Bullet-Proof Warehouses

STATVIEW

Organizational transformation, complex reporting requirements and growing respect for evidence-based medicine are driving data warehousing implementations. But the current dearth of healthcare-specific applications and turnkey systems still makes the data warehouse a custom project.

Blistering budget cuts in the military four years ago hit the medical services units particularly hard. Slated to lose about a quarter of their resources, managers searched for ways to live with less, while still caring for 8.5 million beneficiaries. Medical services managers realized they had to improve business processes and embrace managed care. The question was how.

The ultimate answer consisted of a titanic, multi-phase effort to integrate the collection of IT systems across the Military Health System headquartered at the Pentagon in Arlington, Va. The key integration tools: data warehouses. Four years and nearly millions of dollars later, a collection of local and regional data warehouses pump financial stats and patient profiles into administrative and clinical systems.

An enterprise data warehouse, now in final development, should become operational in the first quarter of 1999. Although project directors hope for total buy-in by clinicians and administrators, they realize they still have a long way to go. Caregivers usually are skeptical when they first encounter a computer system that changes how they do their daily jobs. "They go through a grieving process that includes denial and shock," says Michael Mauro, system architect for the integration project and its enterprise data warehouse (for more details about this project, see Case in Point "Warehouses for the New World Order" on page 50). But shock often turns to delight, Mauro adds. "Once they see the light, they become ravenous for data."

Data power
The axiom that information is power describes the driving force behind data warehouses--central repositories of enterprisewide information that with the right analysis tools can help organizations spot operational ills and uncover market trends. The military isn’t the only sector hungry for accurate internal data.

Civilian healthcare is looking to data warehouses to help attack inefficiencies. Insurers led the charge in the private sector, wading in as early as 1991, when few turnkey data warehousing systems or knowledgeable consultants existed. But the custom data warehouse insurance companies did build helped them identify fraudulent claims and increase operational efficiency and risk prediction.

Today, as warehouse systems and data-analysis tools mature, healthcare organizations are finding that financial operations aren’t the only beneficiaries of a central data storehouse. Clinicians now use warehouses for evidence-based medicine--a new discipline that taps historical data to validate diagnoses and treatment regimes. "If data is properly captured in one big database, the data is worth its weight in the most precious metal for commercial medicine and for research," Mauro says.

Pharmaceutical companies, progressive integrated delivery systems and health plans build warehouses for extracting valuable clinical and cost management statistics. Other companies dedicate warehouses to chronic diseases, such as multiple sclerosis, diabetes, chronic wounds and asthma. In the future, data warehouses in healthcare may underpin the entire delivery system, from the patient’s bedside care to organizational management, says Marshall Ruffin, MD, president of The Informatics Institute, Bethesda, Md.

Table 1 Data warehouses and data mart uses in healthcare

Risk management

Clinical

Public health

Building blocks
Whether used for clinical or financial analysis, the basic components of a data warehouse are the same. At a warehouse’s heart is a database that’s relational or multidimensional (i.e., image, video and audio as well as alphanumeric information). The database usually runs on a dedicated server and mirrors data residing on other enterprise databases used for day-to-day operations.

Augmenting the data warehouse are software tools that extract, cleanse, transform and move selected data elements from the operational systems. Query and reporting applications, known as decision support systems (DSS), let users search the database for specific nuggets of information or compare data to uncover hidden trends, such as finding departments that habitually go over budget or recognizing that a significant percentage of the patient population is older than 50. With its data stored in standardized formats that make comparisons easy, the warehouse provides a "single version of the truth" or a single representation of the business entity.

At their best, data warehouses can help healthcare organizations improve service and lower costs, says Wayne Eckerson, vice president of technology services, The Data Warehouse Institute, Gaithersburg, Md.

But before you can enjoy these benefits, you have to pony up big bucks. Data warehouses commonly cost a million dollars at large organizations. However, there are notable exceptions. Using the Web and in-house expertise, The School of Medicine at the University of Virginia, Charlottesville, created a warehouse for less than $500,000. (See Case in Point, "Success on a Shoestring" below).

Quickie marts
Other organizations save money by resisting a behemoth, organizationwide data warehouse in favor of small, tightly focused data centers called data marts. If the National Zoo were a data warehouse, its aviary would be a data mart. Similar in architecture, data marts are subsets of data warehouses--not miniaturized data warehouse models. Data marts usually address a single, specific business function or process or a department--clinical data only, for example--as opposed to the enterprise view taken by the data warehouse.

