Advanced analytics are a powerful secret weapon against healthcare fraud

Healthcare fraud has been one of the best performing U.S. growth industries for the last decade or more. Its explosive ascendancy in some regions of the country-south Florida and Los Angeles as prime examples-has made it the criminal career path of choice for fraudsters looking to reduce risk while increasing returns. The most entrepreneurial criminals actually are abandoning drug trafficking or more dangerous activities to enter the safe but lucrative arena of healthcare fraud.

The effect on our industry is chilling. In 2002, the Office of the Inspector General of the U.S. Department of Health and Human services identified $12.1 billion in fraudulent claims paid by Medicare. Estimates of annual losses to fraud are higher, ranging from 3% to 5% of national healthcare expenditures. This translates to $54 billion to $90 billion based on 2004 expenditures of $1,794 trillion projected by the Centers for Medicare and Medicaid Services (CMS). In comparison, credit card fraud, which is perceived as a huge problem, is $788 million in annual losses versus 100 times that number for healthcare fraud.

To help identify and deter Medicare fraud, CMS established Program Safeguard Contractor programs to support the Medicare Integrity Program. Another program developed by CMS is Medicare/Medicaid initiative, an innovative method that links state-run Medicaid data and federal-run Medicare data for a comprehensive view across both programs to detect even more suspicious activity.

The private sector of our industry also devotes resources to recovering or preventing payment of fraudulent healthcare claims. But that same year, our best anti-fraud efforts resulted in a paltry industry-wide $356 million in fraudulent claims recovered or identified before payment was issued (NHCAA’s 2002 Anti-Fraud Management Survey). That’s less than four-tenths of 1%-little more than a drop of water

Advanced analytical and intelligent software has been used in a number of industries for decades. One of the more successful applications today is the detection of fraud in the financial services industry. Products such as Fair Isaac’s Falcon Fraud Manager have dramatically reduced credit card fraud over the last decade.

The fact is that a handful of healthcare providers have discovered unscrupulous and often illegal ways to take advantage of imperfections in the healthcare system for their own financial gain, sometimes to the detriment of patients in cases where medically unnecessary procedures are performed for billing purposes. Now that legislation has been adopted across America to ensure prompt payments to providers, it becomes more difficult to adequately screen suspicious claims prior to payment. And though the number of fraudsters is small, the fraudulent practices they engage in have enormous impact on the industry in terms of cost and quality of care.

And while detecting fraud, abuse and error in submitted claims is its principal purpose, maintaining an organization’s payment integrity in handling valid claims – the clear majority – is equally important. In our own client base, we’ve detected policy and system errors of more than $1 million at a single payer within the first month of implementation. The need for payment integrity is especially important since prompt payment legislation levies hefty fines on companies and organizations that don’t comply.

Healthcare payers using advanced decisioning applications find themselves a step ahead of the game. Claims and entities, such as providers, pharmacies and patients, are scored. Claims adjusters, fraud investigators, case managers and other personnel can then focus their review on claims or cases with the highest likelihood of fraud, abuse or error, increasing efficiency and productivity dramatically.

Fair Isaac’s Payment Optimizer solution deploys “predictive models,” a type of statistical technique that uses historical claims and claims-related data to predict the risk of the current claim or behavior being assessed. A typical application for an insurance company will analyze hundreds of records in just milliseconds.

But advanced decisioning systems really flex their muscles through profiling technology. The profiling technology compresses terabytes of claims and claims-related data into essential informational packets. Since the profiling technology is dynamic, the system is able to detect fraud with increasingly greater accuracy over time and with additional aggregate data, giving organizations a greater opportunity to catch fraud, abuse and erroneous claims prior to payment, and before losses mount.

Prospective and Retrospective Detection

Once claims are already paid, the time, effort and expenses exhausted when trying to recoup losses from fraud rise exponentially. It’s in the best interests of healthcare payers to detect such claims before they’re paid, whenever possible. And with advanced decisioning systems, claims that are most at risk for fraud, abuse or error can be quickly detected with very high accuracy.

Organizations also use prepayment advanced decisioning systems to improve payment integrity, or the ability to pay correct and legitimate claims without slowing claims processing. The analytic-based solution enables thorough analyses of claims, resulting in identifying more fraud, abuse and errors prior to payment. The analyses reveal unusual patterns, such as in the behavior of providers and patients, and in the relationship between a patient’s activity across multiple providers in a high speed mode. The speed of analyses assures that claims processing is not compromised, thereby ensuring compliance with states’ prompt payment legislation

Advanced analyses enable healthcare payers to reap extraordinary benefits after payment as well. The intelligent processes of these analyses enable organizations to successfully review suspicious providers without expending laborious hours chasing red herrings. And when payers utilize post-payment analysis, they are able to detect fraud that’s only apparent over time, and not necessarily detected in smaller data sets.

Industry Specific Detection

When used with medical claims, predictive models allow organizations to detect unusual medical practices and charges that are inconsistent with a given peer group, so orthopedic surgeons are compared to and against one another, oncologists to other oncologists, chiropractors with other chiropractors. Payment Optimizer does not rely on the payer’s data to determine a provider’s peer group. A sophisticated blend of predictive models and expert decision logic is used to understand a provider’s practice patterns to determine the appropriate peer group for comparison. In one case, a payer using Payment Optimizer discovered a psychiatrist seeing 112 patients in a single day.

Curbing fraud and abuse among pharmacy services is particularly important in this day and age, as prescription drug benefits play an integral role in both Medicare and commercial benefit coverage. Like medical and dental models, pharmacy claims and claims-related data are analyzed in-depth using profiling technology. Problems such as claims for drugs never dispensed can be spotted. Audits can find pharmacies that habitually bill for drugs not provided and even diverted medicines for resale.

Benefiting the System

When payers succeed in cracking down on fraud, everyone wins. The system runs more smoothly and efficiently, and the cost savings to payers enable more affordable coverage plans and reduce costs to employers while improving quality of care. While advanced analytic tools have been used successfully for years in disease management, healthcare organizations are now finding them equally useful in claims processing and fraud detection. In an industry with $90 billion a year lost to fraud, errors and abuse, it’s high time for healthcare organizations to study the potential impact advanced decisioning systems may have on their bottom line.

Dr. Andrea Allmon is Fair Isaac Corporation’s director of healthcare operations and product management of Healthcare Solutions.