How to use data mining to identify trouble spots
How to use data mining to identify trouble spots
The increasing use of computer software by public and private payers to analyze data and uncover problematic trends in billing and reimbursement is fundamentally changing the fraud and abuse enforcement landscape. The good news is that providers can use these same "data mining" techniques technically to uncover problems on their own.
"The OIG and HCFA’s [Health Care Financing Administration’s] fiscal intermediaries are becoming more and more sophisticated in technology, just as we are," asserts Bret Bissey, chief compliance officer at Deborah Heart and Lung in Browns Mill, NJ. "I think this is a natural progression in the life cycle compliance."
The old way of looking for fraud and billing errors by examining samples and waiting for the hotline to ring is increasingly giving way to this technique, which lets payers and providers alike look at the breadth of claims rather than just a sample, says Mark Rucci, vice president of the consulting group of NiiS/APEX Group Holdings Inc. in Princeton, NJ.
"Not only do providers make mistakes in billing, Medicare makes mistakes in paying claims," he warns. "Hospitals should not only be concerned with billing correctly, but getting paid correctly."
But while data mining takes a proactive posture by looking for patterns that might reveal underlying problems by itself, it rarely leads to proof of fraud or even a mistake in billing, Rucci adds. A second-level manual review is almost always required.
Rucci reports that private payers as well as Medicare and Medicaid increasingly are employing these techniques prior to releasing a check. For example, he says Medicare code edits are now looking for mutually exclusive procedures billed simultaneously as well as comorbidities and other requirements.
According to Rucci, most data mining occurs post-payment. That is when issues such as same-day stays surface because data mining looks at claims historically. Previously, a random sample of claims would be reviewed manually and followed by a more thorough review of targeted claims. "You get a thorough review that way, but only on a sample of claims," he says.
Increasingly, electronic audits review all transactions. But Rucci says that even the most sophisticated commercial software packages available handle only certain issues. "Even though it may include a long list such as upcoding and unbundling, they can’t catch every possible thing that could be wrong with a Medicare claim," he explains.
"There is a definite movement and training program in this area that is well under way in the FBI [Federal Bureau of Investigation]," warns Edward Peloquin, director of the health care services department at Withum & Smith in Princeton, NJ. He says the FBI now is linking several commercial insurance companies with other databases and electronically mining those data to look for information on a national or regional basis, as well as sometimes between individual hospitals.
Peloquin says the FBI is employing this technique with the cooperation of special investigation units and others to review large volumes of data and then relate it back to individual providers.
According to Rucci, data mining now is being used to uncover a range of problems in addition to upcoding and unbundling. For example:
s Duplicate claims. Duplicate billing can take a variety of forms; sometimes bills are inadvertently submitted a second time, or interim billing for a long hospital stay results in overlapping bills. Data mining now is being used to run through a database of claims to examine the entire history of a patient to determine whether any services overlap.
s Eligibility data. Eligibility is a major issue for hospitals. Rucci says data mining can spot things such as over-age dependents as well as newborns, which sometimes are treated as a separate person on a claim. "You could actually have the same services billed once under the mother’s name and once under the child’s name and some unsophisticated systems won’t catch that," he says.
Data mining also can run all claims paid and the service dates against the dates of actual eligibility. The search looks for claims for people not on the eligibility file, claims outside the period of eligibility, and claims for dependents.
s Actuarial analysis. According to Rucci, one of the purest forms of data mining is looking at data from an actuarial point of view. That involves separating claims into very fine service categories from intensive care to skilled nursing and looking for patterns of unusual activity.
"That is data mining at its purest," he asserts. "There may be too many of something or not enough of something given certain norms, and that may warrant looking into how those kinds of claims are being billed," he explains. "That can uncover billing or reimbursement issues or highlight problems in how services are bundled."
s Utilization analysis. Related to actuarial analysis is utilization analysis, which looks at data against historical patterns of the data themselves. For example, hospital data over a three-year period can be aggregated to uncover inappropriate payments.
"Payers and providers using this technique as an investigative tool alsoare getting smarter and realizing that when they get a data dump, they are not getting everything that they think they are getting," adds Peloquin. "That is one reason that we are seeing probe audits, and one of the reasons why there will be more data requests and less on-site confiscation of records and documents."
Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.