WASHINGTON Launched last summer, a $77 million computer system to stop Medicare fraud before it happens had prevented just one suspicious payment by Christmas. That saved taxpayers exactly $7,591.

Hoping for much better results, a disappointed Sen. Tom Carper, D-Del., said, "I wondered, did they leave out some zeros?"

Lawmakers had expected the system to finally allow Medicare to stanch a $60-billion-a-year fraud hemorrhage. Now they're worried about its future performance.

Medicare officials say it's unfair to grade the new technology on a single statistic.

"Suspending payments is only one way of stopping the money," says Ted Doolittle, deputy director of Medicare's anti-fraud program. "There's lots of ways of stopping the money, and we are using them all. Looking at payment suspensions only - that's an unsophisticated view that doesn't give you a full picture of our activities."

When other benefits of the system are taken into account, such as cases referred to investigators and changes to payment software that result in automatic denial of suspect claims, the potential savings in the first six months of operation easily exceed $20 million, Medicare officials indicated in a Jan. 27 letter to Carper. However, officials now acknowledge they don't know how much of that money has actually been recovered.

'Pay and chase'

Other experts point out that the mission of the new system was to stop bogus payments before they leave the Treasury's coffers, ending what's known as "pay and chase," where the agency automatically pays claims, even suspicious ones, and then reviews them weeks after the fact.

That can be a self-defeating way to do business. Law enforcement is usually several steps behind the fraudsters, who sometimes manage to flee the country with millions plundered from the government. The new computer system was meant to elevate Medicare's game, putting it in the same league as major credit card companies that can freeze accounts proactively.

"The whole idea for creating this technology was they were going to be able to end pay-and-chase," said Hank Walther, former head of the Justice Department's health care fraud division. "But we haven't yet seen evidence of its success."

Medicare has "got to explain to us clearly that they are implementing the program, that their goals are well-established, reasonable, achievable, and they're making progress," added Carper, chairman of a subcommittee that oversees federal financial management. "We're not sure if they've done those things."

Payment suspensions

Formally known as the Fraud Prevention System, the new screening technology was mandated by Congress. It's housed in the Baltimore area and staffed by as many as 150 people.

The goal was to allow Medicare to analyze large numbers of claims and spot patterns of potential problems. Does a storefront wheelchair retailer in Los Angeles, for example, have lots of customers in San Francisco, more than 350 miles away?

The system is supposed to issue an alert when something looks wrong so Medicare can investigate the claim before payment is sent out. That's critical because Congress has also directed Medicare to pay promptly, usually within 14 to 30 days.

Doolittle, a former federal prosecutor, said the system has generated some 2,500 leads and identified 600 suspicious cases, some which may mature into major investigations.

In their letter to Carper, Medicare officials said screening technology is now being used to evaluate all Medicare inpatient, outpatient and medical-equipment claims before payment. But payment suspensions did not begin until December 2011.

Medicare awarded an initial $77 million contract for the new system to defense giant Northrup Grumman and a group of other companies. About $27 million has been spent out of a four-year budget that could reach $100 million. IBM is also a contractor.

Sen. Tom Coburn, R-Okla., has questioned whether Grumman has the financial services experience necessary to take the lead on the job.

"My point is there was off-the-shelf stuff they could have bought and applied," said Coburn. "You don't have to recreate the wheel."

Responded Doolittle, "We are inventing the first wheel. The use of predictive modeling doesn't exist on this scale in industry."

Coburn says he's not impressed even by the figure of $20 million or more in potential savings that officials point to.

"If it is fully deployed, we ought to be seeing savings of $5 billion a month," he said. "It will be two to three years before we get an effective predictive system."