Introducing the researchers Monday, national health IT czar David Brailer said hed commissioned the studies in part because of widespread assumptions that health IT could promote fraud by automating fake claims.
The researchers examined fraud prevention techniques in the highly automated and interconnected financial services sector for ones that could apply to health care. They also interviewed vendors and federal agencies who monitor health care fraud.
Copies of the reports are available here.
In the financial services sector, said Linda Kloss, CEO of AHIMA (American Health Information Managers Association), fraud costs 7 cents for every $100 processed.
In health care, an estimated 3 percent to 10 percent of all paid claims are believed to be a fraudulent, costing the health care industry between $51 billion and $170 billion each year.
Fraud takes many forms, such as incorrectly reporting procedures and diagnoses to boost payments and billing for services that werent performed.
It is committed by patients, providers, and several kinds of intermediaries.
One study examined how fraud could be impacted by computer-assisted coding, or software that automatically generates codes from physicians documentation of patient visits.
It is generally used when the kind of documentation varies little, like endoscopies.
The report says that data profiling, advanced analytic models, and rank scoring methods could be used by artificial neural networks to detect fraud.
However, savings will not be realized until the level of EHR adoption and common standards reach a higher threshold.
Thats partly because the fraud detection systems would rely on national statistics to find fraudulent behavior.
In addition, the report concludes that coding products should be certified for consistency and reliability.
The conclusions, Brailer said, were heartening.
"Coding systems have the potential to prevent fraud, not just detect it."
Kloss said that fraud preventing systems should not be tailored to quash known scams because swindlers would quickly adapt.
Instead, she said, systems must be built to inherently proactively detect and prevent fraud.
She implied that much of the techniques to prevent fraud would also be those required to enhance security, such as the ability to audit who saw which patient and who coded in diagnosis information.
"There arent magic bullets; its a whole set of practices, many of which are good health management practices," she said.
Brailer said that many feel that fraud can be prevented if patients have unique health identifiers to track their health care.
However, he said, the use of such identifiers is unlikely because many patients feel more secure with "federated identities," or separate identifiers in separate health systems.