By Neil Munro
If time is money, data is gobs of money.
Just ask the federal Health Care Financing Administration or the Texas state government, which are trawling through multibillion-dollar medical programs and millions of patient reports in search of fraudulent schemes, doctors on the lam, promising medical procedures and estimates of federal health care spending.
Given this outlook, it's only natural that dollar signs are dancing before vendors' eyes, especially those selling databases tailored for PC networks.
Incumbent vendors of mainframe databases such as IBM Corp., Armonk, N.Y., or Computer Corp. of America, Framingham, Mass., find themselves besieged by Oracle Corp., Redwood Shores, Calif.; Sybase Inc., Emeryville, Calif.; and Informix Software Inc., Menlo Park, Calif.
And companies in the trillion-dollar health care industry are racing one another to buy these nifty databases, industry officials say. Pressured by competition, companies that provide health care services "are trying to get their information closer to the caregivers and the people who make the business decisions," said Carol Franduto, Sybase's health care business manager.
Government pressure to cut the cost of federal Medicare and Medicaid programs is also driving health care providers to improve their efficiency with greater use of information technology, said Michael Smith, who manages the health care business for Informix. "We're seeing more and more interest from the government," Smith said.
Let's start at the top, with the HCFA Customer Information System and its database of 3.5 billion Medicare transactions, titled the National Claims Historical Repository. It takes roughly 1,000 characters to describe each of the medical bills dating back to 1991, giving HCFA's data center in Baltimore the task of storing 3.5 terabytes of data, representing $700 billion worth of government spending on health care for retired folks.
That database is expected to grow by 800 million records - representing $194 billion in payments - once the 1996 Medicare bills are fully included, said Joe Broseker, HCFA's director of information systems division and guardian of the database.
| Program Benefit Payments|
|Fiscal Year ||Medicare ||Medicaid |
|1980 ||$33.9 ||$24.0 |
|1985 ||69.5 ||39.3 |
|1990 ||107.2 ||68.7 |
|1991 ||113.9 ||90.5 |
|1992 ||129.2 ||115.9 |
|1993 ||142.9 ||125.8 |
|1994 ||159.3 ||137.6 |
|1995 ||176.9 ||149.8 |
|Current Budget ||Medicare ||Medicaid |
|1996 ||193.9 ||160.1 |
|1997 ||213.1 ||173.7 |
|Source: Health Care Financing Administration |
Of the 3.5 terabytes, roughly 2.5 terabytes are stored online, organized by Oracle 7 database software developed by Oracle. The older data is stored "near-line" on automated tape- and cartridge-handling devices, said Broseker. The older data is organized by Model 204 database software owned by Computer Corp. of America. The software was selected before managers grew confident with Oracle's ability to handle mainframe-sized databases, he said.
Broseker's data center has a herd of IBM ES-9000 series of mainframes capable of processing 200 million instructions per second and feeding data to 100 simultaneous online users nationwide.
Broseker's volcano of data is used to look for signs of fraud, evidence of medical treatments and to gauge the fees to be paid by Medicare managers to the commercial health care companies that are taking more Medicare patients, he said.
And the flow of data will only increase - Medicare costs are expected to grow rapidly during the next several years, breaking the $200 billion-per-year barrier this year.