A roughly 70-year-old health insurer that started out compensating sick or injured railway mail clerks is now helping set an industry standard for sophisticated fraud-detection software technology.
The Government Employees Hospital Association, founded in the 1930s with a few dozen members, reaped savings of hundreds of thousands of dollars in 2004 from a software system that it piloted. The system has helped enable the insurer to detect billing fraud before it sends out payments. "It's been a good investment," GEHA President Richard Miles said.
Many insurers use software in an effort to detect fraud. Most of the systems available detect fraud after a payment has already been sent. The software the GEHA uses tries to identify a fraudulent claim before the payment is processed.
"It's looking at data before you send a check out the door," said Bob Greene, the GEHA's data analysis manager. "Others look retrospectively at data, not in real time."
In 2003, the not-for-profit GEHA, which is now the third-largest insurer for federal employees and provides insurance for about 435,000 people, paid out nearly $1.5 billion in claims to providers all over the U.S. But when it started out, most of its members worked out of Kansas City, Mo., as railway mail clerks.
For the clerks, injuries were common because their profession was dangerous, said Miles, who has the minutes from the first association meeting framed in his office.
The clerks would ride trains and pick up bags of mail as they passed through cities or towns. The mailbags often contained money, making the clerks targets for thieves.
"(The clerks) would get robbed, so they would carry guns," Miles said. When a clerk was sick or injured, co-workers "would literally pass a hat to collect money to help each other out," Miles said.
Clerks then founded the GEHA's forerunner, the Railway Mail Hospital Association, which was incorporated in 1939 and became a part of the Federal Employees Health Benefits Program in the 1960s. The first members paid 50 cents per month in dues and were given $5 per day for missed work.
Now the insurer has members in every state. "We're about the size of a regional Blues plan," Greene said. "But we're spread out all over."
Being spread out makes it more difficult for the insurer to detect fraud. The GEHA would have a better opportunity to see billing patterns and notice irregularities that often indicate fraudulent billing if members were from the same area and seeing the same providers.
Greene said some of the computer programs that proved to be effective for other insurers didn't have the same success with the GEHA because the insurer's member base is so scattered. Then Greene came across an advertisement for a product called the Payment Optimizer made by the Fair Isaac Corp. "I called them and said if you can help us, then you can help anybody," Greene said.
Fair Isaac also produces credit-scoring technology for detecting credit card fraud. The software uses a similar methodology for detecting healthcare fraud.
The software gives each billing transaction a fraud-risk score. The transactions with high scores are flagged, signaling to staff members which claims should be reviewed.
The GEHA has been running the system since spring 2003. Greene says it's difficult to put an actual dollar amount on the savings because most is in the form of better-spent working hours.
"If you have 100 claims with three fraudulent cases, you'd have to spend all that time on finding the fraudulent cases," Greene said. "Now, if you have 100 claims, those fraud cases float to the top."
Although the software detects fraud and abuse, the GEHA found the majority of claims that are flagged have payment or billing errors. This has allowed the GEHA to improve its false-positive rate, the number of cases that appear to be fraudulent but aren't, Greene said.
Using software to defend against fraud is part of a growing trend.
"The days of someone hunched over a paper claims form are over," said Kirk Nahra, a lawyer with Wiley, Rein & Fielding. Nahra is also the general counsel of the National Health Care Anti-Fraud Association, which estimates about 3% to 10% of every dollar spent on healthcare in the U.S. is lost to fraud, totaling $39 billion to $150 billion a year.
Most of the fraud-detection software on the market utilizes a more traditional approach, looking at billing patterns and then examining the outliers, or payments by providers who charge abnormally high prices for a service, Greene said.
Mohit Ghose, director of public affairs for America's Health Insurance Plans, a 1,300-member trade group for insurers, said the prospective feature of the Fair Isaac system was interesting, but using software for this purpose isn't new.
"Fighting fraud is a priority," Ghose said. "The idea of using software has been around for a while."
The last time the AHIP asked its members if they used fraud-detection software was in 1999, and about 75% of the members said they had some sort of software in place, Ghose said. "One would assume that number has gone up," he said.
About 35 of the 41 Blue Cross and Blue Shield Association plans use fraud-detection software, said Byron Hollis, the national antifraud director for the Blues association. Hollis said depending on the size of the plan and the type of software, such systems can cost from $30,000 to $400,000.
IBM recently unveiled prospective software that works with its Fraud and Abuse Management Systems, which many insurers use. Similar to the Fair Isaac software, the IBM software will give each claim a fraud-risk score before a payment is made.
Paul Slagle, a founder of EDIWatch, which develops fraud-detection software for health claims, said IBM- and Fair Isaac-type systems aren't foolproof. The systems give odds as to whether a claim may be fraudulent but don't know if it's definitely a bad claim, he said.