Shining a light on health care fraud
Last update: June 12, 2005 at 5:40 PM
Warren Wolfe
June 13, 2005
If you screened a large group of people with the idea of scoping out those most likely to commit health care fraud, how would you separate the suspect from the honest?
Which provider would be most likely to submit a bill for, say, a stomach-probing endoscopy that never happened? Or who might collect workers' compensation benefits while secretly moonlighting at a second job?
You would get a clue from their pattern of behavior -- and you might compare it with patterns seen in other patients, doctors or other providers for further hints.
Such data mining was pioneered in the financial services industry by Fair Isaac Corp., the Minneapolis-based provider of business analytics and information technology best known for developing FICO, the most commonly used credit score. Almost all of the country's major banks and credit card issuers use the score to help determine the creditworthiness of their customers.
It's a technology the company hopes will score big in rooting out health care fraud.
That area is just a small piece of Fair Isaac's $706 million in annual revenue, but company officials believe that the potential to build the business is ripe. After all, Americans spend a staggering $1.7 trillion on health care annually.
Unlike credit card fraud where banks and financial services companies take the hit, health care fraud hits consumers squarely in the pocketbook through increased insurance premiums. Last year alone, health care fraud cost U.S. consumers between $50 billion and $170 billion, depending on which estimate you choose to believe.
"Health care fraud has really shifted in a very frightening way," said Andrea Allmon, Fair Isaac's product management director for health care and insurance analytics.
"It used to be that if providers want to defraud insurance companies, they would typically bill for things they didn't do," Allmon said. "Or, they would do a few unnecessary procedures, like everyone who came in would get a urinalysis, whether they needed it or not."
Now, "it's shifted to some very, very bright fraud schemes," Allmon said.
One recent high-profile case was the "Rent-A-Patient" scam in California, which involved some Minnesotans (though it did not involve Fair Isaac's technology).
Patients, many of them immigrants or minorities, were approached in company lunchrooms or solicited through community newspapers with the tug: Win a Free Trip to Disneyland! or Free Cosmetic Surgery! Those targeted, some of whom hailed from Minnesota, were typically covered by health insurance.
The otherwise healthy "patient" not only received a trip to Southern California or a face-lift, along with a kickback, but they agreed to undergo a needless surgical procedure, usually an endoscopy or colonoscopy. The riskiest surgical procedure commonly performed was one that corrects "sweaty palms," which requires collapsing the patient's lung in order to sever the nerve that controls perspiration in the hands.
A series of outpatient surgery clinics then submitted fraudulent claims on the patient's behalf.
"It was really unbelievable; it's hard to believe people would agree to that," said Dave Bohnenstingel, manager of the special investigations unit for Blue Cross Blue Shield of Minnesota.
The Eagan-based insurer was integral in bringing the scam to light, he said. "We cooperated with the [FBI] extensively, and we caught the scam fairly early on. We reviewed a lot of claims and we definitely felt that the providers knew that what they were doing was wrong." Blue Cross used software to detect the scam developed by ViPS, a division of WebMD that competes with Fair Isaac.
FBI agents raided suspected clinics last year and brought federal charges against several individuals allegedly involved.
But the scam serves as a good example at how brazen and widespread health care fraud has become -- and how difficult it can be to detect.
Technology transfer
Fair Isaac launched into the health care fraud-busting business seemingly by happenstance. In the mid-1990s, the Australian Health Commission asked the company if it could transfer its core technology for detecting credit card fraud to the health care field.
"We were able to create a product that worked very well there and we brought it back to the U.S. thinking the U.S. market would embrace it quickly," Allmon said.
But the U.S. health care market wasn't interested in a real-time fraud detector, especially since it usually took more than 45 days to process a claim, she said. At the time, no one was in a particular hurry.
"If a provider has ever been found suspect of doing something wrong, [the insurer] would put them on 100 percent review, which would slow down all their claims and force someone to sit there and poke through everything they've done," Allmon said. "It was sort of a blunt instrument."
Since then, many states have enacted laws that require companies to pay insurance claims within 30 to 45 days or face stiff fines. An industry study estimates that 70 percent of claims are paid within two weeks of receipt, and 90 percent within three weeks.
This has put some insurers in the position of choosing between efficiently processing claims and reporting lower losses.
Fair Isaac said its predictive modeling techniques can identify and mathematically represent underlying relationships in historical data to find patterns and make predictions about future behavior. Or, as the company states plainly in its annual report: "We use math to solve business problems."
Louis Saccoccio, executive director of the National Health Care Anti-Fraud Association, applauds the use of real-time technology to combat fraud. "Regrettably, most software is retroactive so that companies and the government are put in the position of 'pay-and-chase' -- the claim is already paid by the time they realize something might be wrong. So when you try to recover, many times it's a difficult thing to do."
Fair Isaac's Payment Optimizer software doesn't detect fraud -- rather it recognizes abnormal patterns that could be fraudulent before the claim is paid. It can review payment trends retroactively as well.
