Oracle's Consultants at oxhp.com

Oxford Health Plans, Inc., did what Oracle's consultants advised.

Here's what happened, according to "The Wall Street Journal".

/* begin copyright violation section */

The Wall Street Journal
Thursday, December 11, 1997

HOW NEW TECHNOLOGY BECAME A SERIOUS PROBLEM FOR OXFORD by Ron Winslow and George Anders
Staff Reporters of THE WALL STREET JOURNAL

The annals of business are filled with Frankenstein stories -- -- tales of technology run amok. But the computer-system horrors at Oxford Health Plans, Inc., take the genre to a new level.

Only two months ago, Oxford was basking in Wall Street admiration for its blazing growth. This week, the healthmaintenance  organization disclosed that it will post a loss of $120 million or more for the current quarter, on top of a surprise third-quarter loss of $78.2 million -- its first loss since going public in 1991.

Oxford's shares closed Wednesday at $17.1250, down $2.9375, or 15%, on the Nasdaq Stock Market. Wednesday's closing stock price was down 75% from its level just before the company first revealed its troubles in late October.

How did disaster strike so quickly? As Oxford's business was faltering, it never saw the warning signs. One reason was a long list of troubles with a computer system that went on-line last year: how it was designed, how it was installed and how Oxford executives managed it.

The computer problems left Oxford unable to send out monthly bills to thousands of customer accounts and rendered it incapable of tracking payments to hundreds of doctors and hospitals.

In less than a year, uncollected payments from customers tripled to more than $400 million, while the sum Oxford owed care-givers swelled more than 50%, to more than $650 million.

For any company grappling with new information systems, Oxford offers a lesson in how not to proceed -- and in how hotshot technology can create even worse problems than the ones it was intended to solve.

PLAN FOR GROWTH: Oxford's dazzling growth was both its distinction and its undoing. The HMO began planning the new computer system in 1993, when it had just 217,000 members. The system didn't rev up until October of last year, by which time the HMO's membership had swelled to about 1.5 million. Thus, the new system was already outdated and outmanned.

Problems popped up immediately. Processing a new member sign-up was supposed to take just six seconds but instead took 15 minutes (a lag that later was fixed). And even as Oxford's backoffice infrastructure was overwhelmed, the company continued signing up hordes of new customers that its system couldn't handle -- more than half a million new members in the past year.

"If you drive a train at 150 miles an hour without good tracks, you derail," says David Friend, a global director at Watson Wyatt & Co., Bethesda, Md.

TAKE BABY STEPS, NOT BIG ONES: Build your new information highway from exit to exit, not coast to coast. Oxford locked into a design for the entire system in late 1993, which made it difficult to adjust to subsequent technological improvements as the project moved forward. Moreover, it tried to convert the bulk of its membership-billing database in one fell swoop -- -- some 43,000 accounts covering 1.9 million members.

Computer specialists say that such a sweeping conversion is far too difficult; companies should switch just the records for, say, one county or one business group at a time. "Put all of your small customers on the new system and see how it goes," suggests John Salek, of REL Consultancy in Purchase, N.Y.

Oxford's all-or-nothing approach misfired, making it all the harder to do further repairs without creating even bigger snags. Stephen F. Wiggins, Oxford's founder and chairman, acknowledges that a more incremental conversion would have been preferable.

BEWARE OF "DIRTY DATA": Oxford's old software was riddled with seemingly innocuous errors in member records that turned out to cause enormous problems in the new system. For instance, the old VMark uniVerse database tolerated errors that let a patient's Social Security number be entered in a box reserved for date of treatment. But the new database, from Oracle Corp., spit out such inconsistencies and refused to process the data until they were corrected.

The new software is "very unforgiving. If you don't get it right, you don't get it in," says Seth Lefferts, an information-systems manager at Oxford.

So when the program detected a single mistake in, for example, a 1,000-member account, it kicked out the entire group -- -- delaying billing and claims processing for all 1,000 members. When technicians fixed the error and re-entered the account, the new software would spit it out yet again when it detected the next mistake in a member's record. And so on.

The volume of individual mistakes "wasn't huge," says Paul Ricker, Oxford's vice president of information systems. "But the effect was rather extreme."

Oracle says it continues to work with Oxford and the HMO's other vendors to get the system working optimally.

DON'T LOSE TRACK OF RECEIVABLES: Once the Oracle software began balking at the old system's erroneous data, Oxford was forced to stop billing some customers for months at a time and often sent flawed invoices to many others. The HMO had no backup system, not even a platoon of pencil-wielding clerks, to fill the gap. Yet the company, following standard accounting practices, continued booking the unbilled income as quarterly revenue.

Then came the rub: When Oxford started to catch up on long-overdue accounts, contacting customers for the first time in months, many refused to pay and others said they had quit the HMO long ago. Hence, the company had to write off $111 million in uncollectable bills and admit it had overestimated its membership by 30,000.

Mr. Wiggins now says he should have "hired an army of temps, put them at a bank of IBM Selectrics, and had them type out bills. That would have made sure that everybody that owed us money had a slip of paper saying so."

DON'T FORGET THE LITTLE GUY: When the data overload swamped Oxford, the company focused first on trying to catch up with its biggest customer accounts and major providers. That seemed to make sense: Why not place top priority on fixing the biggest sources of your revenue and costs?

The risk of such logic, as Oxford now concedes, is that small customers are the ones most likely to disappear when service gets bad. They don't yelp; they simply stop paying and sign up elsewhere. As it turned out, the vast majority of the $111 million hit for uncollectable accounts came from small groups and individuals.

DON'T IGNORE PAYABLES: Oxford's failure to process claims on time angered many of the star physicians and renowned teaching hospitals whose participation was a key selling point of the HMO. Some providers say they still haven't been paid for care they delivered well over six months ago. A big doctor group at Columbia University's respected College of Physicians and Surgeons was owed $16 million at one point; New York Cornell Medical Center was owed as much as $17 million earlier this year.

This meant more than mere embarrassment. Oxford lost track of its actual medical costs -- critical information for reacting to surprise problems, setting reserves and projecting future liabilities. That last item, known as costs "incurred but not reported," or IBNR, is especially crucial in the insurance business. Oxford executives "weren't getting the statistics and data they needed to make accurate estimates on the IBNR," says Rob Levy of consulting firm William M. Mercer & Co. "So they were winging it. That's a disaster to any business."

Mr. Wiggins says that the company followed accepted practices in estimating such costs and that it thought it was making conservative projections. However, one particularly disturbing trend went undetected: a sharp 14% rise in Medicare costs at a time when Oxford's vaunted marketing machine was recruiting new elderly patients at a rate of more than 100 a day.

FORTIFY RANKS OF 'PROPELLER HEADS': Hire a data chief who knows how to manage a giant conversion. Many of Oxford's informationsystem  managers cut their teeth when the company was tiny, but a wholly different pacing and organization are needed when a project involves 150 programmers instead of a dozen. Intermediate goals need to be spelled out precisely, and multiple projects need to be coordinated.

Oxford dealt with data crises by piling more people onto the trouble areas, but it didn't seem capable of anticipating the next snafu.

Mr. Wiggins says his company is now looking for seasoned managers for information systems, medical management and other crucial areas of the company.

Copyright 1997 Dow Jones & Company, Inc. All Rights Reserved

/* end of copyright violation section */
Received on Thu Dec 11 1997 - 00:00:00 CST