Data marts are easier, quicker and cheaper to build than their big brothers. Devote three to six months of development time, and you can get a data mart up and running compared to spending a year or more for a data warehouse. Data marts can be powerful confidence boosters, too. They allow the organization to go through the planning and implementation processes and get its feet wet before embarking on the larger warehouse project.

But starting with a data mart and expanding into a data warehouse is neither easy nor economical. Experts say mart-to-warehouse morphing works only if you plan from the beginning for the mart to be a subset of a data warehouse. "Doing a one-off data mart means painting yourself into a corner," says John Ladley, senior program director of Meta Group, Stamford, Conn. "Given the integration of the data and how many areas share the data, it is absolutely essential that healthcare organizations take a holistic approach," he adds.

Keys to success
Given the expense and complexity of data warehouses and marts, how can you increase your chances of a successful implementation? Experts cite six main areas to focus on.

First, don’t take planning shortcuts. Too often, organizations rush head-long into the project, bypassing the up-front planning cycle, says Jerrel Little, director of worldwide healthcare marketing for database vendor Informix Corp., Menlo Park, Calif.

Second, match your warehouse architecture with your organization’s culture. A warehouse is definitely not a one-size-fits-all entity, says Mary Hope, consultant for the London-based IT analysis firm, Ovum Group. For example, some organizations benefit from having a top-down structure using a centralized, enterprisewide warehouse with data marts hanging off of it (see figure 1, page 48). This can be an appropriate choice for a hierarchical organization. But Hope warns that shoehorning a centralized warehouse into a decentralized organizational structure with local data ownership will be as effective as playing football in ballet slippers.

Instead, Hope favors a federated data warehouse structure with independent, department-based data marts (see figure 2, page 48). Better suited to a more decentralized environment, a federated structure supports a higher degree of local ownership, but has a global overview. Managed with a metadata catalog--which she likens to a cloud sitting on top of the entire architecture--the federated warehouse enables the end user to see what data is available and to dip into any of the pots of data for which they have authorization and permission.

Third, be careful who you listen to. Once you commit to a data warehouse, you’ll find a surfeit of willing advisors. Most vendors associated with data warehousing have a consulting or outsourcing arm. But analysts caution that while vendors understand their particular piece of the puzzle, few can provide the wide range of products and services necessary to build a data warehouse to your satisfaction. Accordingly, some organizations hire systems integrators to guide the warehouse launch.

After you’ve diligently interviewed potential consultants and run background checks, make sure you structure your contracts with regular milestones that can help you access your project’s progress. Peg three months out as the first critical target, says Jim Davis, program manager of data warehousing for SAS Institute, Cary, NC. "Continued incremental successes at timed intervals are critical to the project--to gain funding and management approval--and the momentum to move into the next phase."

Fourth, training is essential. In the past, power users such as financial analysts and strategic planners were the denizens of data warehouses. But today, department managers, physicians, nurses and medical researchers are logging on to warehouses. As a result, data analysis tools must now support a wide range of skill levels, notes Mark Moorman, SAS program manager for business intelligence. This can create problems for warehouse developers and administrators. In an effort to engage all levels of users with the data make sure your project budget includes money for training. Untrained users can’t take full advantage of the warehouse.

Fifth, keep your expectations realistic. Software and database vendors talk like warehouses are plug and play products. Conversely, consultants might try to scare you with reports that 70 percent of warehouse projects end in failure. Neither view is entirely accurate. For example, a recent Meta Group study found no project that was an absolute failure or absolute success, according to Ladley.

Sixth, cleanse your data. Even the best data warehouse can’t reach its potential if it stores inaccurate or inconsistent data (i.e., sales data in one field representing fiscal-year results, while another field lists calendar-year totals). Ninety-nine times out of 100, success or failure comes down to data quality, says Ladley. He adds that healthcare organizations still must develop their own mechanisms to get the data out and push it into the warehouse.

A moving target
A properly built data warehouse can generate multiple benefits to the organization. But be prepared: Even if you achieve initial success, don’t waste time patting yourself on the back, SAS’s Davis warns. "A data-warehouse project is never complete. As the business changes, the warehouse must change."

One-stop shopping

DATABASE AND HEALTHCARE application vendors see gold in data warehouses and are rushing to deliver turnkey systems that promise to end the headaches associated with multiple component, multiple vendor solutions.