It usually scores each claim from zero to 1,000 with higher numbers more likely for fraud. Questionable claims, along with the reasons for the scoring, are sent to a claims adjuster or investigator for review. If fraud is suspected, the adjuster may refer it to law enforcement officials for further investigation or the to district attorney for prosecution.
'A very busy guy'
Common fraud, at least in the public's mind, involves providers -- such as doctors, chiropractors, laboratories, medical equipment firms and hospitals -- billing insurance companies for services never performed, for more expensive procedures than those received or misrepresenting services.
"In the broader scheme, it's fairly minor, compared with the dollars that are extracted from the system from organized crime and from unethical providers," Allmon said.
"It's only when you bring all of the data together that you get a clearer picture of what each provider's doing," she explained. "Take a doctor who bills maybe 15 hours of office visits on one day. You could say, 'He's a very busy guy.' But if he also bills another 15 hours of nursing home visits to Medicare on the same day, then he's not only a busy guy, he's probably a thief."
Grainy footage
One type of health care fraud that often hits the evening news -- complete with grainy, surreptitious surveillance footage -- is workers' compensation insurance fraud. Often a worker will go on disability and then work a second job while collecting another check from the government.
"It seems to be cyclical," said Mark Kulda, director of public affairs for the Insurance Federation of Minnesota. "If the economy is good, a fraudster can actually get a job. But when there are no jobs, that's when you see an increase."
Fair Isaac's model in this area compares "good and bad (perhaps fraudulent) behavior" on claims "even if it's just a one-time occurrence for an individual rather than looking for an organized-crime type of approach," said Kevin Lisle, Fair Isaac's product manager for insurance analytics.
To date, four states are using Fair Isaac's fraud detection products, while 10 are using another technology called the MIRA Claims Advisor, which helps predict more accurate loss reserves related to workers' comp claims. North Dakota recently licensed the company's MIRA product.
The company's presence in the workers' comp space was bolstered by the 2003 acquisition of Diversified HealthCare Services Inc., a provider of medical bill review products for the workers' comp insurance industry.
Utah is using Fair Isaac's VeriComp fraud detection software. "It's really been very helpful to us," said Joel Campbell, assistant vice president of special investigations for the state's workers' compensation fund.
Of the 40,000 workers' comp claims filed in Utah last year, Campbell's unit investigated about 600 cases. He estimated that the state has saved, on average, $6 million to $7 million a year, thanks in part to the unit's fraud-detection efforts.
"You see all kinds of people committing workers' comp fraud -- from people with criminal backgrounds to people who have never run a red light," he said.
Warren Wolfe
June 13, 2005
If you screened a large group of people with the idea of scoping out those most likely to commit health care fraud, how would you separate the suspect from the honest?
Which provider would be most likely to submit a bill for, say, a stomach-probing endoscopy that never happened? Or who might collect workers' compensation benefits while secretly moonlighting at a second job?
You would get a clue from their pattern of behavior -- and you might compare it with patterns seen in other patients, doctors or other providers for further hints.
Such data mining was pioneered in the financial services industry by Fair Isaac Corp., the Minneapolis-based provider of business analytics and information technology best known for developing FICO, the most commonly used credit score. Almost all of the country's major banks and credit card issuers use the score to help determine the creditworthiness of their customers.
It's a technology the company hopes will score big in rooting out health care fraud.
That area is just a small piece of Fair Isaac's $706 million in annual revenue, but company officials believe that the potential to build the business is ripe. After all, Americans spend a staggering $1.7 trillion on health care annually.
Unlike credit card fraud where banks and financial services companies take the hit, health care fraud hits consumers squarely in the pocketbook through increased insurance premiums. Last year alone, health care fraud cost U.S. consumers between $50 billion and $170 billion, depending on which estimate you choose to believe.
"Health care fraud has really shifted in a very frightening way," said Andrea Allmon, Fair Isaac's product management director for health care and insurance analytics.
"It used to be that if providers want to defraud insurance companies, they would typically bill for things they didn't do," Allmon said. "Or, they would do a few unnecessary procedures, like everyone who came in would get a urinalysis, whether they needed it or not."
Now, "it's shifted to some very, very bright fraud schemes," Allmon said.
One recent high-profile case was the "Rent-A-Patient" scam in California, which involved some Minnesotans (though it did not involve Fair Isaac's technology).
Patients, many of them immigrants or minorities, were approached in company lunchrooms or solicited through community newspapers with the tug: Win a Free Trip to Disneyland! or Free Cosmetic Surgery! Those targeted, some of whom hailed from Minnesota, were typically covered by health insurance.
The otherwise healthy "patient" not only received a trip to Southern California or a face-lift, along with a kickback, but they agreed to undergo a needless surgical procedure, usually an endoscopy or colonoscopy. The riskiest surgical procedure commonly performed was one that corrects "sweaty palms," which requires collapsing the patient's lung in order to sever the nerve that controls perspiration in the hands.
A series of outpatient surgery clinics then submitted fraudulent claims on the patient's behalf.
