data from different sources and put them in Excel spreadsheets. That is replete in the industry," Schlegel says. "We've come to the realization that we can't rip spreadsheets out of people's hands."

Schlegel says one option is making spreadsheeting more centralized, where users know where the reports are coming from, how they're being defined, and scheduling them to run. "Ironically," he says, "Microsoft has been behind in capturing the BI platform — except in Office 2007. That's where it catches up."

So is the 2007 Microsoft Excel a "poor man's version" of BI? Chris Sullivan, national director, industry solutions for Redmond, Wash.-based Microsoft Healthcare and Life Sciences says it's definitely filling a need. "One of the things that is significantly different in Office 2007 is it's now really embracing the net platform and looking at it as an extension of that back end," he says. "It provides a great opportunity for hospital organizations from a reporting perspective because in most in healthcare enterprises, a lot of the reports in the business intelligence analysis are done at the end-user level."

Sullivan adds that too often hospitals run into a bottleneck when it comes to analyzing data: they need to ask IT to run reports, which are usually delivered too late to do be useful. "A good example is a chief nursing officer asking, 'How many agency nurses do we need?’” he says. "The answer to that is based on their census data — and typically that data is at least two to three days too late to be of real use." Sullivan says the 2007 version of Excel gives functionality and tools for real time reporting. "Leveraging sequel analysis services, they can pull that data out, extract it, put it in a graphical format and run reports that allow them to get a better handle on their operations."

Since BI can utilize anything from Microsoft Excel to a single enterprise data warehouse, which technologies will drive this change? According to Einbinder, the challenges aren't technical. "I think the technology is there, but we're barely tapping it," he says. "The challenge has been the data." Einbinder says true BI needs high quality, coded data. "Look at the EMR adoption — if only 30 percent of places have EMRs and they're not using it to generate data, you're not going to realize the full potential. The analytic curve is going to lag the adoption curve for the clinical system."

In the end, many feel it may well be business drivers that impel BI adoption — things like pay for performance. In P4P, the business drivers require analytics and business problems demand data. There are also many regulatory drivers and a relentless set of efficiency drivers. "The trend today is the shrinking of the ecosystem," Mowry says. "Everyone within the healthcare ecosystem has to get closer together. That means providers, physicians in their offices, clinics, payer organizations and the consumer."

Mowry agrees that P4P will propel BI forward. "You'll need to be able to identify quality because the federal government is going to be paying for quality. And you're going to have to be able to prove your outcome. In that ecosystem, the need for quality data is more significant today than it's ever been. And it will only continue to grow."

Sidebar

For More Information

Online Resource Center: http://www.healthcare-informatics/Finance Search Our News Archives by Organization Type: Acute Search our Web Site by keywords: business intelligence, ERP