Relational Database Helps Analyze Wisconsin Medicaid

Like other states, Wisconsin has for years just paid its Medicaid bills - with almost no questions asked. About $2.3 billion moves through the state's health insurance claims pipeline each year. But that is about to change. The state is implementing a five-module, $5.8 million information system that will for the first time allow state officials to analyze data flowing through the Medicaid pipeline, question expenditures and spot fraud.

"Historically, we focused on pure transaction processing, so the goal was cost-efficient processing of health care claims. Data was just a waste by-product. Very little analytical processing was done," said Kevin B. Piper, director of the Wisconsin Bureau of Health Care Financing in Madison.

Although other states are implementing managed-care systems to curb escalating Medicaid costs, Wisconsin is taking a more aggressive and comprehensive approach, said Milford Sprecher, an analyst who tracks the state and local government market at Federal Sources Inc. in McLean, Va.

"Many states are just transaction houses [for Medicaid] because of resource constraints, but Wisconsin has really stepped up to the plate on this," Sprecher said. The federal government is paying 90 percent of the cost of the new Wisconsin systems.

Sprecher predicted that other states will follow Wisconsin's lead in the next few years.

Wisconsin's prime contract is with Unisys Corp., but the system will run primarily on a Hewlett-Packard Co. HP 9000 system and Oracle Corp.'s Oracle Version 7.1 relational database management system. Most of the system will be implemented early next year.

Currently, analytical data is fragmented, incompatible and incomplete, but the new system promises a single and complete database in one format, Piper emphasized. He said the database will start at 350G bytes and will grow to 1T byte.

Layered on top of the Oracle database will be end-user query and analysis tools. An executive information system will give managers aggregate data, while a decision-support system for "data junkies" will let state employees drill down to individual claims and doctors for detailed investigation, Piper said.

Vendors of the analysis tools include Kenan Systems Corp. in Cambridge, Mass., and GMIS Inc. in Malvern, Pa. Analysts use Compaq Computer Corp. PCs running Windows.

Another module will review the payments for prescription drugs and the final module - to be implemented in 1996 - will be an expert system for spotting fraud and abuse.

In general, the key benefit of the system will be to reverse the situation in which "analysts spend 90 percent of their time trying to get information out of the mainframe and only 10 percent using it," Piper said.

Now it takes "a tremendous amount of programming" to extract data reports from the mainframe system, and analysts may have to wait two or three months to get an answer, he added.

"We're sitting on top of a $2.3 billion gorilla - the most complicated social welfare program ever created - and the politics of the situation require that we find answers in minutes, not weeks," Piper explained.

Developing For The Future

The most challenging part of the project is making sure the system is adaptable because the Medicaid program is constantly changing, he said. Therefore, it is important that analysts can view a wide variety of data elements in now-unforeseen ways. "We have to be careful not to lock out any data elements," Piper said.

The second challenge is to make sure the diverse group of state employees who can benefit from the new system - anyone from auditors and clinicians to contracting officers - will fully exploit its powers. "This is definitely not an incremental improvement; this is a quantum jump," Piper said.

Reprinted with permission from Computerworld magazine.

Components

MADISON, Wis. - Wisconsin's Medicaid Evaluation and Decision Support project has the following elements:

+ An executive information system and a decision-support system that will allow staffers to analyze all aspects of the Medicaid program, including expenditures, utilization, providers, recipients, quality of care, claims processing and operational statistics.

+ A central online relational database where virtually all Medicaid data will be stored.

+ A drug utilization review system to monitor the prescription drug use of 500,000 recipients.

+ An expert system to detect fraud and abuse.