"People aren’t shopping for technology anymore; they’re shopping for applications that business people understand," says Wayne Eckerson, vice president of technology services, The Data Warehousing Institute, Gaithersburg, Md.

Unfortunately, most healthcare organizations will be disappointed if they try to find an industrial-strength, shrink-wrapped warehouse application today. "Just because it has the name ’clinical’ on the front end doesn’t mean that the product can meet the needs of the clinical organization," cautions Greg Rogers, SAS marketing analyst for health/insurance.

Integrated packages, from vendors such as Erisco and Synertech, are rich on the operations side, but fall short when it comes to making decisions based on their stored data, reports John Ladley, senior program director, Meta Group, Stamford, Conn. Some have had serious performance problems, making it impossible to get reports out. To Ladley, this is like giving a starving man a can of beans but no can opener.

On the decision side, Ladley sees old guard and traditional products such as MedStat and HBOC’s GMIS threatened by new technologies that bring do-it-yourself to complex utilization, NCQA (National Council for Quality Assurance) and HEDIS (Health Plan Employer Data and Information Set) reporting, bypassing the products’ expensive buy or lease packages. He looks for new technologies and re- architected packages in this market.

Some companies are learning the market even as they sell, packaging a healthcare application built from consulting-derived results and templates. A much more difficult step, says Eckerson, is to identify the business process and build an application that can support a complete business process in healthcare. To accomplish this, the application must do away with templates and models in favor of time-consuming coding.

Nevertheless, these types of applications may not be far off. Technology companies are building applications themselves or partnering with application providers already in the market. Alternately, healthcare application vendors are beefing up their online transactional processing (OLTP) applications with better reporting modules and decision support tools.

Wait six to nine months, and vendors such as Synertech, EDS and HBOC may have industrial-strength products or at least partnerships with other vendors that will provide total solutions.

In the meantime, warehouse builders have a number of programs to choose from to help cobble together data storehouses and the tools necessary to extract meaningful information.

Data Warehouse Vendors

Here’s a representative list of vendors with marketable products courtesy of The Data Warehouse Institute, Gaithersburg, Md.

Healthcare data warehouse-enabled applications (i.e., a dedicated database of cleansed, integrated, historical data with OLAP or reporting tools):

• Belmont Research, San Bruno, Calif., www.belmont.com
• Openair Software, Inc., Stoughton, Mass., www.openair.com
• SAS Institute, Cary, N.C., www.sas.com
• Speedware, San Ramon, Calif., www.speedware.com

Turnkey decision support systems Vendors that provide most of the tools needed to build a pre-integrated data mart include:

• Broadbase, Menlo Park, Calif., www.broadbase.com
• Enterworks, Ashburn, Va., www.enterworks.com
• IBM Global Business Intelligence Solutions, Somers, N.Y. www.ibm.com/bi
• Information Builders, New York, www.ibi.com
• Management Science Associates, Inc., Pittsburgh,
• MineShare, Santa Monica, Calif., www.mineshare.com
• Openair Software, Inc., Stoughton, Mass., www.openair.com
• Oracle Corp., Redwood Shores, Calif., www.oracle.com
• QueryObject Systems Corp., Uniondale, N.Y., www.queryobject.com
• SAS Institute, Cary, N.C., www.sas.com
• ShowCase Corp., Rochester, Minn., www.showcasecorp.com
• Silvon Software, Inc., Westmont, Ill., www.silvon.com
• Sybase, Emeryville, Calif., www.sybase.com

CASE IN POINT

Warehouses for the New World Order

When the Military Health System, Arlington, Va., undertook its sprawling healthcare integration effort four years ago, it was prepared to overhaul its culture to make itself more efficient and to survive in an environment of budgetary blood letting. What it wasn’t prepared for, however, was becoming a data warehousing pioneer.

The plan was to begin building an enterprisewide data warehouse two years into the project. But when the time came to choose warehouse technologies, project managers could find few data warehouse products designed for the staggering workload and millions of transactions of the Military Health System. HBOC’s product came closest, reports Michael Mauro, enterprise data warehouse (EDW) system architect for the Corporate Executive Information System (CEIS). But no healthcare vendor was geared for this project’s magnitude.