"It was really unbelievable; it's hard to believe people would agree to that," said Dave Bohnenstingel, manager of the special investigations unit for Blue Cross Blue Shield of Minnesota.
The Eagan-based insurer was integral in bringing the scam to light, he said. "We cooperated with the [FBI] extensively, and we caught the scam fairly early on. We reviewed a lot of claims and we definitely felt that the providers knew that what they were doing was wrong." Blue Cross used software to detect the scam developed by ViPS, a division of WebMD that competes with Fair Isaac.
FBI agents raided suspected clinics last year and brought federal charges against several individuals allegedly involved.
But the scam serves as a good example at how brazen and widespread health care fraud has become -- and how difficult it can be to detect.
Technology transfer
Fair Isaac launched into the health care fraud-busting business seemingly by happenstance. In the mid-1990s, the Australian Health Commission asked the company if it could transfer its core technology for detecting credit card fraud to the health care field.
"We were able to create a product that worked very well there and we brought it back to the U.S. thinking the U.S. market would embrace it quickly," Allmon said.
But the U.S. health care market wasn't interested in a real-time fraud detector, especially since it usually took more than 45 days to process a claim, she said. At the time, no one was in a particular hurry.
"If a provider has ever been found suspect of doing something wrong, [the insurer] would put them on 100 percent review, which would slow down all their claims and force someone to sit there and poke through everything they've done," Allmon said. "It was sort of a blunt instrument."
Since then, many states have enacted laws that require companies to pay insurance claims within 30 to 45 days or face stiff fines. An industry study estimates that 70 percent of claims are paid within two weeks of receipt, and 90 percent within three weeks.
This has put some insurers in the position of choosing between efficiently processing claims and reporting lower losses.
Fair Isaac said its predictive modeling techniques can identify and mathematically represent underlying relationships in historical data to find patterns and make predictions about future behavior. Or, as the company states plainly in its annual report: "We use math to solve business problems."
Louis Saccoccio, executive director of the National Health Care Anti-Fraud Association, applauds the use of real-time technology to combat fraud. "Regrettably, most software is retroactive so that companies and the government are put in the position of 'pay-and-chase' -- the claim is already paid by the time they realize something might be wrong. So when you try to recover, many times it's a difficult thing to do."
Fair Isaac's Payment Optimizer software doesn't detect fraud -- rather it recognizes abnormal patterns that could be fraudulent before the claim is paid. It can review payment trends retroactively as well.
It usually scores each claim from zero to 1,000 with higher numbers more likely for fraud. Questionable claims, along with the reasons for the scoring, are sent to a claims adjuster or investigator for review. If fraud is suspected, the adjuster may refer it to law enforcement officials for further investigation or the to district attorney for prosecution.
'A very busy guy'
Common fraud, at least in the public's mind, involves providers -- such as doctors, chiropractors, laboratories, medical equipment firms and hospitals -- billing insurance companies for services never performed, for more expensive procedures than those received or misrepresenting services.
"In the broader scheme, it's fairly minor, compared with the dollars that are extracted from the system from organized crime and from unethical providers," Allmon said.
"It's only when you bring all of the data together that you get a clearer picture of what each provider's doing," she explained. "Take a doctor who bills maybe 15 hours of office visits on one day. You could say, 'He's a very busy guy.' But if he also bills another 15 hours of nursing home visits to Medicare on the same day, then he's not only a busy guy, he's probably a thief."
Grainy footage
One type of health care fraud that often hits the evening news -- complete with grainy, surreptitious surveillance footage -- is workers' compensation insurance fraud. Often a worker will go on disability and then work a second job while collecting another check from the government.
"It seems to be cyclical," said Mark Kulda, director of public affairs for the Insurance Federation of Minnesota. "If the economy is good, a fraudster can actually get a job. But when there are no jobs, that's when you see an increase."
Fair Isaac's model in this area compares "good and bad (perhaps fraudulent) behavior" on claims "even if it's just a one-time occurrence for an individual rather than looking for an organized-crime type of approach," said Kevin Lisle, Fair Isaac's product manager for insurance analytics.
To date, four states are using Fair Isaac's fraud detection products, while 10 are using another technology called the MIRA Claims Advisor, which helps predict more accurate loss reserves related to workers' comp claims. North Dakota recently licensed the company's MIRA product.
The company's presence in the workers' comp space was bolstered by the 2003 acquisition of Diversified HealthCare Services Inc., a provider of medical bill review products for the workers' comp insurance industry.
Utah is using Fair Isaac's VeriComp fraud detection software. "It's really been very helpful to us," said Joel Campbell, assistant vice president of special investigations for the state's workers' compensation fund.
Of the 40,000 workers' comp claims filed in Utah last year, Campbell's unit investigated about 600 cases. He estimated that the state has saved, on average, $6 million to $7 million a year, thanks in part to the unit's fraud-detection efforts.
"You see all kinds of people committing workers' comp fraud -- from people with criminal backgrounds to people who have never run a red light," he said.
posted by josh at 8:14 PM
