Would Oz Use Data Warehousing?

P> If doctors disliked people looking over their shoulders in the 1980s, then they'll hate the late 1990s because now it's not a colleague with a second opinion but someone with the data warehouse doing the second guessing.

"Our main business today is handling medical data," said Pete Wagner, director of Network Communications for HCIA. "We still do financial analysis, but now it's just a very small part of what the company does. We made a jump to the clinical side, and the company now focuses on data related to the cost and quality of care. We sell data and analyses of it."

And HCIA has a good share of this nascent medical data warehousing market. "It would be very difficult for anybody to break into this market and compete with HCIA because there's a tremendous hurdle to making the arrangement to acquire the data," said Wagner. "We have long-term contracts with hospitals, which commit the hospitals to sending the data to HCIA. A typical pattern is as follows: 'You're getting a bunch of money from Medicare and Medicaid and so forth, and you need to process the data for these programs.' We are the vendor who keeps the data for them."

The main buyers of medical data are insurance companies, health-care providers and health-care suppliers. Each wants to know what's the "norm" for an array of treatments and procedures; each has a different interest in what to do with the same data. Not surprisingly, each generally has a slightly different "norm."

The big boon, however, seems to fall into the arms of the appreciative medical insurance industry. And it's not just about money, it's also about the power of doctors to implement procedures and prescribe medicine.

HCIA's Wagner explains: "One immediate benefit [of data warehousing] is that hospitals have an immediate tool to compare their procedures to other hospitals against a benchmark we may be able to develop. We might be able to say the length of stay for an appendectomy is X number of days, and the hospital might want to ask, 'What's our length of stay, and how does it compare?' If they find a significant negative variance, they can drill down into the data to determine whether they're using procedures or drugs or whatever that might not be appropriate or the best. In a sense, you can think of HCIA as decision support for health-care providers."

This encroachment raises the dander of doctors who resent anyone -- especially someone whose glasses glow from the light of a computer screen -- telling them how to practice medicine.

"[The American Medical Association] is concerned about the practice of medicine and who's making decisions about the practice of medicine," said James Stacey, spokesman for the AMA. Speaking from the AMA's Washington, D.C., headquarters, Stacey added firmly, "The decision about what should be done and should not be done for patients should be made on a professional basis."

Stacey said the AMA itself has for years compiled medical data in an effort to promote collegiality and share information about improved medical procedures. "Generally, there's nothing new in this whole idea of capturing information and using it appropriately," said Stacey. "The AMA itself has been working on a program called Practice Parameters for several years now. This is an attempt to gain the best medical information on parameters and practice so that physicians have a benchmark on what the appropriate protocol is for a particular procedure. The issue is the appropriateness of care. Generally, the AMA feels that this is a useful guide, but that all patients are not alike. Someone may not heal as quickly as someone else. It depends on the patients. To lay down an absolute about any given procedure is to miss the point of the very unique person that every individual is."

But the bottom line for the mass insurer clients of HCIA is lower insurance costs across the board. "Most of our people are analysts, and invariably they give our [medical insurance] clients lower costs and higher quality of care," said Wagner, who eagerly left the issue of cost shrinkage to get back to the "gee whiz" of technological know-how.

"We have a terabyte -- that's a thousand gigabytes of data -- going back four years spread across four Sun servers, the largest of which is a SparcCenter 2000 with over 600 gigabytes of data, 2 gigabytes of RAM, and 32-bit processing," said Wagner. "The most common tool for analysis is Informix4GL as well as InformixSQL running under UNIX. We're currently migrating to Informix 7.1 from version 5, and we expect significant performance improvement when we're all done because Informix 7.1 has significantly greater efficiencies in a multiprocessor environment." Wagner said he expects the system to glean as much as two terabytes by the end of this year.

"This is not the largest data warehouse, but it is the largest private medical data warehouse," continued Wagner. "You can safely call it that. The government may well have a bigger one."

Whether the government does or not, the day has come where the doctor in your hospital room is increasingly beholden to some guy at a computer terminal "helping" him minister to your health.