Nuts and bolts
CEIS administrators enlisted the help of EDS, a Dallas systems integrator, to help design and select warehouse components. Flexibility and scalability were essential. "Whatever business problems you’re facing today, they’ll be different six months from now--and they’ll change even more within 12 months," Mauro explains. "The minute you start to reengineer and change from fee-for-service to managed care, the business metrics you use today will be outdated tomorrow."

Mauro’s group settled on a warehouse that includes IBM’s RS/6000 hardware platform and Informix’s parallel server database system. To populate the system, it chose DataStage, a lexicon-based extraction/transformation tool from Ardent Software. Flexibility is also important on the front end where, among other concerns, the project leaders needed query, reporting and online analytical processing (OLAP) tools to support both thin and fat clients. They selected WebIntelligence from Business Objects. Mauro says these select commercial tools helped his staff avoid a lot of code-writing.

To determine the data elements to support the business reengineering and organizational transformation process, CEIS sifted through tens of thousands of transactional elements to derive a short list that included diagnostic ID, procedure ID and an encounter case mix index. Only 300 data elements were selected for the first version of the warehouse.

Stars and falling prices
Organizing the data for access and retrieval was another major challenge. Project leaders decided that a star schema data model, as opposed to a more traditional relational data model, was to be the only practical solution for such a large warehouse, reports Mauro. The advantage: Users can examine the enterprise from the top down without knowledge of the questions.

But this choice came with drawbacks. Namely, retrieving data can be a problem. A user may be able to determine everything there is to know about an individual or an event, but the system is not optimized for that type of retrieval--for which a properly indexed, normalized data model is better. Mauro credits lower mass storage technology costs for enabling the project to economically compensate for star schema’s shortcomings. Multiple databases, normalized and designed to complement star schema’s redundant data, now work with five star schemas built on information needs: clinical, population, claims, pharmacy and financial.

Disease managment wins big
Mauro says when users realized that they could use the data for disease management at this early stage, he was elated. He observes, "I have seen behavior modification that was not dictated; rather, it was self-engineered based on information and evidence."

CASE IN POINT

Success on a Shoestring

When the School of Medicine dropped its Clinical Research Database into the IT pool at the University of Virginia, Charlottesville, administrators didn’t imagine the ripples they would create. From humble beginnings intended as a resource database for clinical researchers, the database’s appeal soared as people throughout the university discovered its treasure trove of information.

The database outgrew its original focus--as users from the Medical Center and the School of Medicine came to rely on what’s now known as the clinical data repository (CDR).

The project began with a modest goal: give Medical Center staff a storehouse of clinical data. Build versus buy was never an issue. "We built because we didn’t have much money to buy," says Kenneth Scully, database administrator at the Information Technology and Communications-Academic Computing Health Sciences (ITC-ACHS) and the Department of Health Evaluation Sciences.

The Web’s potential attracted Scully even though Web development tools were few. In the end, he wrote a custom Web interface with HTML, Java Script, CGI programs and Perl scripts. Two-and-a-half years and about $350,000 later, he has linked all of the Medical Center’s clinical, administrative and departmental legacy databases into a single Sybase relational database that resides on a dedicated IBM RS/6000 platform.

The CDR integrated existing data sources, reflecting a strong focus on inpatient data. Ultimately, the system that feeds data into the main storehouse must be broadened to include more administrative and financial information, says Jonathan Einbinder, MD, an assistant professor in the Department of Health Evaluation Sciences and the Department of Internal Medicine.

Measuring success
Nevertheless, the CDR managers are pleased with what they see as a series of small victories. Scully cites user access to multiple data sources from their desktops as particularly satisfying. Einbinder is especially gratified by the growing respect for the CDR’s value in medical management. "The challenge for the CDR is not primarily a technical one, it’s really much more a process and an organizational issue," he says.

As others see the value of the consolidated information, more are requesting access to the data for clinical and management decision-making. The CDR is the only place where a user can browse through the data and bypass involving IT in a process that is often frustrating to both sides. If the information is up on the Web, says Scully, the user can poke around and see what’s available. The goal is to engage more people with the data, adds Einbinder.

Although the CDR has outgrown its research focus, it continues to gain value as a management and operational tool. But its future is unclear. The Medical Center is convinced of the value of a data warehouse; but whether or not it is this CDR is another question--one Einbinder considers unimportant. "The CDR is a success against many odds," he says.

Charlene Marietti is senior technology writer at Healthcare Informatics.