Reaching for Help

Plenty of vendors stand ready to help health-care providers mine their data. Here are some pros and cons of relying on outside assistance.

Companies balking at the money and skills needed to develop advanced data mining tools have the option of paying someone else to build the tools—or even to mine the data—for them. Outsourcing all or part of the data mining process can be less expensive and time-consuming than handling it in-house. Such vendors as Mitchell Field, N.Y.-based CrossZ Software, HealthShare Technology Inc. in Acton, Mass., Philadelphia-based Care Management Science Corp. (CMS) and The Medstat Group in Ann Arbor, Mich., offer ready-made databases of publicly available medical information that hospitals can mine at will. Often, those outsourcers also will integrate a provider's internal data with their own databases. And the fact that outsourcing vendors have other customers can be an added benefit: Organizations that farm out their data mining can foster working relationships with other customers and compare their data.

One hospital that went the outsourcing route is Birmingham, Ala.-based Brookwood Medical Center. Brookwood built its own data warehouse from a variety of different internal sources in 1992, but decided to outsource the data mining component to CMS and other vendors rather than develop it internally, says Sue Esleck, administrative leader for care continuum. "I didn't feel like we could satisfy our physicians by trying to grow that in-house," says Esleck. "They were far too sophisticated and required real, proven data." Brookwood sends semiannual data feeds to CMS and receives a summary report of the data mining results. Esleck also has developed a phone relationship with a fellow CMS customer in Pittsburgh. When questions arise about outcomes measurement, Esleck simply gets on the phone for a collaborative tete-Æ-tete.

But there are dangers inherent in settling for a quick data mining fix. The perceived cost savings may be misleading; companies can end up spending as much as $1 million a month, says Mark Brown, program manager for data mining and applied analysis at the Cary, N.C.-based decision-support software vendor SAS Institute Inc. And companies that outsource everything lose control of the process. They often miss the real story behind the data upon which they base their big business decisions. "If you don't understand how it all takes place, you're making a mistake because you might need to explain it to the CEO some day," says Robert S. Brown Jr., a doctor at the University of North Carolina at Chapel Hill School of Medicine.

"You can't outsource outcomes measurement totally because you've got to understand what you see," adds Esleck. "Somebody in-house has to be able to take the reported information and apply it to determine what needs to be changed. Otherwise it's just data."

- J. Bresnahan

Making the Grade

As if there weren't enough acronyms floating around, now there's HEDIS, the health-care version of ISO 9000. Set by the National Committee for Quality Assurance to help employers choose between HMOs, the Health Plan Employer Data and Information Set measures such factors as what percentage of young children in an HMO's network are fully immunized or how many men have had cholesterol screenings. Norwalk, Conn.-based managed care provider Oxford Health Plans relies on outcomes measurement to verify that its physicians are adequately performing basic services. If a particular pediatrician has a month of poor outcomes, for example, the medical analysis team inspects whether the doctor is giving preventive immunizations. If he is not, a team member contacts the doctor, and even sends letters to the patients' parents urging them to make appointments, says Scott D. Stratton, director of Oxford's medical analysis group.

Avoiding Merger Bloat

When it comes to beating the competition, bigger is often better. But companies that acquire new businesses without streamlining their combined resources can fall prey to that state of inefficiency and waste known as "merger bloat." Indianapolis-based Anthem Inc. was at risk for the dreaded disease after going through three mergers in three years with companies worth $1.5 billion to $2 billion each. But mining the combined data in its 1.3 terabyte NCR Teradata database helped the $6 billion health insurance company target areas to trim—such as coronary artery bypass, a common but expensive and risky procedure. Eighteen different hospitals within a region of Ohio were performing bypasses but the variation in outcomes among them was striking. Anthem chose five hospitals as centers of excellence and shifted all of the bypass patients in the region over to them. As a result, costs fell dramatically and the mortality rate was reduced from more than 4 percent to less than 1 percent, says Anthem's senior vice president, William R. Milnes Jr.

Future Perfect

Everybody knows that running out of the one item clients really want is a major customer-service no-no. In the health-care industry, however, running out of a critical medicine at the wrong time can mean the difference between sickness and health. That's why Melville, N.Y.-based home health organization Olsten Health Services mines its data to predict what products and services it will need in the future based on utilization patterns from the past. Knowing, say, that hemoglobin levels of diabetics are higher in August than December tells Olsten to stock an adequate supply of drugs and personnel in the summer.

Related:

Copyright © 2007 IDG Communications, Inc.

Download CIO's Roadmap Report: Data and analytics at scale